1. Executive Summary

FindWebSiteClients (FWC) is the client-acquisition engine behind two of Annette's revenue businesses, SEOBeliever (AI-age SEO and AEO consulting) and Vitalency (B2B MedSpa Brain OS). It finds local businesses that are losing ground online, either because their website is ugly, outdated, or missing, or because the AI answer engines never mention them, then wins them as clients using a "do-it-first" play: build the proof before the pitch.

Two wedges set FWC apart from every existing prospecting tool. First, an AI-visibility audit (the Citation Snapshot) that checks whether ChatGPT, Claude, Perplexity, Gemini, and Grok actually recommend a business over its competitors when someone asks a natural local-intent question. Nobody else in this category measures that. Second, an auto-built redesign that is a real, working Astro webpage from the moment it's generated, not a picture of one, so the screenshot we send in outreach and the site we deploy on a "yes" are the same code.

We've already built and run the engine. Phase 1 through 3 are live: a $0 qualification pipeline, a Claude vision scorer, and the full CITE prompt battery across five AI engines, backed by 157 automated tests. We've run it for real. A single Boulder sweep scanned 1,938 local businesses and flagged 556 quick-win candidates for the cost of 150 free Google Places API calls. We've produced full Citation Snapshot audits, redesign mockups, and geo-grid rank reports for real Boulder prospects across medspa, dermatology, dental, and orthodontic verticals, and we've verified, with real DataForSEO data, exactly which additional verticals (TRT and hormone clinics, IV therapy, vein clinics, and several home-services categories) still have wide-open Map Pack competition worth pursuing.

FWC doesn't charge its own customers today. It's the acquisition front-end for SEOBeliever's ladder (free Citation Snapshot through an $18,000 Buildout with $1,500 a month Monitoring) and for Vitalency's B2B medspa engagements. Every dollar it earns shows up as a closed deal in one of those two businesses. The architecture is deliberately "internal-first, SaaS-ready": clean module seams mean the same qualification and AEO-audit core could later be licensed to other agencies as a standalone product, without a rewrite, if that path makes sense.

The near-term goal isn't to scale prospecting for its own sake. It's to convert the pipeline we already have (real, review-scored, non-competing prospects across at least four verticals in Boulder) into the first handful of paid SEOBeliever and Vitalency engagements, and to prove the do-it-first play works before we decide whether to expand it into more metros or spin it out as its own product.

2. Problem and Opportunity

The Local Business Owner Doesn't Know What's Actually Broken

Most local business owners know, vaguely, that their website looks dated or that a competitor seems to show up more online. What they don't know, because almost nobody is measuring it yet, is that AI answer engines now sit inside how their next customer finds a provider, and that they can be completely invisible there even while winning on Google's Map Pack. Our own research on one Boulder dermatology practice found exactly this: it wins the local Map Pack outright but scores dead last on AI-citation share among its named competitors. A business owner who never asked "does ChatGPT recommend me" has a real, fixable problem they don't know exists.

Generic Prospecting Tools Sell a List, Not Proof

The two incumbent tools in this space, GoHighLevel's prospecting module and uglysitescraper.com, both stop at "here are some businesses with bad websites." Neither measures AI-visibility at all, and neither builds anything real before the pitch. That leaves the actual conversion work (the mockup, the audit, the specific reasons a site looks dated) for the seller to do by hand, one prospect at a time, which is exactly why cold outreach in this space converts so poorly.

Annette's Edge Is a Combined AI-Plus-SEO Practice

SEOBeliever already runs a proven, paid methodology (the CITE prompt battery and the Citation Share Index) as a lead magnet. FWC's job is to run that same methodology at scale, across every qualified prospect in a market, and to pair it with a redesign that's already real code. That combination, AI-visibility auditing plus a genuinely deployable mockup, is the whitespace nobody else occupies.

The Opportunity: Industrialize a Play That Already Works

Annette has already closed real value with this exact do-it-first approach elsewhere (Bone Voyage's domain rating move from 0.9 to 62, the Rebuilding Seminars methodology, Ajijic). FWC's opportunity is to turn that one-at-a-time, founder-led process into a repeatable pipeline: qualify a market, generate the proof for every real prospect in it, and hand the sales conversation a finished asset instead of a cold pitch.

3. Solution: What FindWebSiteClients Does, End to End

FWC runs as a Python engine (findclients/) that Annette drives from the CLI today, built with clean module seams so a web or SaaS layer could wrap the same core later without a rewrite.

