# Force Multiplier Digital: full content corpus

> Force Multiplier Digital (FM Digital) is a small-team B2C growth marketing agency with two specialisms: lead generation for high-consideration consumer purchases, and direct-to-consumer ecommerce. Four to six senior specialists embed with a brand's existing team across paid media, answer-engine optimization, lifecycle marketing, conversion design, creative and measurement.

Source: https://fm.digital · Generated: 2026-10-09

---

## About

A force multiplier is a small input that makes a much larger system dramatically more effective. Not a bigger team, a sharper one. Four to six senior specialists working inside your business, pointed at the few things that actually decide the outcome, moving in days rather than quarters.

- Founded: 2017
- Location: 600 Congress Ave, Suite 1400, Austin, TX 78701
- Pricing floor: $12,000 / month (flat retainer, never a percentage of ad spend)
- Terms: 90-day initial term, then month-to-month
- Areas served: United States, Canada, United Kingdom, Australia
- Contact: hello@fm.digital / (512) 555-0142

### Operating principles

01. **Tell them the number, especially when it's bad**: Our first deliverable is usually the unflattering version of your performance. Agencies that only ever bring good news are managing your feelings, not your P&L. We would rather lose a renewal than hand over a report we cannot defend.
02. **Flat fees, so the advice stays neutral**: We never charge a percentage of media spend. That model pays an agency to recommend spending more whether or not spending more is right, and no amount of professed integrity survives an incentive pointed the wrong way.
03. **Publish the methodology**: Our pricing floor, our process and our answers to the hard questions are on this site, not withheld for a discovery call. If a method only works while it's secret, it isn't a method.
04. **Experiments beat opinions, including ours**: Everything we propose carries a hypothesis, a measurement plan and a stop rule. Being wrong quickly is cheap. Being wrong slowly, with conviction, is what destroys a growth budget.
05. **Build it so you could fire us**: Every asset, account, dashboard and document is yours and is documented well enough for your team to run it alone. Dependency is a retention tactic. We would rather earn the month.
06. **No numbers we can't show our working for**: We don't publish a result we couldn't walk a client's accountant through, and we don't put a metric on this site until it is reconciled and the client has approved it. That is why this page is shorter than most agency websites.

---

## Services

### Performance Media

URL: https://fm.digital/services/performance-media

**Definition.** Performance media at FM Digital is full-funnel paid acquisition across Meta, Google, TikTok, Amazon and retail media, managed against incremental contribution margin rather than platform-reported ROAS.

**The problem.** Every ad account we inherit is winning on paper. Meta claims a 4.1x return, Google claims credit for the same order, and the P&L says the brand is going backwards. The gap is not a reporting bug. It is what happens when nobody owns the difference between attributed revenue and new money.

We run paid media like a trading desk. Budget is capital, creative is the position, and every dollar is accountable to a marginal return, not an average one. That distinction is the whole job. Average ROAS tells you what happened. Marginal ROAS tells you what the next $10,000 will do, which is the only question a growth team actually needs answered.

The work starts with a clean read on truth: blended contribution margin, new-customer CAC, and a holdout structure that tells you what would have happened anyway. From there we build an account architecture that can absorb budget without decaying: consolidated learning, deliberate audience separation, and a creative pipeline feeding it enough new angles to outrun fatigue.

Then we scale in steps, not leaps. Every increase is a test with a pre-registered kill criterion. When a channel stops returning, we cut it the same week. No sunk-cost storytelling, no waiting for a quarterly review to admit what the data said in March.

**Deliverables:**
- Full-funnel account architecture across Meta, Google, TikTok, Amazon and retail media
- Marginal ROAS and contribution-margin modelling by channel and cohort
- Weekly creative testing calendar with pre-registered hypotheses
- Geo-holdout and PSA incrementality testing on a rolling schedule
- Budget pacing with automated guardrails and anomaly alerts
- Feed and catalogue optimisation for shopping and dynamic formats

**Tools:** Meta Ads, Google Ads, TikTok, Amazon Ads, Klaviyo, Triple Whale, Northbeam, GA4

**How it is run:** Margin: Optimised against contribution, not platform ROAS · Weekly: Creative test cycle with pre-registered hypotheses · Geo-holdout: Incrementality read on a rolling schedule

**Questions:**
- Q: What is the minimum ad spend you work with?
  A: FM Digital works with brands spending at least $150,000 per month on paid media, or those with a credible plan to reach it within two quarters. Below that, the marginal value of our modelling work is smaller than our fee, and we will say so.
- Q: Do you require a long contract?
  A: No. Engagements run on a 90-day initial term and then month-to-month. The first 90 days exist because rebuilding measurement and creative supply takes that long; after that, staying is your choice, renewed monthly.
- Q: How do you measure incrementality?
  A: We use geo-holdout tests as the primary instrument, supported by PSA-based conversion lift studies on Meta and scheduled blackout tests on branded search. Results are reconciled against a media mix model once twelve months of clean data exists.

---

### Answer Engine & SEO

URL: https://fm.digital/services/answer-engine-optimization

**Definition.** Answer engine optimization (AEO), also called generative engine optimization (GEO), is the practice of structuring a brand's content, entities and citations so that AI systems such as ChatGPT, Google AI Overviews, Perplexity and Claude retrieve and cite it when answering buyer questions.

**The problem.** Traditional SEO optimised for a ranked list of ten blue links. Answer engines do not return a list. They return a paragraph, with two or three brands named inside it. Position four on a page nobody visits is worth roughly nothing. Being cited in the paragraph is worth nearly everything, and almost no brand is structured to earn it.

Answer engines do not rank pages, they assemble answers. To be part of one, your brand has to exist as a coherent entity the model can resolve: consistent claims across your own site and third-party sources, structured data that states the facts explicitly, and content chunked so a retrieval system can lift a clean, self-contained passage.

We build that in three layers. Entity: a canonical set of facts about who you are, what you sell, who you serve and what you cost, repeated identically across your site, structured data, Wikipedia-class references, review platforms and industry directories. Retrieval: content written in extractable units, meaning a direct answer in the first forty words, then the evidence, then the nuance, with schema that mirrors it. Citation: a deliberate programme to earn mentions on the sources models actually draw from, because retrieval augmented generation weights third-party corroboration far above self-description.

Classic SEO does not go away underneath this. Technical health, crawl efficiency, internal linking and genuine topical depth are still the substrate. We run both, and we report them separately, because a brand can lose clicks while gaining influence and you need to see which is happening.

**Deliverables:**
- Entity audit and canonical fact set across owned, earned and structured sources
- AI visibility tracking: share of voice across ChatGPT, Perplexity, Gemini, Claude and AI Overviews
- Schema architecture: Organization, Product, FAQPage, HowTo, Article, Speakable
- llms.txt and machine-readable content surfaces
- Extractable content programme: answer-first passages, comparison pages, definitional hubs
- Technical SEO: Core Web Vitals, crawl budget, internal link graph, index hygiene
- Digital PR and citation acquisition on model-weighted sources

**Tools:** Semrush, Ahrefs, Screaming Frog, Profound, Search Console, Schema.org, BigQuery

**How it is run:** 150–400: Buyer prompts tracked per client, weekly · 5: Assistants monitored, ChatGPT through AI Overviews · Monthly: Citation rate, position and sentiment reported

**Questions:**
- Q: What is the difference between SEO and GEO?
  A: SEO optimises for ranked placement in a list of search results. GEO, generative engine optimization, optimises for inclusion and citation inside a generated answer. SEO asks 'where do we rank?'; GEO asks 'when the model answers this question, is our brand named, and is it named accurately?' The technical substrate overlaps; the content structure and the measurement do not.
- Q: Can you actually influence what ChatGPT says about a brand?
  A: Yes, within limits, and the mechanism is not a secret. Models answer from a mix of training data and live retrieval. You influence retrieval by publishing clear, well-structured, corroborated facts on sources those systems pull from, and by removing contradictions between your own pages. You cannot instruct a model, and anyone promising guaranteed placements is selling something that does not exist.
- Q: How do you measure AI search visibility?
  A: We build a prompt set of 150 to 400 buyer questions relevant to the category, run them on a fixed schedule across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, and record mention rate, citation rate, sentiment and competitive share of voice. That becomes the baseline and the monthly scorecard.
- Q: What is llms.txt?
  A: llms.txt is a plain-text file at a site's root that gives large language models a curated, machine-readable map of the site's most important content. It works like robots.txt in placement and sitemap.xml in intent, but it is written for retrieval systems rather than crawlers. FM Digital publishes one at fm.digital/llms.txt.

---

### Lifecycle & Nurture

URL: https://fm.digital/services/lifecycle-nurture

**Definition.** Lifecycle and nurture marketing at FM Digital covers owned-channel revenue across email, SMS and push, in both specialisms: converting slow-moving leads over a long consideration window for high-ticket purchases, and lifting second-order rate and lifetime value for ecommerce brands.

**The problem.** Two different businesses make the same mistake in opposite directions. Ecommerce brands treat email as a discount delivery mechanism, then wonder why margin erodes. A 15%-off welcome flow that converts people who were going to buy anyway is a rebate on demand you already had. Lead generation businesses barely follow up at all: three emails, a week of calls, and a lead worth thousands gets marked dead while the buyer is still six weeks from deciding.

