What does an AI visibility program do in 90 days? It measures how ChatGPT, Claude, Gemini and Perplexity answer a fixed list of buyer questions, finds the sources they cite, fixes the brand's own pages, earns coverage on the cited sources, and measures again. The output is a documented record of how often the brand is mentioned, cited and described correctly in the chatbox, before and after.
Last updated November 2, 2026.
What does this AI visibility program guide cover?
- The 90-day plan at a glance
- Weeks 1 and 2: the baseline
- Weeks 3 and 4: the cited-source map
- Weeks 3 and 4: page changes
- Weeks 5 to 10: earned coverage
- Weeks 11 to 13: re-measure and report
- What a full-service program includes
- What the program costs and what drives the price
- Frequently asked questions
What does a 90-day AI visibility program look like at a glance?
A 90-day AI visibility program runs in four phases: baseline, source map and page fixes, earned coverage, and re-measurement. Each phase produces a named output that the next phase depends on, so the order matters more than the speed.
| Phase | Weeks | Work | Output |
|---|---|---|---|
| 1. Baseline | 1 to 2 | Write the prompt list, run it in four engines, log answers and sources, write the fact sheet | Baseline report |
| 2. Source map and page fixes | 3 to 4 | Classify cited domains, confirm crawler access, rewrite the priority pages | Source map and updated pages |
| 3. Earned coverage | 5 to 10 | Place facts, data and expert commentary on the cited outlets, and correct third-party profiles | Placements and corrected profiles |
| 4. Re-measure | 11 to 13 | Run the same prompt list again, compare, decide the next quarter | Before and after report |
No program can guarantee that an engine will name a brand. The plan makes the attempt systematic and the result measurable.
What happens in weeks 1 and 2 of an AI visibility program?
Weeks 1 and 2 produce the baseline: a fixed list of 25 to 50 buyer questions, each run several times in four engines, with every answer logged. The log records whether the brand is named, how it is described, and which domains the engine cites.
Repeat runs are necessary because generated answers change. Pew Research Center noted in its analysis of Google AI summaries that the summary for any given search may change over time. The same weeks produce a one-page fact sheet that lists the company name, category, headquarters, founding date, leadership, products and pricing. The GEO audit checklist lists 20 checks a baseline should cover, and the guide to getting recommended by ChatGPT covers the ChatGPT-specific steps.
Accuracy belongs in the baseline, not only visibility. 5W's Hallucination Index tested five fashion brands on 25 core business questions each and found that 20 percent of answers contained material errors. An AI citation audit automates this baseline across engines.
Why does the program map cited sources before writing any pages?
The program maps cited sources first because engines draw each category's answers from a small and different set of domains, and the brand's own site is rarely the only one that matters. The map shows where the facts about the brand must be accurate.
Pew found that Wikipedia, YouTube and Reddit were the most frequently cited sources in Google AI summaries, making up 15 percent of all sources listed. Category patterns are narrower still. 5W's research found that automotive citations came from KBB, Edmunds, NHTSA and IIHS data rather than press coverage, as the automotive citation analysis shows, and its cybersecurity work found CISA, NIST and CVE.org outranking trade press, as described on the cybersecurity AEO page.
Classify each cited domain into one of five types: the brand's own pages, review or comparison sites, reference sites, government or standards bodies, and news or trade press. Each type needs a different action in phase 3.
What changes on the brand's own pages in weeks 3 and 4?
Three things change on the brand's own pages: crawler access is confirmed, each priority page opens with a short direct answer, and claims gain named evidence. These changes are cheap and they come before any outreach.
OpenAI states that sites opted out of its OAI-SearchBot crawler will not be shown in ChatGPT search answers, so the robots.txt file is checked first. Each priority page then opens with a 40 to 80 word answer to its target question. The median AI summary in Pew's data was 67 words, so that length fits the format. Evidence follows the answer: Aggarwal and co-authors reported in the KDD 2024 paper GEO: Generative Engine Optimization that adding citations, quotations from relevant sources and statistics raised visibility by over 40 percent across various queries, with results that varied by domain.
The method behind these page rules is explained in the guide to what AEO is.
What does earned coverage add in weeks 5 to 10?
Earned coverage puts the brand's facts on independent sites that engines already retrieve. A retrieval system gathers passages from several domains before it writes an answer, so a fact that appears accurately on several independent domains is present in more of the passages the system can use. This is a reasoned mechanism, not a measured guarantee.
