How It Works

Measure. Fix the datasets.
Prove it every month.

AI does not crawl your website to recommend you. It pulls from a short list of sources. We get you into those sources, correctly and consistently, then re-measure across 11 AI models and show you the change.

1
Step 1 ยท Week 1

Measure

We ask 11 AI models the questions your customers actually ask: "best [your service] near [your city]", "who should I call for...", "is [your business] any good". 15 questions, 11 models, 165 data points. Every answer is scored.

11 AI models

Google AI Overviews and AI Mode, ChatGPT, Gemini, Perplexity, Claude, plus Grok, DeepSeek and four more for breadth

Category Clarity

A blind test: does AI even know what you are? Businesses that fail this get recommended for the wrong thing, or not at all

Hallucination check

We catch wrong hours, wrong phone, wrong address, wrong services, or AI saying you do not exist

Who AI recommends instead

Your competitors scored on the same questions, so you see the gap on day one

๐Ÿ“Š

Deliverable: your Day-0 baseline score, model by model, with the exact sources AI is pulling from


2
Step 2 ยท Days 7 to 30, then ongoing

Fix the datasets

Google AI answers from your Google Business Profile. ChatGPT answers local questions from Yelp and Bing Places. Siri reads Apple Business Connect. Best-of lists and your vertical directory feed the rest. We make every one of them complete, correct and identical.

AI Data Feed pass

Google Business Profile, Yelp, Bing Places, Apple Business Connect, BBB and your vertical directory: name, categories, services, hours, pricing and attributes made consistent

Best-of lists and citations

Outreach to the local and vertical best-of lists AI cites most, plus 20 citations a month on Pro

Answer-first pages

3 service pages rebuilt so the answer comes first, with FAQs. Schema and llms.txt included as hygiene

Reviews

Review engine sends requests, drafts responses and auto-posts. Reviews are one of the top signals AI uses to pick who to name

โœ…

Deliverable: written 30-day checklist. Baseline by Day 7, strategy doc by Day 10, data feed audit by Day 14, first pages live by Day 21


3
Step 3 ยท Every month

Re-measure and prove it

75 to 88% of the sources AI cites change day to day, so a one-time fix does not hold. Every month we run the full audit again (twice a month on Pro), compare it to your Day-0 baseline and send you a report you can read in five minutes.

Score movement

Your score versus baseline, model by model. What moved, what did not, what we are doing about it

First-mention alerts

The first time a model names you for a question it did not before, you hear about it

Weekly hallucination watchdog

Wrong answers about your business get caught and corrected at the source before your customers see them

Owner report

Plain language. No dashboard to learn. One page for the owner, detail behind it if you want it

๐Ÿ›ก๏ธ

The promise: 20 points above your Day-0 baseline by Day 120, or billing pauses in month 5 while we keep working free

Why we fix sources instead of chasing rankings

Being recommended by AI is rare, and the reason is structural. The models name businesses from the datasets they are grounded on, not from whoever has the best website.

1.2%
of local businesses named by ChatGPT. Perplexity 7.4%, Gemini 11%, versus 35.9% shown in a Google 3-pack
3.4x
more local AI citations go to Yelp than any other source. ChatGPT now answers local questions from Yelp
88%
of people verify an AI recommendation on Google before acting, so your AI answer and your Google profile must agree

That is why Step 2 is directories, lists, pages and reviews, in that order, and why Step 3 never stops.

Sources: SOCi Local Visibility Index 2026, BrightLocal 2026, local AI citation analysis 2026.

What You Get Every Month
๐Ÿ“Š

11-model audit

15 real customer questions across 11 AI models. Monthly on Core, twice a month on Pro. Scored against your Day-0 baseline.

๐ŸŽฏ

Category Clarity score

Does AI know what you are? Tracked every audit so you can watch it move from "some kind of contractor" to your actual category.

๐Ÿ›ก๏ธ

Hallucination watchdog

Weekly checks for wrong hours, phone, address or services in AI answers. Corrections pushed to the source.

๐Ÿ””

First-mention alerts

The first time a model names you for a question it did not before, you get the alert with the exact answer.

๐Ÿ“„

Owner-readable report

One page in plain language: score, what moved, who AI recommends instead, what we are doing next month.

โญ

Review engine

Requests, responses and auto-posting run for you every month. Included in both products.

See Step 1 for your business, free

The free AI audit is the same measurement we run for customers. 11 models, 165 data points, about 2 minutes. No card.