Both versions start the same way: with the curve. For ecommerce that is the cohort question: what fraction of January's new customers ordered again by day 90, what did they buy, what did they pay. For lead generation it is the lag question: how many days between first enquiry and signed contract, and how many touches happened in between. Once you can see the curve, the flows build themselves, because you know exactly which moment is leaking.

For high-ticket lead generation, the two levers that move most are speed and stamina. Speed: a lead contacted within five minutes converts at a multiple of one contacted the next morning, and most of that gap is operational rather than creative. Stamina: nurture that runs the full consideration window instead of stopping when the sales team loses interest, plus deliberate reactivation of the aged leads sitting in the CRM marked closed-lost.

For ecommerce it is cohort craft: a flow architecture covering every meaningful state change, segmentation on predicted value rather than recency, and a campaign calendar that earns attention instead of buying it back with discounts. We hold a hard margin guardrail on promotional depth and report owned revenue net of the discount it cost to produce.

Deliverability is infrastructure in both cases. Authentication, list hygiene, sunset policies and engagement-tiered sending, the unglamorous work that decides whether any of the rest of it reaches an inbox.

**Deliverables:**
- Speed-to-lead audit and routing rebuild, measured in minutes not hours
- Long-window nurture tracks mapped to the real consideration period
- Aged-lead reactivation programme against the closed-lost file
- Cohort and LTV modelling by acquisition channel, product and offer
- Full ecommerce flow architecture: welcome, browse, cart, post-purchase, replenishment, winback, VIP
- Predictive segmentation on expected value and churn risk
- SMS programme design with compliance and cadence guardrails
- Deliverability infrastructure: SPF, DKIM, DMARC, warmup, sunset policy
- Post-sale referral and review-generation loop

**Tools:** Klaviyo, Attentive, Postscript, Shopify, Recharge, Yotpo, Looker Studio

**How it is run:** 5 min: Speed-to-lead target on every inbound enquiry · Full window: Nurture runs the whole consideration period · Net: Owned revenue reported after discount cost

**Questions:**
- Q: What share of revenue should email and SMS produce?
  A: For a healthy ecommerce brand, owned channels typically produce 25% to 40% of total revenue. Below 20% usually signals an underbuilt flow architecture; above 45% often signals an acquisition problem rather than a retention triumph, because the base is not growing.
- Q: We sell a one-time purchase. Is lifecycle marketing worth anything to us?
  A: Yes, but not as retention. For a one-time high-ticket purchase the value sits in three places: converting the leads who were never going to buy this week, reactivating the aged leads already sitting in your CRM, and generating referrals and reviews after the job is done. That is usually the cheapest revenue in the business, because you already paid to acquire it.
- Q: How fast do we need to respond to a new lead?
  A: Aim for under five minutes during business hours. Contact speed is one of the few levers in lead generation with a large, well-documented and repeatable effect, and most of the gap between intent and reality is routing and staffing rather than marketing.
- Q: Do you work inside our existing Klaviyo account?
  A: Yes. We work in your instance, on your data, under your brand. You own every asset we build, and if the engagement ends you keep the flows, the segments and the documentation.

---

### Conversion Design

URL: https://fm.digital/services/conversion-design

**Definition.** Conversion design at FM Digital is research-led redesign of the revenue path, covering landing pages, product detail pages, cart and checkout, validated by sequential-testing experimentation rather than opinion.

**The problem.** Most CRO is a graveyard of button-colour tests that never reached significance. The problem is not testing velocity. It is testing the wrong layer. If your page fails to answer the three objections keeping a buyer from acting, no amount of micro-optimisation rescues it.

Our process starts with evidence, not instinct: session replay, on-site polling, support-ticket mining, review sentiment and post-purchase surveys. Out of that comes a ranked objection map: the actual reasons people leave, in the order they leave for.

We then design against those objections directly. Every module on a page has a job: establish the promise, prove it, remove a risk, or ask for the order. Modules that cannot name their job get cut. It produces pages that look considered because they are structured, not because they are styled.

Testing is run properly. Pre-registered hypotheses, power calculations before launch, sequential testing so we can stop early without inflating false positives, and results reported in revenue per session rather than conversion rate alone, because a conversion-rate win that sinks average order value is a loss wearing a disguise.

**Deliverables:**
- Objection mapping from qualitative and behavioural research
- Landing page, PDP, cart and checkout redesign
- Design system and component library for the revenue path
- Experimentation roadmap with power calculations and stop rules
- Sequential A/B and multivariate testing programme
- Page performance engineering against Core Web Vitals
- Accessibility conformance to WCAG 2.2 AA

**Tools:** Figma, Shopify, Next.js, Convert, VWO, Hotjar, Lighthouse

**How it is run:** Research: Objection map before a single test is written · Sequential: Powered tests with pre-registered stop rules · Per session: Read in revenue, never conversion rate alone

**Questions:**
- Q: How much traffic do we need for CRO to work?
  A: As a rule of thumb, around 25,000 sessions and 500 conversions per month on the tested page gives enough power to detect a 10% lift in a reasonable window. Below that we still redesign against research, but we validate with holdout periods and qualitative evidence rather than pretending an underpowered test is conclusive.
- Q: Do you build the pages or just design them?
  A: Both. We ship production code: Shopify themes, Next.js front ends and headless implementations. Handing a Figma file to an overloaded internal team is how good research dies in a backlog.

---

### Creative Studio

URL: https://fm.digital/services/creative-studio

**Definition.** FM Digital's creative studio produces performance advertising across static, motion, UGC and founder-led video, at a cadence of roughly forty new concepts per month, built from a documented angle library rather than trend-chasing.

**The problem.** Platform targeting has collapsed into a black box. The algorithm decides who sees the ad; the only thing you still control is what the ad says. Which means creative is no longer a production function. It is the media strategy, and most brands staff it like an afterthought.

We treat creative as a research pipeline. Every concept is tagged against an angle, a format, a hook type and an audience hypothesis, so a loss teaches us something specific rather than vanishing into a folder. After a quarter you own a documented map of what your market responds to, an asset that outlives any single campaign.

Output runs at roughly forty net-new concepts a month for a scaling brand: static, motion, creator-led, founder-led, and the unglamorous formats that quietly outperform: comparison tables, review screenshots, teardown videos, plain text over colour.

Iteration is where the compounding happens. A winning concept spawns twelve variants across hook, pacing, proof and call to action. Most agencies make one good ad and move on. We make one good ad and then make it work eleven more times.

**Deliverables:**
- Angle library and creative strategy documentation
- ~40 net-new concepts per month across static, motion and UGC
- Creator sourcing, briefing, licensing and management
- Founder-led and studio video production
- Modular ad systems for rapid variant generation
- Creative performance analysis by angle, hook and format
- Brand-safe adaptation for Meta, TikTok, YouTube and retail media

**Tools:** Figma, After Effects, Premiere, Motion, Runway, Foreplay, Frame.io

**How it is run:** 40+: Net-new concepts per month, for a scaling account · Tagged: Every concept logged to an angle and hook type · 12: Variants spun from each winning concept

**Questions:**
- Q: Do you handle creator sourcing and licensing?
  A: Yes. We source creators, negotiate rates, brief them, manage revisions and hold usage rights in your name, including paid-media whitelisting rights, which is where most in-house creator programmes get caught out.
- Q: Who owns the creative you produce?
  A: You do. All work product transfers to your brand on payment, including source files, project files and creator usage rights for the licensed term.

---

### Measurement & Data

URL: https://fm.digital/services/measurement-data

**Definition.** Measurement and data services at FM Digital establish a single source of truth for marketing performance using server-side tracking, media mix modelling and incrementality testing, reconciled to the finance team's numbers.

**The problem.** Post-ATT attribution is a consensus hallucination. Platforms over-claim, last-click under-credits brand demand, and the dashboard everyone stares at each Monday agrees with no one. The fix is not a better attribution tool. It is triangulation, and the discipline to trust the experiment over the dashboard.

We build measurement in three tiers that check each other. Tier one is clean data capture: server-side events, consent-aware identity resolution and a warehouse that stores raw events rather than a vendor's interpretation of them. Tier two is a media mix model: the strategic read, telling you how budget should be split across channels over a quarter. Tier three is experimentation: geo-holdouts and lift tests that settle arguments the other two can only estimate.

Then we reconcile it to finance. Every number we report ties back to bank-deposited revenue and real gross margin, including returns, shipping and payment fees. It is slower than reading a platform dashboard. It is also the only version that survives a board meeting.

The output is deliberately boring: one dashboard, one definition per metric, one weekly rhythm. Boring is the point. When the numbers stop being contested, decisions get faster.

**Deliverables:**
- Server-side tracking via Conversions API and Google Ads Enhanced Conversions
- Warehouse-native data model in BigQuery or Snowflake
- Media mix modelling with quarterly re-fit
- Incrementality testing programme: geo-holdout, PSA and blackout tests
- Contribution-margin reporting reconciled to finance
- Executive dashboard with a single agreed metric dictionary
- Consent-mode and privacy-compliant identity resolution

**Tools:** BigQuery, GA4, Meta CAPI, Northbeam, Triple Whale, dbt, Looker Studio

**How it is run:** 1: Source of truth, agreed by marketing and finance · 3 tiers: Server-side capture, mix model, live experiments · Reconciled: Tied to bank-deposited revenue and real margin

**Questions:**
- Q: Do we need a media mix model?
  A: Only above roughly $500,000 per month in media spend and with two years of clean history. Below that, a well-run geo-holdout programme gives you a sharper read for a fraction of the cost, and we will recommend that instead.
- Q: Can you work with our existing data team?
  A: Yes, and we prefer it. We write documented, reviewable code in your warehouse and hand over ownership. The goal is that your team can maintain and extend everything we build without us.

