AI SEO Strategy: A 90-Day Roadmap
An AI SEO strategy is a sequence, not a list. Every guide agrees on the ingredients: technical access, entity clarity, quotable content, off-site mentions, and measurement. What most skip is the order, and order is what decides whether 90 days of work produces results or a pile of half-finished initiatives. This roadmap puts the fast, foundational work first and starts the slow, compounding work early enough to matter.
What 90 days can and cannot do
| Realistic in 90 days | Takes longer |
|---|---|
| A measured baseline of how AI engines describe you | A Knowledge Panel |
| Technical access and indexation fixed | Consistent presence in competitive category answers |
| A clean, connected entity foundation | Model memory catching up with new facts |
| One complete topic cluster, indexed | Rankings for the most competitive head terms |
| First off-site mentions and reviews | A mature review and press footprint |
Set expectations against the left column. If someone promises the right column in a quarter, ask how.
Scroll the diagram sideways to see all of it.
Days 1 to 30: foundation
The first month fixes what blocks everything else and records where you started. Resist the urge to publish content in this phase; content published onto a site engines cannot read or identify is wasted.
Week 1: capture the baseline
Build a prompt set of 30 to 50 questions real buyers ask, run each several times across ChatGPT, Perplexity, Gemini, and Google AI Mode, and log presence, citations, accuracy, and which competitors are named. Export 16 months of Search Console data, verify Bing Webmaster Tools, and crawl the site. This baseline is the only way to show, on day 90, what changed. Methods in measuring brand visibility in ChatGPT and share of model.
Week 2: fix access, rendering, and indexation
Work through the first three sections of the technical SEO checklist for AI search: crawler access including CDN bot settings, indexation in both Google and Bing, and server-side rendering. These are the fixes most likely to move AI answers quickly, because they decide whether you can be retrieved at all.
Weeks 3 and 4: build the entity foundation
State your core facts plainly on one canonical page, then connect them in structured data: one Organization node with a stable @id, Person nodes for founders, and a short sameAs list. Audit every profile and directory for inconsistencies. If your name is shared with another entity, start entity disambiguation here, before you add more signal. The entity SEO guide is the hub for this phase.
Days 31 to 60: build one cluster
The second month builds one complete topic cluster, not five partial ones. A finished cluster signals depth; scattered posts signal nothing in particular.
Pick the cluster closest to revenue
Choose the topic that sits directly above your most important service page. Topical authority transfers best to the commercial page the cluster links to.
Map the question space
List the primary questions and their follow-ups, from AI engines, “People also ask”, and your sales and support teams. Most will have little or no search volume, which is expected; they are the queries engines generate during fan-out. See topic clusters that AI engines cite.
Brief, publish, and mark up
Write a brief for every page using the AI SEO content brief template. Publish a pillar and six to ten supporting pages, including at least one comparison page and one honest decision page covering cost and fit. Add Service schema to the service page the cluster supports.
Wire the links
Every supporting page links up to the pillar and across to its closest siblings; the pillar links down to all of them; everything routes to one service page. Descriptive anchor text throughout.
Days 61 to 90: earn corroboration and measure
The third month starts the slow work in earnest and closes the loop on measurement.
Earn independent mentions
Complete and refresh your review profiles, identify the category roundups that rank in both Google and Bing, and start outreach for inclusion. Fix directory inconsistencies found in month one. The playbook is in digital PR for AI search. None of this pays off inside 90 days, which is exactly why it starts now.
Run a first refresh pass
Revisit the pages that existed before the plan. Update facts, add sourced figures, restructure passages to lead with the answer, and set honest updated dates.
