All resources / Guide

What does a 90-day AI search plan look like?

A small team can run AI search work in 90 days with a fixed prompt set, a technical check, a short list of pages that answer real buyer questions, and a few third-party mentions. Because answers vary from run to run, log how often you appear across many runs each week, and compare quarters rather than weeks. The plan promises no mentions; it gives you a repeatable process and honest numbers.

By Corvan · Published · PDF guide, 7 chapters

After this guide you will be able to run a fixed set of 30 buyer prompts across four AI engines and log the results in a sheet, with competitors as your control. You will have checked the settings that decide whether AI tools can read your site, and you will know which five pages and three outside sites come first. You will also have a 13-week calendar, a one-hour routine, and the arithmetic that says which changes are big enough to report. Nothing here promises a mention, because answers change from run to run and this work is slow to show.

Key takeaways

  • Freeze 30 buyer prompts after week one, because a changed prompt set manufactures a trend.
  • OAI-SearchBot decides ChatGPT search appearance and GPTBot is for training, so make those two calls separately.
  • By hand you get 90 answers per engine a quarter, about plus or minus 9 points, so only big moves show.
  • Read single weeks as a log and compare quarters, with three competitors in every run.

What is inside

  1. What should a 90-day plan answer before you start?
  2. How do you run a baseline audit in week one?
  3. Which technical settings decide whether AI tools can read you?
  4. Which pages should you write or fix first?
  5. Which third-party sites should mention you?
  6. How do you measure without fooling yourself?
  7. What does a weekly one-hour routine look like?

The full text is in the PDF. Fill in the form to open it.

Where the facts in this piece come from

  1. Semrush, ChatGPT traffic analysis from 17 months of clickstream data
  2. Search Engine Journal, AI recommendations change with nearly every query (SparkToro)
  3. Cloro, AI visibility sample size
  4. The Hoth, why your AI visibility number is wrong
  5. OpenAI, overview of OpenAI crawlers
  6. Google Search Central, AI features and your website
  7. Peec, most-cited domains in AI search
  8. Sistrix, AI citation drift
  9. HasData, the AI crawler block index

FAQs about What does a 90-day AI search plan look like

1. How many AI answers do I need before a visibility number means anything?

Cloro, which sells AI answer data, puts it at about 300 answers per period for a margin of roughly plus or minus 5 points, and says to read each engine on its own. A small team working by hand collects about 90 per engine a quarter by our arithmetic, about plus or minus 9 points, so it can see only large changes.

2. Will a 90-day AI search plan get my company mentioned in ChatGPT?

We cannot promise that. Sistrix found ChatGPT replaced about 74 percent of its cited domains each week, and AI search work takes one to three months to show movement. The plan gives you a checked site, better pages, some outside mentions, and a number to compare next quarter.

3. Does blocking GPTBot take my site out of ChatGPT search?

OpenAI says the two settings are independent: OAI-SearchBot controls whether you appear in ChatGPT search, and GPTBot is for training its models. So you can allow the first and block the second in robots.txt, the text file that tells crawlers what they may fetch.

About Corvan

Corvan makes Corvan AI, AI agents that run your marketing: they plan your campaigns, write every post in your voice, and show you which posts booked meetings. The team also runs go-to-market services: consultation, ad operations, and AI search ranking. Every piece under Resources is written by the team that runs the work, with sources. Meet the team on the team page, or read what it returned for Drive Me Barcelona.