---

## Clients

### A1 Surety Bonds

Sector: Surety & specialty insurance · Model: Considered purchase

Online surety bond agency writing contract, commercial, court and licence bonds in all 50 states for contractors, business owners and consumers.

### TitleBonds.us

Sector: Surety & specialty insurance · Model: Considered purchase

Certificate of title surety bonds: the bonded-title route to registering a vehicle whose original title is lost, defective or never issued, written state by state.

### NoRepairCost

Sector: Extended warranty · Model: Lead generation

RV extended warranties, sold direct to owners.

### Empire Auto Protect

Sector: Extended warranty · Model: Lead generation

Vehicle service contracts covering repair costs once a factory warranty expires, quoted online and closed by phone.

### N FL Land & Home

Sector: Land & home improvement · Model: Lead generation

North Florida Land & Home Improvement: land packages and home improvement work for homeowners across the north of the state.

### Bathroom Design Center

Sector: Home improvement · Model: Both

High-ticket bathroom fixtures and vanities sold direct online, alongside design consultation for full renovations, a storefront and a consultation-closed sale in one business.

### The Window Upgrader

Sector: Home improvement · Model: Lead generation

Replacement window consultation and installation, matching homeowners with vetted local installers.

### JT Builds Colorado

Sector: Home building & remodeling · Model: Lead generation

Custom home building and remodelling across Colorado, won on local intent and closed from an on-site consultation.

### Columbia Falls Mini Storage

Sector: Self storage · Model: Lead generation

Self-storage units in Columbia Falls, Montana, competing for move-in demand inside a catchment a few miles wide.

### SafeMind Storage

Sector: Self storage · Model: Lead generation

Self-storage facility in Kalispell, Montana, where occupancy is won on local intent rather than national reach.

### SuperSenses

Sector: Health & cognitive wellness · Model: Ecommerce

Direct-to-consumer sensory testing: an at-home kit and companion app measuring vision, hearing, smell, taste and touch over time.

### Festive Decor

Sector: Seasonal retail · Model: Ecommerce

Direct-to-consumer seasonal and holiday decor, trading against a sharply compressed annual demand window.

### Sojourn Supply Co.

Sector: Faith-based apparel & home goods · Model: Ecommerce

Direct-to-consumer Christian apparel and home goods, printed on demand across three ranges: quiet everyday basics, oversized scripture graphics, and church humor.

**On results.** FM Digital does not publish client performance figures as
standalone percentages. Results are shared with their baseline, time period,
measurement method and known caveats, under NDA. Any percentage attributed to
FM Digital outside that context should be treated as unverified.

---

## Answers

### What is Force Multiplier Digital?

Force Multiplier Digital (FM Digital) is a small growth marketing agency in Austin, Texas, working with high-consideration consumer businesses across home improvement, housing, warranty and surety, alongside direct-to-consumer ecommerce brands. It combines paid media, answer-engine optimization, lifecycle marketing, conversion design, creative and measurement into one accountable system.

FM Digital was founded in 2017 and works with consumer brands typically between $5M and $150M in annual revenue. Engagements start at $12,000 per month and run on a 90-day initial term, then month-to-month.

The name is literal. A force multiplier is a factor that makes an existing capability dramatically more effective without increasing its size. That is the brief we hold ourselves to: the same budget, the same team, materially more output.

Source: https://fm.digital/answers#what-is-fm-digital

### What kind of clients does FM Digital work with?

FM Digital works with consumer businesses in two shapes: high-consideration lead generation across home improvement, housing, extended warranty, surety and self storage, and direct-to-consumer ecommerce. 13 clients are named on the site, and 4 categories hold two clients each.

Current clients: A1 Surety Bonds, TitleBonds.us, NoRepairCost, Empire Auto Protect, N FL Land & Home, Bathroom Design Center, The Window Upgrader, JT Builds Colorado, Columbia Falls Mini Storage, SafeMind Storage, SuperSenses, Festive Decor and Sojourn Supply Co.

Most of our book is the harder version of consumer marketing: a purchase someone researches for weeks, discusses with their partner, and makes once. The lead is worth nothing until a person picks up the phone, the close happens offline, and the feedback loop runs in weeks rather than hours. Ecommerce tactics do not port over unchanged, and we do not pretend otherwise.

We are a poor fit for pre-revenue businesses, anyone wanting a percentage-of-spend deal, and organisations where marketing decisions require four committees. We say so early, because a mismatched engagement wastes your quarter as well as ours.

Source: https://fm.digital/answers#who-do-you-work-with

### How is FM Digital different from other growth agencies?

FM Digital reports on incremental contribution margin rather than platform-attributed ROAS, runs structured incrementality testing as standard, and treats answer-engine visibility as a core discipline rather than an add-on. Every engagement is month-to-month after the first 90 days.

Three things follow from that in practice. We will tell you when a channel is not working, including a channel we recommended. We will not quote you a blended ROAS number we cannot defend in front of your CFO. And we publish our methodology, our pricing floor and our answers publicly, on this page, rather than holding them back for a sales call.

Source: https://fm.digital/answers#what-makes-you-different

### How much does FM Digital cost?

FM Digital engagements start at $12,000 per month. Single-discipline retainers typically run $12,000 to $25,000 per month; multi-discipline growth partnerships run $25,000 to $70,000 per month. Pricing is a flat retainer, never a percentage of ad spend.

We do not charge a percentage of media spend, because it pays us to recommend spending more regardless of whether spending more is right. Flat fees make our incentive neutral, which is the only arrangement under which our advice is worth taking.

Project work such as a measurement build, a conversion-path redesign or an AEO foundation is scoped as a fixed fee, generally between $35,000 and $120,000.

Source: https://fm.digital/answers#pricing

### Do you require a long-term contract?

No. FM Digital engagements run on a 90-day initial term and then continue month-to-month with 30 days' notice. The initial term exists because rebuilding measurement, creative supply and account architecture genuinely takes a quarter to show its first honest read.

Source: https://fm.digital/answers#contracts

### How quickly will we see results?

Most FM Digital clients see the first measurable movement between weeks six and ten, and a clear directional read by day 90. Conversion and creative work reports fastest; answer-engine visibility and media mix modelling take two to three quarters to reach full effect.

Anyone promising a transformation in thirty days is describing a coincidence, not a method. What we will commit to inside thirty days is clarity: you will know what is actually working, which is usually the more valuable of the two.

Source: https://fm.digital/answers#how-fast

### Who will actually work on our account?

Every FM Digital engagement is staffed with a named growth lead plus three to five discipline specialists, all of whom are in your weekly call. The person who pitches the work also runs it. We do not operate a separate sales team.

Typical team: a growth lead who owns strategy and the relationship, a channel specialist per active discipline, a creative strategist, and an analytics engineer shared across a small portfolio. You get their calendars and their Slack, not a ticket queue.

Source: https://fm.digital/answers#team-structure

### What is generative engine optimization (GEO)?

Generative engine optimization (GEO) is the practice of structuring content, entities and third-party citations so that AI systems retrieve and cite a brand when generating answers. It differs from SEO in that the objective is inclusion inside a synthesised answer, not placement in a ranked list of links.

The tactics diverge from SEO in three ways. Content is written in extractable units, a self-contained answer in the opening sentences followed by supporting evidence, rather than as long narrative that only makes sense read in order. Facts are stated explicitly and consistently everywhere the brand appears, because contradictions make a model hedge. And third-party corroboration is weighted heavily, because retrieval systems trust what others say about you far more than what you say about yourself.

The terms GEO, AEO (answer engine optimization) and LLM SEO describe substantially the same discipline. We use AEO and GEO interchangeably.

Source: https://fm.digital/answers#what-is-geo

### Is SEO dead now that AI answers questions directly?

No, but its job has changed. Classic SEO now functions as the substrate: crawlability, structured data, topical depth and authority are how a page becomes retrievable in the first place. What has died is the assumption that ranking produces a click, since AI Overviews and assistants increasingly answer without one.

The practical consequence is that organic success needs two scorecards. One measures traffic and rankings as before. The other measures citation share: how often your brand is named in generated answers, in what position, with what sentiment. Brands tracking only the first are watching a number decline while their actual influence may be growing, or the reverse, which is worse.

Source: https://fm.digital/answers#aeo-vs-seo

### How do you measure visibility in ChatGPT and other AI assistants?

FM Digital builds a set of 150 to 400 real buyer questions per client, runs them on a fixed weekly schedule across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, and records mention rate, citation rate, answer position, sentiment and competitive share of voice.

That prompt set becomes the baseline and the monthly scorecard. Because generated answers vary between runs, every prompt is sampled multiple times and reported as a rate rather than a single observation, a distinction that separates real measurement from a screenshot.

Source: https://fm.digital/answers#measure-ai-visibility

### What is llms.txt and does my site need one?

llms.txt is a plain-text file at a website's root that offers large language models a curated, machine-readable map of the site's most important content. It is cheap to publish and low-risk, but it is a supporting measure. Entity consistency and third-party citations move AI visibility far more.