Phase 3 work usually has three strands.
- Placements on the news, trade and review outlets that appeared in the source map.
- Original data or research that other sites can cite, because named statistics are the kind of evidence the KDD 2024 paper found raised visibility.
- Corrections to out-of-date company profiles, directory listings and reference entries.
Any contribution to a platform such as Reddit or Wikipedia follows that platform's rules, with accurate and disclosed information.
How is progress reported at day 90?
Progress is reported as a before and after comparison on the same prompt list, because that is the only fair comparison when answers vary. Clicks alone understate the effect, since Pew found users clicked a link inside an AI summary in only 1 percent of visits.
| Metric | Baseline (week 2) | Day 90 | Source of the number |
|---|---|---|---|
| Mention rate | Recorded | Recorded | Prompt log |
| Citation share | Recorded | Recorded | Prompt log |
| Answer accuracy | Recorded | Recorded | Prompt log checked against the fact sheet |
| AI referral sessions | Recorded | Recorded | Analytics referral report |
| Impressions and position | Recorded | Recorded | Google Search Console |
The report then states what moved, what did not, and which sources or pages explain the difference. The next quarter's plan starts from that explanation. GEO optimization services run this cycle on a continuing basis.
What does a full-service AI visibility program include?
A full-service program includes six deliverables that run from a baseline audit to monthly reporting. The table maps each deliverable to what the client receives.
| Deliverable | What it covers | What the client receives |
|---|---|---|
| Audit | How AI assistants describe the brand today | A baseline report with answers, citations and errors |
| Source mapping | Which sites the assistants cite for the category | A ranked list of target outlets and pages |
| On-site fixes | Answer-first pages, structure, dates and facts | Rewritten or restructured pages |
| Earned media and links | Coverage and mentions on the cited sources | Placements and a link record |
| Tracking | A fixed prompt list run on a schedule | Citation and mention data over time |
| Reporting | Monthly summary and next steps | A written report and a review call |
Three things are normally outside a program's scope. Paid placement inside an AI answer is not a service any provider can sell. Content production beyond the agreed page count is usually billed separately. Third-party costs such as wire distribution, research data and tool licenses are often billed on top.
What does an AI visibility program cost, and what drives the price?
5W's AI visibility programs start at $20,000 per month, its GEO retainers run from $15,000 to $75,000 per month, and its PR retainers start around $15,000 per month. Scope sets the final figure, not a standard rate.
| Cost driver | What changes |
|---|---|
| Number of prompts tracked | More prompts mean more data to collect and review |
| Markets and languages | Each market has its own sources and answers |
| Competitors benchmarked | A wider set adds analysis time |
| Pages rewritten each month | More pages mean more editing and fact checks |
| Earned media volume | More placements mean more outreach |
B2B technology companies can read the sector version in GEO services for B2B tech. A buyer comparing providers can use 5W's guide to the 12 questions to ask an AI visibility agency and its comparison of AI visibility tools and agencies to test any proposal against the same plan.
What does an AI visibility program not promise?
An AI visibility program does not promise a citation, a ranking in any engine, or a timeline. Engines choose their own sources and change them over time, so a credible program commits to a method, a measurement plan and a before and after report, and it treats any guaranteed placement in an AI answer as a warning sign.
Frequently asked questions
What does "in the chatbox" mean?
The chatbox is the reply area of an AI assistant such as ChatGPT, Claude, Gemini or Perplexity. Being in the chatbox means the assistant names or cites a brand in its answer to a buyer's question.
How long before an AI visibility program shows results?
No one can promise a timeline. The 90-day plan is built so that the first measurable comparison is available at the end of week 13, and later quarters build on what that comparison shows.
Does the program need new content or only fixes?
Most programs start with fixes to existing pages, since those are the fastest changes and the source map shows which pages matter. New content is added where the prompt log shows a question that no page answers.
Which engines does the program cover?
The baseline in this plan covers ChatGPT, Claude, Gemini and Perplexity, and Google AI Overviews can be tracked through search data. The list should match the engines the company's buyers use.
Is an AI visibility program the same as SEO?
The two overlap but are not the same. SEO targets rankings in search results, while an AI visibility program targets citations and accurate mentions in AI answers, and it relies on the same crawlable, well-structured pages that SEO builds.
Can a program guarantee citations in AI answers?
No program can guarantee citations. Engines select their own sources, so the commitment is to a documented method and a before and after measurement, not to a specific placement.