1. Source prospects. Pull local businesses by niche and metro. Today via a free local-file source and the GBP Prospect Finder Chrome extension (a companion tool that hits Google Places Text Search directly, no paid API required for the first ~1,000 calls a day). Paid sources (Outscraper primary, Google Places fallback) are interface-complete and waiting on a greenlight.
2. Qualify for free. Every website gets checked for reachability, SSL validity, mobile-friendliness, load time, page weight, footer-copyright staleness, and CMS fingerprint, no paid API required. These roll into a 0 to 100 ugliness score plus a concrete, non-insulting issues list that doubles as cold-outreach material ("their footer still says 2013", "no SSL certificate").
3. Score the aesthetics. A Claude vision pass reads the actual homepage and gives specific, human-readable reasons it looks dated, a judgment the free signals alone can't make.
4. Run the Citation Snapshot (Wedge 1). The CITE prompt battery, SEOBeliever's own method, queries ChatGPT, Claude, Perplexity, Gemini, and Grok with real local-intent questions and scores the business's Citation Share Index against three named competitors, geo-located by text, never by IP, with refusals and empty answers reported as data rather than dropped.
5. Map the local rank grid, when the report calls for it. RankGridEngine, a sister tool, sweeps an NxN grid of coordinates around the business and renders a heat map of where it actually ranks in Google Maps, at about $0.045 a business.
6. Build the redesign (Wedge 2). For qualified prospects that clear the bar, generate a real working Astro site from their public business data (name, services, hours, reviews) and screenshot it for outreach. The redesign is never regenerated from scratch on a "yes": the same project deploys straight to Cloudflare Pages with the prospect's domain and a working contact form.
7. Draft the opener, never send it. Every outreach message is built from that specific prospect's real, evidence-grounded issues, then held for Annette's explicit "send." Nothing goes out automatically.
8. Hand off the win. A closed deal gets logged into SEOBeliever's or Vitalency's own pipeline and serviced through their existing ladders. FWC's job ends at the handoff.

4. Positioning and Differentiation

The category has two entrenched players, and FWC sits in the gap between what they do and what an AI-plus-SEO operator can actually prove.

• Vs. GoHighLevel's prospecting module: Deep, patented lead scoring and a full CRM, but it's platform bait. It gets you into a $97 to $497 a month system with a $29 per sub-account add-on, and it has no AI-visibility check and no generated mockup at all. It's built to lock you into GoHighLevel, not to hand you a finished asset.
• Vs. uglysitescraper.com: A basic AI ugliness score and a CSV export, at $29 to $119 a month for 250 to 1,000 credits. No AI-visibility measurement, no mockup, no deploy path. It sells a list, not a pitch.

FWC's differentiation:

1. The AEO wedge nobody else has. We're the only tool in this category that measures whether ChatGPT, Gemini, Perplexity, Claude, or Grok actually recommend the business, and it's the automation of a lead magnet SEOBeliever has already proven converts.
2. The mockup is real code, not a picture. Every other "here's what your site could look like" pitch is a render someone still has to build if the prospect says yes. Ours deploys as-is.
3. No lock-in. FWC is standalone and cheap to run, roughly 1 to 4 cents a fully scored lead. There's no platform Annette has to migrate her business onto.
4. Vertical depth on the health and wellness side. SEOBeliever's one-client-per-vertical rule and Vitalency's medspa focus mean FWC isn't chasing every category; it's built to find the single best non-competing prospect in medspas, dermatology, cosmetic dentistry, HRT and TRT clinics, and IV therapy, verticals where the ticket size and the AI-citation gap both justify the work.

5. Target Market and Beachhead (ICP)

First metro: Boulder, Colorado, chosen because it's where SEOBeliever and Vitalency already operate and where FWC has already run real, verified sweeps.

First vertical ICP, verified with real data (2026-07-09): we ran a DataForSEO sweep of 38 candidate Boulder business types (cost: $0.076) specifically to find where the Map Pack is still open rather than already dominated by three or more 100-plus-review incumbents. The clearest fit for SEOBeliever's health-and-wellness focus: TRT and hormone clinics (7 businesses over 100 reviews) and IV therapy and wellness lounges (8 businesses over 100 reviews). Both are real, sellable, non-hospital businesses with either review strength or clear review-velocity headroom, and neither vertical is defended the way dentistry, medspas, or veterinary care already are in Boulder.

A parallel, already-productized vertical: nail salons. A 6-thread research pass validated the vertical (45% of consumers now use AI tools like ChatGPT for local-business recommendations, up from 6% a year earlier, per BrightLocal 2026, and only 14% of marketers measure AI citations at all) and produced a concrete offer, spun into its own feeder project, NailSalonMarketing, which supplies FWC with vertical-specific research and pairs the free Citation Snapshot with a cheap missed-call-text-back and review-engine bundle sized for a one-to-five-chair salon.