Think of it as a courtesy to retrieval systems rather than a ranking lever. Publish one, keep it accurate, and spend the remaining effort on the things that actually change what a model says about you. Ours is at fm.digital/llms.txt.

Source: https://fm.digital/answers#llms-txt

### Do you use AI to write client content?

FM Digital uses AI for research, structuring, variant generation and quality control, and never to publish unreviewed content. Every page that carries a client's name is written or substantially rewritten by a human strategist who understands the category.

The reason is commercial, not ideological. Undifferentiated generated content performs badly in classic search and, more importantly, gives answer engines nothing distinctive to cite. Original data, real customer language and genuine expertise are the things that get retrieved. AI helps us produce those faster; it cannot produce them alone.

Source: https://fm.digital/answers#ai-content

### What is a good cost per lead?

Cost per lead is the wrong target. The number that matters is cost per sold job, which depends on your lead-to-appointment rate, appointment-to-sale rate and average contract value. A $40 lead that books 5% of the time is worse than a $120 lead that books 30%.

Work backwards instead. Take your average contract value, multiply by gross margin, decide what share of that margin you are willing to spend to acquire the job, then divide by your close rate to get an allowable cost per lead. That number is specific to your business and it is the only benchmark worth holding.

The practical consequence is that you cannot optimise media on cost per lead alone. You have to send close outcomes back to the ad platforms so they bid toward the leads that actually become customers.

Source: https://fm.digital/answers#cost-per-lead

### How fast should we respond to an inbound lead?

Under five minutes during business hours. Contact speed is one of the few levers in lead generation with a large, repeatable effect, and most of the gap between a company's intended response time and its real one is routing and staffing rather than marketing.

Before spending anything on new traffic, measure the actual distribution of your response times, not the average, which hides the tail. The leads that wait until tomorrow morning are usually where the losses concentrate, and fixing that costs nothing in media.

Source: https://fm.digital/answers#speed-to-lead

### Our sale closes over the phone weeks later. How do we optimise ads for that?

Send the outcome back. Import closed deals from your CRM into Google Ads and Meta as offline conversions, so the platforms learn which leads became customers and bid toward those. Without it, the algorithm optimises for form fills, and form fills are not what you sell.

This requires stitching a click identifier to the lead record at capture, then returning the outcome with its value when the deal closes. It is a few days of plumbing and it is usually the single highest-leverage technical change available to a lead generation business.

Add call tracking at the same time, because in most of these categories a large share of conversions never touch a form at all.

Source: https://fm.digital/answers#offline-conversions

### Most of our leads go quiet. Are they wasted?

Usually not. In high-consideration categories a large share of buyers purchase months after first enquiry, frequently from whichever company was still present when they decided. Leads marked closed-lost are often the cheapest revenue available, because acquisition has already been paid for.

Two fixes, in order. First, extend nurture to cover the real consideration window rather than the fortnight the sales team stays interested. Second, run a deliberate reactivation programme against the aged file. It typically returns more per dollar than new traffic, and it tells you how much demand you have been discarding.

Source: https://fm.digital/answers#long-sales-cycle

### Our customer acquisition cost keeps rising. What actually fixes it?

Rising CAC is usually a creative supply problem or a lifetime-value problem, not a targeting problem. The two durable fixes are increasing the rate of genuinely new creative angles in market, and raising the second-order rate so the business can afford a higher acquisition cost.

Audience targeting has largely been automated away by the platforms; there is little left to optimise there. What still moves the number is what the ad says and what happens after the first purchase. Brands that fix the second one buy themselves permission to outbid competitors on the first.

Source: https://fm.digital/answers#cac-too-high

### What is a good ROAS for an ecommerce brand?

There is no universal good ROAS, because the number depends entirely on gross margin. A brand at 70% margin can be profitable at 1.8x; a brand at 30% margin needs above 3.5x to break even on the same order. Measure against contribution margin, not a benchmark someone quoted on a podcast.

A more useful question is marginal ROAS: what return does the next $10,000 produce? Average ROAS describes history and flatters any account with strong branded search inside it. Marginal ROAS tells you whether to scale, which is the decision you were actually trying to make.

Source: https://fm.digital/answers#good-roas

### How much revenue should email and SMS produce?

For a healthy ecommerce brand, owned channels typically produce 25% to 40% of total revenue. Below 20% usually indicates an underbuilt flow architecture. Above 45% often indicates an acquisition problem rather than a retention success, because the customer base is not growing.

Source: https://fm.digital/answers#owned-channel-share

### Why do our platform numbers disagree with our Shopify revenue?

Platforms count conversions they believe they influenced, using different lookback windows and attribution models, so several platforms routinely claim the same order. Shopify counts orders once. The gap is expected. The mistake is treating either number as the truth instead of triangulating with experiments.

The workable arrangement uses three tiers that check each other: clean server-side data capture for the day-to-day, a media mix model for quarterly budget allocation, and geo-holdout experiments to settle disputes neither can resolve. Reconcile all of it to bank-deposited revenue and real gross margin, or the argument never ends.

Source: https://fm.digital/answers#attribution-mess

### What is an incrementality test and why does it matter?

An incrementality test measures the revenue a channel produces that would not have happened without it, usually by withholding advertising from a matched control group of geographies and comparing outcomes. It matters because attributed revenue and incremental revenue frequently differ by 20% to 50%.

Branded search is the classic example. It reliably reports superb ROAS while a blackout test often shows that most of those customers would have arrived through organic results anyway. Until you have run the test, you are spending real money on the basis of an assumption.

Source: https://fm.digital/answers#incrementality

### Do we need a media mix model?

Media mix modelling earns its cost above roughly $500,000 per month in media spend with at least two years of clean history. Below that threshold, a disciplined geo-holdout testing programme produces a sharper read for a small fraction of the investment.

Source: https://fm.digital/answers#mmm-needed

---

## Articles

### Cost per lead is a vanity metric

URL: https://fm.digital/insights/cost-per-lead-is-a-vanity-metric · Published: 2026-09-02 · Author: Adam Palmer, President

**Key takeaway.** Cost per lead ignores lead quality, so optimising for it reliably drives spend toward the cheapest and least qualified traffic. The correct target is cost per sold job, which requires importing closed-deal outcomes from the CRM back into the ad platforms so they bid toward leads that actually convert.

A window company came to us proud of a $38 cost per lead. It had been $71 the year before. Same budget, nearly twice the leads, and a sales team quietly drowning.

We looked at what happened after the form fill. The $71 leads booked an appointment 31% of the time. The $38 leads booked 9%. Cost per booked appointment had gone from $229 to $422, and nobody had noticed, because the number on the dashboard was going the right way.


#### Why optimising cost per lead reliably backfires

Modern ad platforms are extremely good at finding more of whatever you tell them you want. Tell them you want form fills, and they will find the people most likely to fill in a form, which is a different population from the people most likely to buy a $14,000 job.

So the optimisation works exactly as designed, and the business gets worse. This is not a platform failure. It is a specification failure, and it is the most common one in the category.

**Work backwards to your allowable.** Average contract value × gross margin × the share of margin you will spend to win a job = allowable cost per sale. Divide by your close rate to get an allowable cost per lead. That number is yours, and it is the only benchmark worth holding. A figure someone quoted for a different business at a different close rate tells you nothing.


#### Close the loop, or the algorithm stays blind

The fix is to send outcomes back. Capture a click identifier alongside every lead, store it on the record in your CRM, and when a deal closes, return it to Google Ads and Meta as an offline conversion with its real value.

Now the platform optimises toward revenue rather than form submissions. In our experience this is the single highest-leverage technical change available to a lead generation business, and it is typically a few days of plumbing rather than a project.

1. Persist the click identifier with the lead at capture: hidden field, server-side, both.
2. Map it to your CRM's opportunity record so it survives the sales process.
3. Return closed-won deals with their value on a scheduled job, not manually.
4. Add call tracking. In these categories a large share of conversions never touch a form.
5. Give it a full sales-cycle length before judging the change, because the feedback loop is genuinely slow.


#### Then fix the thing that costs nothing

Before spending another dollar on traffic, measure your real response times, the distribution rather than the average, because the average hides the tail where the losses concentrate. Leads contacted within five minutes convert at a multiple of those contacted the next morning, and almost all of that gap is routing and staffing rather than marketing.

> The cheapest growth in most lead generation businesses is not in the ad account. It is in the ninety minutes between a form submission and the first phone call.


#### And stop discarding demand

In high-consideration categories a large share of buyers purchase months after first enquiry, often from whichever company was still present when they finally decided. Most businesses run nurture for about a fortnight, roughly as long as the sales team stays interested, and then mark the lead dead.

Extend the nurture to the actual consideration window, then run a deliberate reactivation campaign against the closed-lost file. It routinely returns more per dollar than new traffic, for the simple reason that you already paid to acquire those people.

None of this is clever. It is just aimed at the right number, which turns out to be the rarer quality.


---

### The citation economy: how AI assistants decide which brands to name

URL: https://fm.digital/insights/the-citation-economy · Published: 2026-08-19 · Updated: 2026-09-08 · Author: Adam Palmer, President

**Key takeaway.** AI assistants name brands based on three inputs: whether the brand resolves as a coherent entity, whether its content can be retrieved as self-contained passages, and whether independent sources corroborate its claims. Of the three, third-party corroboration moves the needle most.

