What an AI visibility program looks like, week by week

What an AI visibility program looks like, week by week

LA
Jedi Solomon
··5 min read

An AI visibility program for a small or mid-size business runs in four phases over 90 days: a baseline in weeks 1 and 2, technical fixes and the first pages in weeks 3 and 4, publishing and third-party mentions in weeks 5 to 8, and a re-test in weeks 9 to 13. The target is a number you can check: your brand named or linked on a set of buyer prompts where you are not cited by ChatGPT, Gemini or Perplexity today.

WeeksWorkWhat you hold at the end
1 to 2Prompt panel, baseline runs, offer scoringA count of prompts where you are cited now, and a gap list
3 to 4Crawler access, schema, first 5 answer-first pagesPages live and readable by crawlers
5 to 8Weekly publishing, outside mentions, listing cleanupA steady cadence of reviewed pages
9 to 13Re-test, fix the missesYour cited count against the target

Why an AI visibility program starts with a baseline

Write 25 buyer prompts: the questions a customer types before choosing someone like you. "Best project management software for a 20-person agency." "Where to buy organic cotton work shirts in bulk." Pick prompts you do not win today, because a panel you already win tells you nothing. Run each one three times in ChatGPT, Gemini and Perplexity, and log who gets named. Use one row per prompt and one column per assistant, so the gaps show up at a glance. The AI visibility audit page shows how we set that panel.

Then score your offers. Can a stranger read your site and learn what you sell, who it is for, what it costs or roughly costs, and what your return, warranty or service terms are? Assistants quote what is on the page. A missing price or a policy buried in a PDF is a hole. Write down every gap. That list drives weeks 3 to 8.

Weeks 3 and 4: access, schema and the first five pages

Check robots.txt and your CDN or firewall rules for the crawlers from OpenAI, Google and Perplexity. Sites block them by accident more often than you would expect. Next, publish schema that matches each page: Organization, Product or Service, and FAQ where the page has real questions. Structured data correlates only weakly with AI citations, so treat it as hygiene that supports rich results and feed consistency. Skip llms.txt. Google discourages the file.

Last, write the first five answer-first pages: the answer in the opening two sentences, then the specifics. Choose the prompts where buyers compare or ask about price. A pricing explainer, a comparison, a "best for" page, a policy page and a service-area page cover most businesses. At LynkAI this is the 30-day line: crawler access fixed, schema live and five pages published, or the $1,500 audit fee is refunded (terms here).

Weeks 5 to 8: publish steadily and get mentioned elsewhere

Publish a few pages a week and have a person review each one before it goes live. Work down the gap list. Each page answers one question, and every claim on it is something you can show.

JCS Fashions, a 4,878-product Shopify apparel store in Milpitas, California, published about 290 quality-gated articles between April and September 2026. Its guide-page clicks went from 0 in February to 231 in August. That is a big catalog, and a small consulting firm needs far fewer pages. The full JCS Fashions story has the monthly numbers.

Mentions on other sites carry weight too. Ahrefs, studying 75,000 brands, found a 0.66 correlation between third-party brand mentions and AI Overview visibility, against 0.22 for backlinks. Spend these weeks on review profiles, industry roundups, partner pages and the directories where your category is listed. If you run a local or multi-location business, complete the listings the engines read: Gemini grounds local answers on Google Maps, while ChatGPT and Perplexity use Bing Places and Yelp.

Weeks 9 to 13: re-test and fix the misses

Re-run the same 25 prompts on a single day, three runs each. Count a prompt as cited when you are named or linked in at least 2 of 3 runs. Our 90-day guarantee is 15 of 25, or we keep working unbilled until it is reached.

For each miss, look at who the engine named and which page it drew from. The cause is usually a missing page, a thin page or no outside mention. Fix those and run the same prompt again to confirm the change. Expect these AI citations to show up as sessions before they show up as orders. On JCS, sessions from AI assistants went from 20 in January to 569 in October, and AI-assistant orders were 5 in October. Total Shopify sales over the same months went from $1,706 to $26,371, across all channels with paid ads included. October 2025 was $16,558, so season is part of that rise.

Start with the baseline this week

Weeks 1 and 2 can be done by you in an afternoon: write the 25 prompts, run them three times each, and record who is named. If you want the result without the spreadsheet, the free AI visibility scan shows where your business stands in ChatGPT, Gemini and Perplexity, and it is a fair starting point for any AI visibility program you run yourself or hand off.