Why this ICP, concretely:

• These are single-location, owner-operated businesses with real ticket sizes (TRT and hormone therapy, IV drips, cosmetic dental work, medspa treatments), which justifies SEOBeliever's $1,800 to $18,000 ladder.
• Our own Boulder data shows the review-count "reputation game" and the AI-citation gap are different problems. Evolution Dermatology, one of our real prospects, wins the Google Map Pack outright but scores dead last on AI-citation share among named competitors. High reviews don't close the Citation Gap.
• SEOBeliever's one-client-per-vertical, one-per-market exclusivity rule means FWC's job isn't to hand Annette a pitch-list of competing businesses. It's to select the single best non-competing prospect per vertical per metro, which is exactly what the verified sweep did.
• Wide-open adjacent verticals (vein clinics at zero businesses over 100 reviews, EV charger installers at zero, financial advisors at zero, tattoo removal at one, radon mitigation at two) exist as a second-tier expansion list for standalone web-build work outside the health-and-wellness core, once the first vertical is proven.

6. Market Size

All figures below are directional and exist to size the opportunity, not to forecast it.

• Total Addressable Market (TAM): U.S. spend on local-business digital marketing services (SEO, web design, local listings, reputation management) is commonly estimated in the tens of billions of dollars annually. This isn't FWC's real market; we aren't a horizontal agency competing for all of it.
• Serviceable Addressable Market (SAM): The slice that's actually relevant is local, single-location, high-ticket health-and-wellness businesses (medspas, dermatology, cosmetic and general dentistry, HRT and TRT clinics, IV therapy, and adjacent categories like nail salons) in metros where SEOBeliever's methodology and Vitalency's medspa focus apply. A conservative estimate, based on the number of qualifying businesses per metro (our own Boulder sweep alone surfaced 556 quick-win candidates out of 1,938 businesses scanned) times the number of metros this model could realistically expand into, puts the SAM in the low hundreds of millions of dollars annually across North America.
• Serviceable Obtainable Market (SOM): Our real constraint is SEOBeliever's exclusivity rule (one client per vertical per market), not lead volume.

• Assumption: we prove the model in Boulder across 4 to 6 verticals in Year 1, then expand to 8 to 10 metros over 3 years with 15 to 25 verticals active per metro at full maturity.
• Value per client: an average blended client value of $3,000 to $5,000 in initial engagement plus a growing share on $1,500-a-month Monitoring.
• Conclusion: the 3-year SOM is realistically $200,000 to $400,000 in annual attributed revenue to SEOBeliever and Vitalency combined, a small, deliberately-paced slice built one non-competing client at a time rather than a volume play.

7. Revenue Model and Pricing

FWC itself doesn't bill anyone. It's the acquisition engine for two revenue businesses, and its "pricing" is really about how efficiently it produces a paying client for them.

How the money actually flows: a qualified, non-competing prospect that FWC surfaces converts into a SEOBeliever engagement on SEOBeliever's existing ladder: free Citation Snapshot (the give-value-first proof), then a $1,800 Audit, a $4,500 Sprint, a $9,500 Blueprint, $1,500 a month Monitoring, or an $18,000 Buildout. A qualified medspa prospect can instead route to Vitalency's B2B medspa engagement model. FWC's redesign mockups (Wedge 2) support standalone web-build work outside either ladder, for prospects whose fit is a new site rather than an AEO or SEO engagement.

Why this shape, not a direct FWC price: SEOBeliever's ladder is already validated and its vocabulary (Citation Gap, Citation Triangle, Citation Share Index) is the client-facing product. Building a second, parallel pricing model inside FWC would confuse the funnel and risk cannibalizing the $1,800 Audit by giving its findings away for free. The free Citation Snapshot intentionally diagnoses the gap and never prescribes the fix; the fix is what SEOBeliever sells.

Future optionality (illustrative, not committed): because the architecture is internal-first, SaaS-ready, a licensed version of the qualification-plus-AEO-audit engine could eventually be sold to other agencies or consultants at a monthly seat or usage price. That's a real option the module boundaries preserve, not a current plan, and it would need its own build-vs-license decision before pursuing.

8. Unit Economics

FWC's economics are almost entirely about how cheaply it can produce a genuinely qualified, sellable prospect, because the actual revenue is captured downstream by SEOBeliever and Vitalency.