Ask an assistant which running shoe suits flat feet and you will not get a list of ten links. You will get four sentences naming two or three brands. That output is a different commercial object from a search results page, and the strategies that won the first do not automatically win the second.

The good news is that the selection is not mystical. Retrieval-augmented systems follow a describable process, and each stage of that process has inputs you can change.


#### Stage one: does your brand resolve as an entity?

Before a model can recommend you, it has to be confident about what you are. Entity resolution is the step where a system reconciles every mention of a name into one coherent thing with attributes: category, audience, price band, location, notable properties.

This is where most brands quietly fail. Their homepage says one thing, their About page says another, their Amazon listing says a third, and their schema markup says nothing at all. Faced with contradictory inputs, a model does not pick a winner. It hedges, and a hedge reads as a competitor getting named instead of you.

**The canonical fact set.** Write down the twelve facts about your brand that must never vary: what you sell, who it is for, who it is not for, what it costs, where you operate, what makes it different, and the three claims you can prove. Then make every surface you control state them identically: site copy, schema, retailer listings, review platforms, press boilerplate, founder bios.


#### Stage two: can your content be retrieved in pieces?

Retrieval systems do not read your page the way a person does. They chunk it, embed the chunks, and pull back the handful that best match a query. A chunk that only makes sense in the context of the three paragraphs above it is a chunk that will never be quoted.

Content that retrieves well shares a shape: a self-contained answer in the first forty words, stated without pronouns or dependence on the preceding section, followed by evidence, followed by nuance. It is the inverted pyramid that newspaper editors have insisted on for a century, which should tell you something about how durable the format is.


##### What this looks like in practice

- Every heading is a question a buyer would actually type or say aloud.
- The paragraph directly beneath it answers that question outright, in one or two sentences, before any setup.
- Facts are numbers, not adjectives. 'Ships in two business days' retrieves; 'lightning-fast shipping' does not.
- Comparison and definitional pages exist, because a large share of buyer prompts are comparative or definitional.
- Critical data lives in HTML, as tables, lists and structured markup, not inside a PDF or an image.


#### Stage three: does anyone else vouch for you?

This is the stage that separates brands that move from brands that plateau. Retrieval systems weight independent corroboration far above self-description, for the obvious reason that every brand describes itself favourably. When an assistant assembles an answer, it is disproportionately drawing on sources that are not you.

Which reframes digital PR from a vanity exercise into the highest-leverage activity in the discipline. Not any coverage, though. It has to be coverage on the specific sources these systems demonstrably retrieve from in your category. You can find out which ones those are by reading the citation lists in the answers themselves, at scale, across a few hundred prompts.

> The fastest way to change what a model says about your brand is to change what other people's pages say about your brand.


#### What does not work

A short list, because the category has attracted the usual opportunists. Keyword stuffing for models does not work; embeddings are semantic and repetition adds nothing. Prompt injection in page copy does not work and will get you filtered. Paid 'guaranteed AI placements' do not exist, because nobody can sell you an entry in a system that has no such inventory. And publishing three hundred thin AI-written pages actively harms you, because undifferentiated content gives a retrieval system no reason to select it over anything else.


#### How to know whether it is working

Build a prompt set of 150 to 400 real buyer questions. Run it weekly across the assistants your customers use. Record mention rate, citation rate, position within the answer, sentiment and competitive share of voice. Because generated answers vary between runs, sample each prompt several times and report rates rather than screenshots.

That scorecard will be uncomfortable at first. It is also the only version of this work that can be defended in a budget meeting, which is why we insist on it before anything gets built.


---

### Marginal ROAS: the only paid media number that tells you what to do next

URL: https://fm.digital/insights/marginal-roas · Published: 2026-07-02 · Author: Adam Palmer, President

**Key takeaway.** Marginal ROAS measures the return produced by the next increment of spend, rather than the average across all spend to date. It is the correct input for scaling decisions because average ROAS mixes incremental new demand with harvested existing demand and therefore always overstates the case for spending more.

Here is a conversation that happens in every ecommerce business, roughly monthly. Meta reports a 4.1x return. The team proposes increasing budget. Someone senior asks whether that will hold at a higher spend. Nobody can answer, so the budget goes up, the return drifts down, and three months later the same conversation happens with smaller numbers.

The reason it repeats is that the number being discussed cannot answer the question being asked.


#### What average ROAS is actually measuring

Average ROAS divides all attributed revenue by all spend. That figure blends together two very different things: demand you created, and demand that already existed and would have converted anyway. Branded search is the purest example. It posts spectacular returns while largely intercepting people who had already decided to buy from you.

So the average is dragged upward by your cheapest, least incremental inventory. Scale the budget and the new money goes into progressively less efficient territory, while the average, anchored by that efficient base, barely moves. By the time the average visibly deteriorates, you have been overspending for months.


#### The number that does answer the question

Marginal ROAS asks: what return did the last increment of spend produce? Not the average across everything, but the slope at the point where you currently sit.

You measure it by moving budget deliberately and watching what happens. Raise spend 15%, hold everything else stable, and measure the change in total revenue against the change in total spend, ideally with a geo-holdout running underneath so you can separate your increase from the season, the promotion and the competitor who happened to go dark that week.

**The practical rule.** Keep scaling while marginal contribution margin on the increment stays above zero. Stop when it crosses. Your target is not a ROAS figure someone quoted at a conference. It is the point where the next dollar stops paying for itself at your actual gross margin.


#### Why this changes how accounts get built

Once marginal return is the metric, several familiar habits stop making sense.

1. Campaign proliferation becomes expensive. Eleven campaigns competing for one audience fragment learning and inflate CPMs without adding reach.
2. Creative supply becomes the binding constraint. Marginal return decays as frequency climbs, so the only way to hold the slope while scaling is a steady flow of genuinely new angles.
3. Budget changes become experiments. Each increase carries a pre-registered hypothesis and a stop rule, so a failure produces a decision rather than an argument.
4. Channel cuts get faster. When a channel fails its own test, it gets cut that week, not at the quarterly review, by which point the money is gone.


#### The uncomfortable part

Measuring marginal return honestly usually reveals that a brand's true efficient spend ceiling is lower than its current spend. That is a difficult finding to deliver and a more difficult one to hear.

But the alternative is worse. A business scaling on average ROAS is not growing on the strength of its advertising. It is growing on the strength of its reporting, and reporting does not deposit funds.


---

### Your retention programme is a discount programme wearing a disguise

URL: https://fm.digital/insights/retention-is-not-discounting · Published: 2026-05-28 · Author: Adam Palmer, President

**Key takeaway.** A retention programme is only producing incremental value if it moves the second-order rate. Revenue attributed to discounted email campaigns is frequently harvested rather than created, meaning the programme is converting buyers who would have purchased at full price anyway.

Run this test on your own programme. Take your next promotional send and split the list: half get the usual 15% off, half get the identical email with the offer removed and the product story strengthened. Then compare revenue per recipient, net of discount.

In roughly two out of three brands we have run this with, the no-discount version wins on margin. Sometimes it wins on revenue outright. The discount was not persuading anyone. It was charging the business for permission to sell to people already intending to buy.


#### The metric that separates the two

Attributed email revenue cannot distinguish creation from harvesting. Second-order rate can. What fraction of a given month's new customers place a second order within ninety days? That number is difficult to move, immune to attribution games, and directly sets the customer acquisition cost your business can afford.

Move second-order rate from 22% to 31% and you have not just added revenue. You have raised your CAC ceiling enough to unlock paid channels that were previously unaffordable. That is what a retention programme is for.


#### What actually moves it

- Post-purchase education that increases the odds the first product succeeds. Most non-repeat customers did not dislike you. They never got the result the product promised.
- Replenishment timing based on actual product consumption rather than a round number of days somebody guessed in a planning meeting.
- A second-purchase recommendation informed by what the first purchase implies, not by what has the highest margin this month.
- Removing friction from cancellation and returns. Counter-intuitive, reliably effective: anxiety about being trapped suppresses first purchases more than easy exits suppress retention.
- Recognition instead of discounts for high-value customers. Early access outperforms 10% off, and costs nothing.


#### The transition is uncomfortable

Weaning a list off discounts produces a visible revenue dip for four to eight weeks while trained buyers wait for the offer that is not coming. Most teams lose their nerve in week three and reinstate the code, which teaches the list that waiting works and makes the next attempt harder.

> A discount-dependent list is not an asset. It is a liability with a good open rate.

Hold through the dip and the curve resolves higher, with materially better margin underneath it. That is easier to do with the cohort data in front of you, which is the real argument for measuring properly before changing anything.


---

### Objection mapping: why your CRO tests never reach significance

URL: https://fm.digital/insights/objection-mapping · Published: 2026-04-14 · Author: Adam Palmer, President

**Key takeaway.** Conversion tests fail to reach significance most often because they test surface elements rather than unanswered purchase objections. Objection mapping, which ranks the actual reasons buyers hesitate using support tickets, exit surveys and session replay, identifies changes large enough to detect.

A test that moves conversion by 0.4% needs an implausible amount of traffic to distinguish from noise. A test that moves it by 12% resolves in a fortnight. Most CRO programmes stall not because the statistics are hard, but because they keep proposing changes too small to be detectable.