Re-run the baseline and report
Run the same prompt set the same way and compare against week one. Report leading and lagging indicators separately, because in the first quarter the leading ones move first.
| Leading indicators (expect movement) | Lagging indicators (expect little yet) |
|---|---|
| Pages indexed in Google and Bing | Citations in competitive answers |
| Impressions for the new cluster | AI referral traffic |
| Presence rate across the prompt set | Share of model against leaders |
| Accuracy of how engines describe you | Pipeline attributed to AI |
The weekly plan at a glance
| Week | Focus | Output |
|---|---|---|
| 1 | Baseline | Prompt set results, Search Console export, crawl |
| 2 | Technical | Access, indexation, and rendering fixes shipped |
| 3 to 4 | Entity | Canonical facts page, connected schema, profile audit |
| 5 | Cluster planning | Question map and page list |
| 6 to 8 | Cluster build | Briefs written, pillar and supporting pages published |
| 9 | Links and schema | Internal links wired, Service schema live |
| 10 to 11 | Off-site | Reviews, directories, roundup outreach |
| 12 | Refresh | Existing pages updated |
| 13 | Measure | Baseline re-run, 90-day report, next cluster chosen |
How the roadmap lines up with Google’s own guidance
One of the top results for “AI SEO strategy” is not an agency post at all. It is Google Search Central’s May 2025 article on succeeding in Google’s AI search experiences. It is worth reading in full, and it is reassuring: nothing in it asks for tactics beyond sound SEO. Each of its recommendations maps onto a week of this plan.
| Google’s recommendation | Where it sits in this roadmap |
|---|---|
| Focus on unique, valuable content for people | Weeks 5 to 8: the cluster is built from real buyer questions and first-hand facts |
| Provide a great page experience | Week 2: technical fixes, including speed and mobile |
| Ensure Google can access your content | Week 2: robots, rendering and indexation |
| Manage visibility with preview controls | Week 2: decide deliberately on nosnippet and similar controls rather than inheriting defaults |
| Make sure structured data matches visible content | Weeks 3, 4 and 9: schema generated from the same facts shown on the page |
| Go beyond text | Weeks 6 to 8: original diagrams, screenshots and images in the cluster pages |
| Understand the full value of your visits | Weeks 1 and 13: measure conversions and assisted value, not only clicks |
| Evolve with your users | Weeks 12 and 13: refresh what buyers now ask, and choose the next cluster from it |
What Google’s article does not cover is everything outside Google: Bing indexation, other assistants, and the independent mentions that make a brand easy for any engine to corroborate. Weeks 2, 10 and 11 cover those, because a strategy that only serves one engine is a channel plan, not an AI SEO strategy.
Who does the work
One strategist can run the plan part-time with a writer and a few hours of developer time in weeks two and nine. What usually stalls in-house plans is not skill but continuity: the measurement cadence and the off-site work are the first things dropped when other priorities arrive. In-house GEO vs hiring an agency covers how to decide, and our pricing shows what an agency-run version of this roadmap costs.
Where the 90-day AI SEO roadmap leads next
Day 90 is a checkpoint, not a finish line. The foundation you built is reused by every future cluster, so the second quarter moves faster: a new cluster, a quarterly refresh of the first, and the off-site work continuing to compound. If you want this roadmap run for you, from baseline to 90-day report, that is how our AI SEO service engagements begin.
Frequently asked questions
What should an AI SEO strategy include?
A baseline of how AI engines describe you today, technical access and indexation fixes, an entity foundation of consistent facts and connected structured data, topic clusters that cover the questions buyers ask, off-site corroboration, and a measurement cadence that tracks presence and citations as well as traffic.
How long does AI SEO take to work?
Technical and indexation fixes can show up in AI answers within days to weeks. Content clusters usually need one to three months to be indexed and start earning citations. Entity recognition and off-site corroboration compound over several months, so a 90-day plan should expect leading indicators, not a finished result.
What should I measure in the first 90 days?
Leading indicators: indexation in Google and Bing, impressions for the new cluster, presence rate across your prompt set, and accuracy of how engines describe you. Lagging indicators such as citations, AI referral traffic, and pipeline follow later.
Can a small team run a 90-day AI SEO plan?
Yes. One strategist working part-time with a writer and occasional developer help can run the plan below. Larger sites take longer per phase, but the order of the work stays the same.
Is AI making SEO obsolete?
No. AI answers are built on retrieved web pages, and retrieval depends on the same crawling, indexing and quality signals SEO has always worked on. What changes is the finish line: being named and cited in an answer counts alongside ranking, which is why this roadmap measures presence in AI answers as well as traditional rankings.
What comes after the first 90 days?
The second cluster, a quarterly refresh of the first, continued off-site work, and the same measurement cadence. The first 90 days build the foundation that every later cluster reuses.