• Cost to find a prospect: roughly 1 to 4 cents for a fully scored lead once paid enrichment is wired in (Outscraper at $0.003 to $0.012 a record, Claude Haiku vision at a fraction of a cent, self-hosted Playwright screenshots at $0). Today, using the free GBP Prospect Finder path, it's effectively $0: our real Boulder sweep processed 1,938 businesses on 150 free Google Places calls.
• Cost to add the AEO layer: the CITE Snapshot itself runs on OpenRouter calls across five engines, a few cents per prospect. A geo-grid Map Pack heat map via RankGridEngine runs about $0.045 a business. SEO traffic and keyword data via DataForSEO runs about $0.02 a business.
• Cost to build the redesign mockup: a few cents of Claude copy generation plus a $0 Playwright render.
• Value per closed client: a single SEOBeliever engagement, even at just the $1,800 Audit tier, is worth roughly 45,000 times what it costs to find and score the prospect that became that client. That ratio is the entire argument for running the qualification pipeline widely: the cost of over-qualifying is trivial next to the value of a single conversion.
• The real bottleneck isn't cost, it's selection. Because SEOBeliever sells one client per vertical per market, FWC's job is precision, not volume: find the single best non-competing prospect, not the largest possible list. Our 556-candidate Boulder sweep and our 38-vertical saturation sweep both exist to serve that selection problem, not to maximize lead count.

9. Go-to-Market

FWC's go-to-market is really "prove the pipeline in Boulder, then decide whether to widen it," not a conventional customer-acquisition funnel, since FWC has no external customers of its own yet.

Phase 1: Prove the Model With the Pipeline We Already Have

We already hold verified, non-competing prospect data for TRT and hormone clinics, IV therapy and wellness lounges, and an active nail-salon vertical test, plus real Citation Snapshot audits and redesign mockups already produced for Boulder medspa, dermatology, and dental prospects. The immediate goal is converting a handful of these into signed SEOBeliever or Vitalency engagements, not finding more prospects.

Phase 2: Systematize the Select-the-One Step

Right now, picking the single best non-competing prospect per vertical per market is judgment work done by an agent reviewing the data. Turning that into code (matching review strength, review velocity, Map Pack openness, and Citation Gap severity into a single ranked pick) is what makes the pipeline repeatable without a person re-doing the analysis every time.

Phase 3: Expand Metro by Metro, Only After Boulder Proves Out

The 38-vertical saturation-sweep method (roughly $0.08 to check dozens of business types in a metro) is designed to be run in any new metro SEOBeliever or Vitalency wants to enter, before committing outreach time there.

Phase 4: Decide on Paid Data Sources

Outscraper and Google Places are already interface-complete stubs; wiring in a paid source removes the free-tier ceiling on how many businesses we can qualify per day, but that's a spend decision, not a blocker to closing the deals already in the pipeline.

10. Defensibility and Moat

• The methodology is SEOBeliever's, and it's already proven. FWC isn't running a generic AI-visibility checker anyone could copy from a blog post. It automates SEOBeliever's own CITE prompt battery and Citation Share Index, a method that already converts as a lead magnet before FWC ever touched it.
• Real, deployable code as the mockup. Competitors can eventually add an AI-visibility check. Building a redesign generator where the outreach screenshot and the deployed site are the identical Astro project is a much deeper technical moat; it requires the same Astro-plus-Cloudflare production pipeline Annette already runs for her own sites, not a bolt-on feature.
• Verified, metro-specific market intelligence. The 38-vertical Boulder saturation sweep and the 556-candidate quick-win sweep aren't generic data; they're grounded, DataForSEO-verified findings about exactly which Boulder verticals are still winnable. That kind of grounded local intelligence has to be re-earned in every new metro, which slows down any competitor trying to copy the playbook wholesale.
• Two established landing zones for the value. Because FWC feeds SEOBeliever's and Vitalency's existing sales motion and pricing ladders rather than inventing its own, a closed deal doesn't depend on FWC building trust from zero; it inherits SEOBeliever's proof and Vitalency's positioning.

11. Competitive Landscape

Our wedge is that we're the only tool combining a proven AI-visibility methodology with a mockup that's already real code, and we're not selling a platform or a list; we're producing the exact asset that turns a cold pitch into a do-it-first conversation.