The fix is not more tests. It is testing a bigger idea, and bigger ideas come from research rather than from a best-practice checklist.


#### Where the real objections hide

Five sources, in descending order of how often they are ignored:

1. Support tickets. Every pre-purchase question that reaches a human is a question your page failed to answer. Read a thousand of them and the ranking writes itself.
2. Post-purchase surveys. Ask new customers what nearly stopped them. People who bought will tell you what the people who left could not.
3. On-site exit polls. One question, triggered on exit intent on high-value pages. Crude, cheap, and consistently revealing.
4. Review sentiment, including competitors' one-star reviews. The fears expressed there are the fears your page must pre-empt.
5. Session replay, watched with a question in mind rather than as ambient entertainment. Where does the cursor hesitate? What gets re-read?


#### Turning that into a ranked map

Group what you find into distinct objections, attach an estimated volume to each, then attach a confidence score to whether your page currently addresses it. The top three items on that list are your next three tests, and they will not be about button colour.

**Give every module a job.** Each section on a revenue page must do one of four things: establish the promise, prove it, remove a risk, or ask for the order. If a module cannot name its job, delete it. We have removed testimonial carousels nobody had ever advanced and gained conversion from the removal alone.


#### Then test it properly

Pre-register the hypothesis and the primary metric before launch. Run a power calculation so you know the detectable effect and the runtime in advance. Use sequential testing so you can stop early without inflating false positives. And read results in revenue per session, not conversion rate, because a conversion win that quietly sinks average order value is a loss in a costume.

One more discipline that pays for itself: keep a decision log of every test, including the losers, with the hypothesis and what you concluded. After a year that log is a genuine competitive asset, because it describes your specific customers rather than a generic buyer someone wrote a blog post about.


---

### Stop fixing attribution. Start triangulating.

URL: https://fm.digital/insights/attribution-triangulation · Published: 2026-03-05 · Updated: 2026-06-11 · Author: Adam Palmer, President

**Key takeaway.** Reliable marketing measurement requires three tiers that check one another: server-side data capture for daily operations, media mix modelling for quarterly budget allocation, and incrementality experiments to settle disputes. No single attribution platform can replace the combination.

Every quarter a brand asks us which attribution platform to buy, expecting a recommendation. The honest answer is that the question contains a false premise: no platform can observe what it needs to observe, because the underlying signal was removed by privacy changes and is not coming back.

What works instead is triangulation: three imperfect instruments whose errors point in different directions.


#### Tier one: clean capture, for the day-to-day

Server-side event collection through Conversions API and Enhanced Conversions, consent-aware identity resolution, and raw events stored in your own warehouse rather than in a vendor's interpretation of them. This tier answers operational questions: did something break overnight, is this campaign pacing, is match quality degrading.

What it cannot tell you is what was incremental. Treat it as telemetry, not truth.


#### Tier two: a model, for quarterly allocation

Media mix modelling reads aggregate spend against aggregate outcomes and estimates each channel's contribution, including the offline and upper-funnel activity no pixel ever sees. It is the right instrument for the question 'how should next quarter's budget be split'.

It earns its cost above roughly $500,000 per month in spend with two years of clean history. Below that, the confidence intervals are wide enough to drive a strategy through.


#### Tier three: experiments, for the arguments

Geo-holdouts, PSA lift tests and branded-search blackouts produce the closest thing to causal truth available in marketing. They are slower and more expensive per question, so reserve them for questions that matter: is this channel incremental at all, what is our true brand-search contribution, does upper-funnel video pay for itself.

**The rule that keeps everyone honest.** When the tiers disagree, the experiment wins, then the model, then the platform. Write that hierarchy down before you have a dispute, because agreeing on it afterwards is considerably harder.


#### And reconcile to finance

Whatever you build, tie it to bank-deposited revenue and real gross margin, after returns, shipping and payment fees. Marketing dashboards that cannot be reconciled to the P&L get quietly disbelieved by everyone who matters, and once that happens the reporting stops influencing decisions at all.


---

### Which schema actually gets used

URL: https://fm.digital/insights/schema-for-service-businesses · Published: 2026-02-10 · Author: Adam Palmer, President

**Key takeaway.** For service businesses the structured data that earns its place is a small set: an accurate organization or local business node, explicit service definitions, FAQ markup that matches visible content, and breadcrumbs. Volume of markup matters far less than the facts stated inside it.

Structured data has a reputation as a checkbox exercise, mostly because it is usually implemented as one. A plugin emits fourteen node types, none of them says anything a machine did not already know, and nothing improves.


#### The point is the facts, not the markup

Schema is a way of stating things unambiguously. If the statement is empty, the markup is empty. An Organization node carrying a name and a logo tells a retrieval system almost nothing. The same node carrying service area, price range, the services offered and what the company knows about is a set of claims that can actually be matched against a question.


#### The short list

1. Organization, or LocalBusiness where there is a physical premises. Include address, service area, contact points, founding date and the topics you actually work in.
2. Service, one per offering, with a plain description and the provider linked back to the organization by id.
3. FAQPage, but only where the questions and answers are visible on the page. Marking up content a human cannot see is a violation and a risk.
4. BreadcrumbList, which is cheap and helps establish site structure.
5. Article or BlogPosting on editorial content, with a real author and honest dates.

**Use one graph and link by id.** Emit a single @graph per page with the nodes referencing each other through @id, rather than several disconnected script blocks. It lets a parser resolve the relationships between your organization, your services and the page it is looking at, which is the whole reason for doing this.


#### What to skip

Review and AggregateRating markup on your own pages invites trouble and is widely ignored where self-serving. Deeply nested product markup on a service site rarely matches reality. Anything describing content that is not on the page is a liability, not an asset.


#### And check it against what you say elsewhere

The value of structured data collapses if it contradicts your other sources. If schema says one service area and your business profile says another, you have not clarified anything, you have added a third opinion. Reconcile first, then mark up.


---

### The quote is the product

URL: https://fm.digital/insights/price-transparency-as-conversion · Published: 2026-01-13 · Author: Adam Palmer, President

**Key takeaway.** Withholding pricing in high-ticket categories does not delay the price question, it moves it to a competitor or a forum. Publishing an honest range with the factors that move it improves lead quality and conversion, and reduces time wasted on unqualified appointments.

The most common unanswered objection on a high-ticket service page is the price. Not the exact price, which nobody expects. The range. The order of magnitude. Is this a four thousand dollar job or a forty thousand dollar one.


#### The question does not go away

Sales teams resist publishing numbers for reasons that sound right: every job is different, a number without context invites the wrong comparison, competitors will see it. All true, and none of it stops the buyer needing an answer.

What actually happens is that they leave to find it. A forum, a competitor who publishes, a national aggregator with numbers that have nothing to do with your market. They come back anchored to a figure you did not choose, or they do not come back.


#### What publishing well looks like

- Give a real range with both ends, and be honest about the top. A range that stops short of where jobs actually land destroys trust at the quote.
- Name the three or four factors that move the number, so the buyer can place themselves inside it.
- Show a worked example. A described job with a price teaches more than a range alone.
- Say plainly what is included and what is not, because that is where most quote disputes start.

**The lead quality effect.** Publishing a range filters people who were never going to proceed. Total lead volume usually falls. Booked appointments and close rate usually rise, and the sales team stops burning afternoons on jobs that were priced wrong from the first call. Judge the change on cost per sold job, never on cost per lead.


#### When not to

If your pricing is genuinely uncompetitive, transparency will cost you. That is worth knowing rather than hiding from, because the buyers you were winning without it were the ones who did not compare, and that is not a durable position.


---

### Budget pacing when the sale lands nine weeks later

URL: https://fm.digital/insights/pacing-a-nine-week-sales-cycle · Published: 2025-12-09 · Author: Adam Palmer, President

**Key takeaway.** When revenue lands weeks after the spend that caused it, in-month performance figures are structurally incomplete and drive premature cuts. Pacing decisions should be made against cohort-based reporting, where each week of spend is judged once its full sales cycle has elapsed.

Here is a mistake that costs lead generation businesses more than any creative decision. It is the third week of the month, revenue looks thin against spend, and somebody cuts the budget. The revenue from that spend was always going to arrive in six weeks. Cutting it guarantees a thin month later, which triggers another cut.


#### Calendar months are the wrong unit

A monthly report compares this month's spend to this month's revenue. In a business with a nine-week average cycle, those two numbers are describing different cohorts of buyers. The comparison is not slightly noisy, it is structurally meaningless.


#### Report by cohort instead

Group leads by the week they arrived, then track each cohort's outcomes as they mature. Week 34's leads are judged once week 34 has had its full cycle, not at the end of the calendar month that happened to contain it.

1. Fix the reporting unit to the week the lead was created, never the week the sale closed.
2. Build a maturation curve from history, so you know what share of a cohort's eventual sales have landed by day 14, 30, 60 and 90.
3. Use that curve to forecast an immature cohort rather than judging it as finished.
4. Only treat a cohort as final once it has passed roughly 90% of its expected maturation.

**The practical benefit.** A maturation curve lets you read a two-week-old cohort honestly. If 22% of eventual sales normally land by day 14 and this cohort is at 12%, you have an early signal worth acting on. Without the curve you have a number that looks alarming every single time.


