How AI engines pick who to recommend — and how we get you picked
No magic, no “secret AI hack.” AI recommendations are the output of a supply chain of sources. We audit that supply chain, fix the broken links, and re-measure every month until your name is in the answers.
Where an AI answer actually comes from
When someone asks ChatGPT “who's the best plumber near me?”, the answer is assembled from a handful of upstream sources. If you're weak in any of them, you drop out of the answer. This is the whole game:
1 · The model's memory
What the AI learned about your market during training — from directories, review sites, news, and 'best of' lists. You can't edit the past, but the sources it trusts keep getting re-crawled. Get on them now and future model updates learn you exist.
2 · Live web search (Bing for ChatGPT)
For 'near me' and 'best X in Y' questions, engines search the live web. ChatGPT's search runs on Bing — the index almost every local business ignores. An unclaimed Bing listing means invisibility on the fastest-growing referral channel.
3 · Your website, as a machine reads it
AI crawlers (GPTBot, ClaudeBot, PerplexityBot) read your site — if your robots.txt lets them, and if your pages contain machine-verifiable facts: services, service area, hours, pricing signals, proof. Most local sites fail both.
4 · Your Google Business Profile
Gemini and Google's AI Overviews lean heavily on GBP data: categories, services, attributes, Q&A, photos, posts. A thin profile reads as an unverifiable business — so the AI names someone else.
5 · The citation layer
Directories and local 'best of' articles the engines quote. Every market has a specific set — and it's different for plumbers in Chicago vs. dentists in Austin. Your audit reveals which sources the engines actually used in your market.
6 · Review signals
Not just star count. Engines weight recency, velocity, and whether reviews mention specific services ('fixed our water heater same-day'). A 4.9★ business with stale reviews loses to a 4.6★ with fresh, specific ones.
Step 1 — We measure what AI actually says
Everything starts with real answers, not guesses:
36+ real buyer questions
We submit the questions your customers actually ask — 'best X in Y', 'who should I call right now', 'affordable but trustworthy' — to ChatGPT, Gemini, and Claude through their official APIs, and record every answer verbatim.
Competitor + citation extraction
Every business named in every answer is counted, so you see exactly who's winning your customers and by how much. We also capture which sources the engines cite — that list becomes your fix plan.
Technical checks
AI crawler access on your robots.txt, structured data on your site, Bing presence, and Google local pack placement. These explain why you're invisible — not just that you are.
Honesty note: AI answers vary between runs. That's exactly why we measure with dozens of prompts across three engines and report mention rates, not single answers — and why anyone promising you “#1 in ChatGPT” is selling snake oil. We promise honest measurement and diligent execution, and we show our work.
Step 2 — We fix the supply chain, in priority order
The exact source list and sequencing come from your audit — that prioritization is the craft. Here's what the work includes:
Technical foundation (days 1–14)
Unblock AI crawlers in robots.txt. Claim and complete your Bing Places listing — ChatGPT's search layer. Install LocalBusiness schema with services, service area, hours, geo, and review markup so engines can verify you as a real, current business.
Google Business Profile depth (days 1–30)
Complete every category, service, and attribute. Seed the Q&A section with the questions buyers ask. Weekly post drafts written for you. Photo guidance. GBP is Gemini's primary local source — depth here moves answers.
Citation placement (days 7–60)
Your audit shows which directories and 'best of' pages the engines cited in your market. We run a placement campaign against that specific list — correcting data, securing listings, pitching inclusion where editorial. We never fake reviews or astroturf. Ever.
Answer-shaped content (monthly)
AI engines lift answers from pages that directly answer buyer questions with specifics and proof. We write them for your site every month — service + city pages structured the way engines quote, with your real differentiators.
Review velocity system (ongoing)
A simple ask-flow for your recent customers plus response drafts for every review. Steady, fresh, specific reviews are the strongest trust signal engines have — and the hardest one for competitors to fake.
Your part: about 15 minutes at onboarding, then approving what we send. If you can forward an email, you can run this.
Step 3 — We prove it moved (or you cancel)
Baseline audit + technical foundation. You get your first full report and the fix plan. Crawler, Bing, schema, and GBP work lands.
Citations land, content publishes, review flow starts compounding. Retrieval-based answers (the 'search the web' kind) typically move first.
Mention rates climb in your monthly reports. Deeper presence compounds — every month of citations, content, and reviews makes you harder to skip.
Your monthly report shows:
- · Your score vs. last month, per engine
- · Competitor mention movement
- · Verbatim answers — the receipts
- · What we did, what's next, what we need from you
Same prompts, same engines, same methodology every month. If it's working, you'll see it. If it's not, you'll see that too — and you can cancel in one click.
Start with the measurement — it's free
See your score, your competitors, and your specific broken links before you spend a dollar.
Run my free audit