12. Risks and Assumptions

• Conversion risk: we have real, qualified prospects but haven't yet closed a paying engagement sourced through this pipeline. Mitigation: prioritize converting the existing TRT, IV therapy, and nail-salon prospects over finding more, and treat the first close as the proof point before expanding metros.
• AI-citation urgency is still modest. Annette's own standing caveat: AI answer engines sway a real but still small share of local-service buying decisions today. Mitigation: pitch the Citation Gap honestly as "get ahead of competitors while it's cheap," alongside things that move revenue now, like Map Pack position and review velocity, never as a guaranteed booking flood.
• Google AI Overviews remains unmeasured. It has no chat API, so the Snapshot needs a geo-located SERP API (DataForSEO or SerpApi) to close this gap; it's registered as skipped today, never faked. Mitigation: wire it in before quoting AI Overviews specifically in any client report; until then, be explicit that the Snapshot covers five engines, not six.
• Exclusivity limits volume by design. SEOBeliever's one-client-per-vertical-per-market rule means FWC can't just sell every qualified prospect it finds. Mitigation: this is treated as a feature, not a bug; the select-the-one logic is the actual product, and it protects SEOBeliever's positioning as non-competing.
• Manual selection doesn't scale past a few metros. Picking the single best prospect per vertical is still judgment-driven analysis today. Mitigation: Phase 2 of the roadmap turns this into deterministic, testable code before expanding past Boulder.
• Key-person dependency. The whole pipeline, sourcing through selection through outreach drafting, currently runs through Annette and Charlie directly. Mitigation: the internal-first, SaaS-ready module boundaries exist specifically so this can be handed to a future hire or licensed out without a rewrite.

13. Roadmap and Milestones

Phase 1: $0 Qualification Engine (Done, 2026-06-07)

• Free signal checks, ugliness scoring, CLI, 46 (now 157 across the full suite) deterministic tests.

Phase 2: Claude Vision Scorer + Screenshots (In Progress)

• Headless Playwright screenshots plus a Claude vision read of the actual homepage, an aesthetic judgment the free signals can't make on their own.

Phase 3: AEO / AI-Visibility Audit, Wedge 1 (Done)

• The full CITE prompt battery live across five engines, the Citation Share Index scoring, and real Snapshot audits already produced for Boulder medspa, dermatology, dental, and orthodontic prospects.

Phase 4: Real Data Sources + Enrichment (Next, Pending Greenlight)

• Wire Outscraper (primary) and Google Places (fallback) behind the already-locked interface. The one real human dependency: an API account and a spend decision.

Phase 5: Redesign Mockup, Then Deploy, Wedge 2 Centerpiece (Partially Proven)

• A real Astro mockup has already been generated and deployed for at least one Boulder prospect; the next step is making that a repeatable generator rather than a one-off build.

Phase 6: Per-Lead Opener + Branded Report (Next)

• Evidence-grounded, drafted-never-sent outreach openers, plus a branded web-and-PDF report, folding in the reporting strength GoHighLevel already has.

Phase 7: GoHighLevel Parity, Then Past It

• Listings and NAP accuracy, reviews and reputation snapshots, competitor side-by-side comparisons, an embeddable free-audit widget, and predictive lead prioritization, closing every gap GoHighLevel covers that FWC doesn't yet, without giving up the two wedges it can't touch.

Ongoing: Convert and Hand Off

• Every won deal gets logged back into SEOBeliever's or Vitalency's own status tracking. That handoff, not more prospecting volume, is the actual near-term milestone.

14. Financial Projections (Illustrative, Directional)

These figures are assumptions built on FWC's own unit economics and SEOBeliever's existing ladder, not a forecast. They illustrate revenue FWC could realistically help produce for SEOBeliever and Vitalency combined if the pipeline converts as planned; they are not FWC's own revenue, since FWC does not bill directly.

Assumptions: Year 1 reflects converting a fraction of the real prospects already in hand (TRT and hormone clinics, IV therapy, one nail-salon test, Boulder medspa and dermatology candidates). Average engagement value rises as trust builds toward Sprint and Blueprint tiers. Monitoring revenue is ramp-weighted, since clients join the $1,500-a-month tier partway through a year rather than on January 1. Expansion into new metros in Years 2 and 3 assumes the 38-vertical saturation-sweep method is repeated and the select-the-one logic is systematized per Phase 2 of the roadmap.

Immediate Next Steps

1. Convert at least one of the existing verified TRT, hormone, or IV-therapy Boulder prospects into a signed SEOBeliever engagement, since the pipeline data already exists.
2. Wire the Google AI Overviews SERP API so the Citation Snapshot covers all six answer surfaces, not five.
3. Turn the select-the-one prospect logic into deterministic code instead of judgment-based agent review, ahead of any second-metro expansion.
4. Run the 38-vertical saturation sweep in a second candidate metro to test whether the Boulder playbook transfers.
5. Build the reusable redesign generator so Wedge 2 stops being a one-off build per prospect and becomes a repeatable step in the pipeline.