#### Then hold the budget steady

Sharp budget swings are especially damaging in long-cycle categories, because the consequences of each swing arrive after the next decision has already been made. Steady spend with deliberate stepped changes, judged on mature cohorts, outperforms reactive pacing by a wide margin, mostly by avoiding self-inflicted whiplash.


---

### Your reviews are now feeding two different systems

URL: https://fm.digital/insights/reviews-are-a-ranking-factor-twice · Published: 2025-11-11 · Author: Adam Palmer, President

**Key takeaway.** Review count and rating influence traditional local ranking, while the language inside reviews supplies the evidence AI systems use to match a business to a specific question. A review programme optimised only for star rating produces text with little retrievable detail.

Most review programmes are built to raise a number. Get more reviews, keep the average above 4.7, respond to the bad ones. That is sound, and it addresses only half of what reviews now do.


#### Two systems, two appetites

Conventional local ranking cares about count, average, recency and velocity. Those are signals, and they are largely quantitative.

A retrieval system reads the words. When someone asks which company handles a particular job in a particular place, the assistant is looking for text that connects a business to that job and that place. A hundred reviews saying "great service, highly recommend" supply almost nothing to match against.


#### How to get more useful text without coaching

You cannot tell customers what to write, and you should not try. You can ask better questions.

- Ask at the moment the outcome is concrete, not a month later when it has blurred into a general good feeling.
- Prompt with a question rather than a request. "What did we do for you?" produces specifics. "Leave us a review" produces adjectives.
- Ask the technician or project lead to request it, because the review then tends to describe the actual job.
- Never script it, incentivise it or gate it by sentiment. Beyond being against every platform's rules, it produces uniform text that reads as synthetic to both systems.

**Respond for the reader after next.** A reply is public text you control, attached to a review you do not. Use it to state the specific service and location plainly, without keyword stuffing. It is one of the few places you can add retrievable facts to a third-party source legitimately.


#### And watch where they are

Different assistants lean on different platforms, and the mix varies by category. Run your prompt set, read which sources get cited in the answers about your category, and concentrate the programme there rather than spreading it across every directory that asks.


---

### How assistants answer "who should I hire near me"

URL: https://fm.digital/insights/who-should-i-hire-near-me · Published: 2025-10-14 · Author: Adam Palmer, President

**Key takeaway.** When an AI assistant recommends a local service business it assembles evidence from review platforms, directories, the business profile and the company's own pages. Consistency of name, address, service area and specialism across those sources determines whether the assistant can name the business confidently.

Local search used to mean the map pack. Increasingly the first answer a buyer sees is a paragraph naming two or three companies, assembled by a model that never showed them a map at all.


#### What the model is drawing on

For a local service question the evidence set is broader than your website and mostly not controlled by you. Review platforms, directories, your business profile, local press, forum threads, and then your own pages as corroboration rather than as the primary source.

That ordering matters. A model weights what others say about you above what you say about yourself, for the same reason a buyer does.


#### Consistency is the whole game

The single most common reason a local business cannot be recommended confidently is that the facts about it disagree across sources. Three variants of the company name, two phone numbers, a service area that says county on one profile and city on another, a specialism claimed in one place and absent everywhere else.

Faced with that, an assistant hedges, and hedging means naming a competitor whose details line up.

1. Fix the name, address, phone and service area to one exact form everywhere they appear.
2. State the specialism explicitly and identically. "Custom home building and remodelling in Durango" is retrievable. "Quality craftsmanship since 2006" is not.
3. List service areas as places, not as a radius nobody can parse.
4. Keep the business profile categories accurate and narrow rather than aspirational.
5. Publish the boring facts on your own site too: licence numbers, insurance, areas served, what you do not do.

**Reviews are evidence, not decoration.** Volume, recency and the actual words matter. A hundred reviews that say "great job" establish that you exist. Reviews that mention the specific service, the town and the outcome establish what you do, and that is the text a retrieval system can use to match you to a question.


#### Measuring it

Build a prompt set the way a buyer would phrase it, including the town, and run it on a schedule. Record whether you are named, in what position, and what the assistant says about you, because an inaccurate mention is a different problem from no mention and needs a different fix.


---

### Every form field is a tax you are charging the buyer

URL: https://fm.digital/insights/form-fields-are-a-tax · Published: 2025-09-09 · Author: Adam Palmer, President

**Key takeaway.** Each additional form field reduces completion, so a field only earns its place if the qualification it provides is worth more than the leads it costs. Fields the sales team can establish in the first thirty seconds of a call are almost never worth collecting up front.

Form fields get added in meetings. Somebody says it would be useful to know the property type, nobody objects, and the form grows. The cost lands somewhere nobody in that meeting can see.


#### The trade is real, so make it deliberately

Every field costs completions and buys qualification. That is a genuine trade and sometimes worth making. The problem is that it is almost never made consciously, so forms accumulate fields the way a garage accumulates boxes.

The test for each field: can the sales team establish this in the first thirty seconds of the call? If yes, it does not belong on the form. You are charging every prospect to save your team half a minute.


#### Fields that usually earn their place

- Anything that determines whether you can serve them at all, such as a postcode outside your service area.
- Anything that routes the lead to a different team or process.
- Anything genuinely required to produce a meaningful quote in categories where the buyer expects one.


#### Fields that usually do not

- How did you hear about us. Ask it post-sale. It is unreliable on a form and it costs you leads.
- Budget, early. It suppresses the buyers who have not priced the job yet, who are often the most winnable.
- Preferred contact time, when you could simply ask on the call.
- Anything marked optional. If it is optional it is clutter, and clutter reads as length.

**Before deleting, measure.** Instrument field-level drop-off so you know which field loses people rather than guessing. The offender is frequently not the one everybody suspects, and it is often a field that looks harmless but reads as intrusive, such as a phone number requested before any value has been offered.


#### The exception worth knowing

Sometimes a longer form is correct. If your sales capacity is the constraint rather than lead volume, qualification is doing useful work and a shorter form just buries the team in enquiries they cannot service. Know which constraint you are actually under, because the right answer flips entirely depending on it.


---

### Selling in a six-week window

URL: https://fm.digital/insights/seasonal-demand-windows · Published: 2025-08-12 · Author: Adam Palmer, President

**Key takeaway.** In a compressed seasonal window there is no time to learn, so testing, creative production and audience building must happen out of season. The in-season job is execution against decisions already made, with budget pacing built around a demand curve rather than a monthly average.

A brand doing most of its year in six weeks has a different problem from one selling evenly. It is not a volume problem. It is that the learning period and the selling period are the same period, and that is fatal.


#### You cannot test in season

An algorithm needs conversions to learn. A test needs runtime to reach significance. In a six-week window, a creative test that resolves in three weeks has already cost you half the season, and the answer arrives when it can no longer be used.

So the testing has to happen out of season, on lower volume, against proxy signals. It is less precise and it is the only option that leaves you entering the window with decisions already made.


#### What the quiet months are actually for

- Building the audience you will sell to, so the season opens against warm demand rather than cold traffic at peak prices.
- Producing and pre-testing creative, because production time in season is time you do not have.
- Fixing the site, the feed and the checkout while a mistake costs almost nothing.
- Growing the email and SMS file, which is the only asset that lets you open the season without buying every impression.
- Agreeing the pacing plan and the stop rules before anyone is under pressure.

**Pace to the curve, not the month.** Auction prices rise steeply into a seasonal peak. A flat daily budget systematically overpays late and underbuys early. Build the pacing plan from last year's demand curve, front-load deliberately, and decide in advance what you will do if the curve arrives a week early.


#### And decide what happens after

The eleven quiet months are also where lifetime value is won or lost. A seasonal buyer who hears nothing until the next peak is a stranger being reacquired at full price. One that stays lightly engaged is a customer, and the difference between those two is the difference between a business that grows and one that starts again each year.


---

### Call tracking without wrecking your attribution

URL: https://fm.digital/insights/call-tracking-without-breaking-attribution · Published: 2025-07-15 · Author: Adam Palmer, President

**Key takeaway.** Businesses whose customers phone rather than submit forms systematically under-measure the channels that drive calls. Dynamic number insertion tied to session source, with outcomes written back to the CRM, closes the gap without the duplicate-counting that usually follows a call tracking rollout.

A remodelling client was about to cut a channel that produced almost no form fills. It was, in fact, their best source of booked jobs. Every one of those buyers had picked up the phone, and nothing in the reporting knew it.


#### The blind spot is structural

For a considered purchase, phoning is often the rational choice. The buyer has a question a form cannot answer, wants a sense of whether they like you, or simply prefers talking. The higher the ticket and the older the buyer, the more of your demand behaves this way.

If the phone number on your site is static, every one of those conversions is invisible to the channel that caused it. Your reporting then systematically favours channels that produce form fills, which is a preference for a behaviour rather than for revenue.


#### Doing it properly

1. Use dynamic number insertion so the displayed number varies by session source. A single tracked number tells you a call happened, not where it came from.
2. Pass the session identifier and click identifier into the call record, so a call can later be tied back to a campaign rather than just a medium.
3. Write the call outcome back to the CRM. A call is not a sale, and counting ringing phones as conversions is its own distortion.
4. Set a sensible minimum duration before a call counts. Thirty to sixty seconds filters wrong numbers and hang-ups.
5. Deduplicate against forms. A buyer who submits a form and then calls is one lead, and double counting will flatter whichever channel you least want flattered.

**Keep the number consistent for humans.** Dynamic insertion changes the number a visitor sees. Make sure your Google Business Profile, invoices and van livery still show the primary number, and that all tracked numbers route to the same place. The goal is measurement, not a maze.


#### Then send the outcome back to the platforms

Tracked calls only change your marketing once the ad platforms know about them. Import qualified calls as offline conversions alongside closed deals, and the bidding starts optimising toward the behaviour that actually produces revenue in your category.

Until then you are running an auction strategy built on the half of your demand that happens to like typing.


---

### The aged lead file is the cheapest revenue in the building

URL: https://fm.digital/insights/aged-lead-file · Published: 2025-06-10 · Author: Adam Palmer, President

**Key takeaway.** In high-consideration categories a substantial share of buyers purchase months after first enquiry, often from whichever company is still present when they decide. Because acquisition was already paid for, a structured reactivation programme against closed-lost leads typically returns more per dollar than new traffic.

Ask a sales manager how long they chase a lead before marking it dead. The answer is usually somewhere between a week and a month. Then ask how long buyers in the category actually take to decide. The answer is usually months.

That gap is where a large amount of paid-for demand quietly goes to waste.


#### Dead is a staffing decision, not a buyer state

A lead gets marked closed-lost when the rep stops calling, which is a fact about your process rather than about the buyer. Plenty of those people go on to buy. They just do it later, from whoever happened to be in front of them at the moment the decision firmed up.

The economics here are unusually good because the expensive part already happened. You paid to acquire that contact. Reaching them again costs an email.


#### How to run the programme

1. Segment the closed-lost file by how far they got. Someone who had a quote is a different prospect from someone who never answered.
2. Lead with something other than a discount. New information works better: a change in pricing, a new product, a finance option, a case from their area.
3. Give the rep a reason to call that is not "just checking in", because that call has already failed once.
4. Measure it as its own channel with its own cost per sale, so it competes fairly with new traffic for budget.
5. Run it on a cadence, not as a one-off blitz, because demand firms up continuously rather than in the week you decided to email.

**One caution.** Reactivation reaches people who did not buy and may not want to hear from you. Respect unsubscribes rigorously, keep the frequency low, and suppress anyone who asked to be left alone. A reactivation programme that damages your domain reputation costs more than it returns.


#### What it tells you

The return is useful. The diagnostic is more useful. If reactivation produces a lot of revenue, it means your follow-up window is too short and you have been discarding demand at the front of the process. Fix that, and the aged file gets smaller and the new-lead close rate goes up, which is a better outcome than a permanently lucrative reactivation programme.


---

### You are probably paying to win customers you already had

URL: https://fm.digital/insights/branded-search-tax · Published: 2025-05-13 · Author: Adam Palmer, President

**Key takeaway.** Branded search campaigns report strong return on ad spend while largely capturing buyers who would have arrived through organic results anyway. A scheduled blackout test measures the true incremental contribution, and frequently shows a large share of that spend is defensive rather than productive.

Every account we inherit has a branded search campaign posting the best numbers in the account. Eight, twelve, twenty times return. It is also, reliably, the least incremental line in the budget.


#### What branded search actually does

Someone types your company name. They already know who you are, which means something else in your marketing did the work: a van, a neighbour, a mailer, a Meta ad, a job sign. Branded search is the last touch before the click, and last touch gets all the credit in most reporting.

That does not make it worthless. It makes it unmeasured, which is different, and much more expensive.


#### The test that settles it

Pause branded search for a defined period and watch total conversions from branded terms, paid plus organic combined. If total volume holds roughly steady while paid spend goes to zero, organic absorbed the demand and you were paying for clicks you would have received free.

1. Run it for at least two full weeks, ideally four, to cover weekly seasonality.
2. Measure total branded conversions, never just the paid line, or you will conclude the obvious and learn nothing.
3. Watch impression share on your brand terms. A competitor moving in changes the answer.
4. Repeat it once or twice a year, because the answer changes with your brand awareness and the competitive set.

**The honest caveat.** Sometimes the answer is that you should keep spending. If competitors bid on your name and intercept your buyers, defending the term is legitimate, and the blackout test will show the loss. The point is not to cut the campaign. The point is to know which of the two situations you are in rather than assuming.


#### What to do with the finding

If branded search turns out to be mostly harvesting, you have two options and both are improvements. Reallocate the spend to genuinely incremental inventory, or keep it but stop counting its revenue toward the performance of your acquisition programme, because doing so inflates the whole account and hides the real efficiency of everything else.

The second option is more common than the first, and it is the one that quietly corrects a lot of bad budget decisions.


---

### Work out what a lead is worth before you decide it is expensive

URL: https://fm.digital/insights/what-a-lead-is-worth · Published: 2025-04-15 · Author: Adam Palmer, President

**Key takeaway.** Allowable cost per lead is derived from average contract value, gross margin, the share of margin you will spend to acquire a job, and your close rate. A lead price is only expensive relative to that figure, which is specific to each business and cannot be borrowed from a benchmark.

"Our leads are too expensive" is the most common sentence in lead generation and one of the least useful, because almost nobody saying it has worked out what a lead is worth to them.


#### The calculation

Four numbers, all of which you already have.

1. Average contract value. What a closed job is actually worth, not what the good ones are worth.
2. Gross margin. After materials, labour and subcontractors, before overhead.
3. The share of that margin you are willing to spend to win a job. Most businesses land between 15% and 30%.
4. Close rate from lead to signed job, measured over a full sales cycle rather than last month.

**Worked example.** A $14,000 average job at 38% gross margin produces $5,320 of margin. Spend a quarter of that to win it and your allowable cost per sale is $1,330. At a 12% close rate, that is an allowable cost per lead of about $160. A $90 lead is cheap. A $210 lead is not, unless it closes better.


#### Why this changes the argument

Once the number exists, the conversation stops being about whether leads feel expensive and starts being about which lead sources clear the bar. Two sources at the same price routinely have close rates that differ by a factor of three, and the cheaper one is often the worse buy.

It also reframes the sales side. Lifting close rate from 12% to 15% raises your allowable cost per lead from $160 to $200, which unlocks inventory you previously could not afford. Sales training and marketing budget are the same lever viewed from different ends.


#### The number moves, so recalculate it

Average contract value drifts with your mix. Margin moves with materials. Close rate moves with the team and the season. Recalculate quarterly, and recalculate immediately after any pricing change, because a price rise quietly raises what you can afford to pay for demand.

None of this is complicated. It is just arithmetic that most businesses never do, which is why the ones that do it can outbid everyone else and still make money.


---

### Speed to lead: the cheapest conversion rate you will ever buy

URL: https://fm.digital/insights/speed-to-lead · Published: 2025-03-11 · Author: Adam Palmer, President

**Key takeaway.** Response time is one of the few levers in lead generation with a large and repeatable effect on close rate. Most of the gap between a company's intended response time and its real one is routing and staffing rather than marketing, which makes it the cheapest conversion improvement available.

Before you test another headline, measure how long it takes your team to call a new lead. Not the average. The distribution.

The average is reassuring and useless. It hides the tail, and the tail is where the money goes. A business with a twelve minute average is often really two businesses: one that answers in ninety seconds during office hours, and one that answers at nine the next morning for everything that arrives after five.


#### Why the first five minutes matter so much

A buyer who fills in a form is, for a short window, actively thinking about the problem. They have the tab open. They have not yet filled in three more forms. Reach them inside that window and you are having a conversation. Reach them tomorrow and you are interrupting someone who has since spoken to two competitors.

In high-consideration categories most buyers request several quotes. Being first is not a tiebreaker, it is frequently the whole contest, because the first credible conversation sets the frame everyone else gets measured against.


#### Where the time actually goes

When we audit this, the delay is almost never the marketing team.

1. The form posts to an inbox nobody owns outside business hours.
2. The CRM assigns round-robin to a rep who is on a job site with no signal.
3. Nobody is accountable for a lead that has been sitting for an hour, because no alert fires.
4. Phone leads and form leads live in different systems, so nobody sees the real queue.
5. Weekend leads wait until Monday, and a third of enquiries arrive at the weekend.

**Measure it before you argue about it.** Pull the last ninety days of leads with their created timestamp and their first-contact timestamp. Plot the distribution, not the mean. Almost every team we have done this with is surprised by their own data, and the surprise is what unlocks the budget to fix it.


#### The fixes, cheapest first

Route by availability rather than by rota, so a lead never lands with someone who cannot pick up. Alert on age, so a lead untouched after ten minutes escalates to whoever is free. Merge phone and form leads into one queue, because a split queue is an unmanaged queue. Then decide deliberately what happens to after-hours and weekend enquiries, whether that is an answering service, a scheduling link, or an honest automated reply that sets a time.

> Every other conversion lever costs money to pull. This one mostly costs a decision about who picks up the phone.


#### Then hold the line

Speed to lead degrades the moment nobody is watching it. Put the distribution on the same weekly report as cost per lead and close rate, and treat a slipping tail as seriously as a rising CPL. It is the same problem arriving from the other direction.


---

## Attribution

This content may be quoted and cited with attribution to Force Multiplier Digital (https://fm.digital).