AEO in 90 Days: A Week-by-Week Plan for a B2B Company Starting From Zero

A B2B company starting from zero can build a working AEO program in 13 weeks. The sequence is: measure first, remove access blocks, make your company unambiguous, rewrite and build answer-first pages, start earning third-party proof, then re-measure with the identical method. This plan gives you the week-by-week schedule, owners, deliverables and a "done when" test for each week.

It is written for a company without a dedicated AEO team: a marketing lead who owns it, a developer for a few hours a week, and subject matter experts for interviews. Enterprises with many business units should also read the enterprise AEO strategy playbook. If the terms are new, start with AEO and GEO for beginners.

The plan at a glance

Weeks 1 to 3: Measure, then remove blocks

Week 1: Kickoff and access

Most AEO programs lose their first month waiting for access. Get it in week one.

• Google Search Console and GA4 (editor access)
• CMS with publishing rights
• CDN or WAF console, and raw server or CDN logs
• CRM fields and web forms
• LinkedIn company page, Clutch or G2 profile, partner directory listings

In the same week, agree on three priority offerings and one or two priority industries, and write down five competitors from CRM loss reasons. The full access list is in the week-one access checklist.

Week 2: Prompt set and baseline

Write 40 prompts from sales calls, RFPs and discovery notes, spread across category, problem, comparison, alternatives and validation prompts. Run each one three times in ChatGPT, Claude, Perplexity and Gemini under a written protocol. Score mention rate, citation rate and share of voice, and save the cited URLs.

Do not skip repeat runs. Answers vary: Orbit Media found Claude kept only 30% of its cited URLs week to week. The full method and scoring template are in how to benchmark your AI visibility. Nothing on the site changes until this baseline is saved.

Week 3: Crawler and index fixes

Run layer 1 of the AEO checklist. The usual findings in week 3:

• A CDN bot setting that challenges AI user agents
• Key copy rendered only by JavaScript
• Money pages missing from the sitemap, or carrying a stray noindex
• robots.txt rules copied from an old template that block search-facing bots

This week has outsized impact. OpenAI states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and Google requires pages to be indexed and snippet-eligible to appear as supporting links in AI Overviews and AI Mode. Done when logs show verified hits from OAI-SearchBot, Claude-SearchBot and PerplexityBot. See AI crawler optimization.

Weeks 4 and 5: Make the company unambiguous and map the pages

Week 4: Entity foundations

1. Write one canonical company description: category noun, audience, offerings, founding year, headquarters, one dated proof point.
2. Paste it into the homepage, About page, LinkedIn, your Clutch or G2 profile, Crunchbase and partner listings.
3. Publish bio pages for the founder and two or three practice leads, each linked to LinkedIn.
4. Add Organization markup with a complete sameAs list on the home or About page.
5. Ask each engine "What is [brand]?" and log any confusion with a similarly named company.

Background: entity SEO and Organization schema.

Week 5: The prompt-to-page map

Take every weight-3 and weight-2 prompt and assign it one target URL. Where no page fits, that is a gap.

PROMPT-TO-PAGE MAP

prompt_id | prompt | family | weight | target URL | status | fix | week
P01 | best CLM for mid-market manufacturers | category | 3 | /industries/manufacturing/ | exists, weak intro | rewrite | 6
P07 | CLM that works inside ServiceNow | category | 3 | (none) | gap | new integration page | 9
P12 | [Leader] vs [Us] | comparison | 3 | (none) | gap | new comparison page | 7
P19 | how is [Us] priced | validation | 2 | /pricing/ | no pricing model stated | rewrite | 6
P23 | alternatives to [Leader] | alternatives | 3 | (none) | gap | new alternatives page | 7

Rank gaps by weight, then by how often the prompt comes up in sales calls. Cap it at eight pages for the quarter: four rewrites and four new pages. Comparison pages deserve priority if Google matters to you: Seer Interactive found comparison queries triggered AI Overviews 95.4% of the time. The architecture is covered in question-based content architecture.

Weeks 6 to 10: Pages and proof

Week 6: Rewrite four money pages

Rewrite before you create. Existing pages already have crawl history and internal links. For each page, interview the subject matter expert for 30 minutes with these questions:

SME INTERVIEW (30 minutes per page)
1. In two sentences, what is this, and who is it for?
2. What do buyers ask on the first call about this?
3. What numbers can we publish? (timelines, volumes, counts, dates)
4. What does it integrate with or depend on, by name?
5. What is the pricing model and what drives cost up or down?
6. Which customer result can we name, and in what timeframe?
7. When is this the wrong choice for a buyer?
8. What do competitors claim that is misleading?

Then rebuild the page: a 40 to 60 word answer block at the top, question-style H2s, a comparison or spec table, the named proof, and a visible last-updated date. Question 7 matters more than it looks. Pages that say when you are the wrong fit read as evidence, not marketing.

Week 7: First new pages

Ship the comparison or alternatives page, one priority industry page and one case study with a measured outcome. Keep comparison pages factual and fair, with a table and a "choose them if" section. Guides: comparison pages AI engines cite and case study pages for AI citations.

Week 8: Third-party profiles

Go back to the cited URLs in your week 2 baseline. Count how often each third-party domain appears. Those are your gatekeepers.

• Complete your Clutch profile (services) or G2 profile (software) with the canonical description, correct categories and portfolio items.
• Send ten customers a personal review request through the platform's own flow. Account managers ask; marketing supplies the link and a short note.
• Update partner directory and marketplace listings: tier, certifications, industries, description.

Review sites carry weight in some engines. Orbit Media found Clutch was Claude's single most-cited third-party domain. More in the off-site assets that get B2B brands cited.

Week 9: Second new pages and the founder series

Ship the pricing model page and the integration page from your gap map. Start a founder or CEO LinkedIn series on the two topics you most want to be cited for, three posts over the next three weeks. Perplexity's most-cited domain in the Orbit Media study was LinkedIn. The system for making executive content a pipeline channel is in the CXO branding playbook.

Week 10: Listicles and transcripts

For every listicle in your baseline citations where you are missing or described wrongly, send a short, factual request to the author: what you do, for whom, one proof point, and why readers of that list would care. No payment offers, no pressure. Then turn your best webinar or podcast appearance into an ungated transcript page with a summary at the top. Guides: listicle placements and webinar content for AI citations.

Weeks 11 to 13: Measure and decide

Week 11: Measurement plumbing

• Build a GA4 custom channel group for AI referrers. Steps are in AI referral traffic in GA4.
• Add "AI assistant (ChatGPT, Claude, Perplexity, Gemini)" to the "how did you hear about us" field and map it to the CRM.
• Run a mini re-check of the 10 highest-weight prompts to spot anything broken before the full run.

Week 12: Full re-benchmark

Run the same 40 prompts, three runs, four engines, under the same protocol as week 2. Same accounts, same location, same week of the month. Calculate margins of error before reading the deltas. With 480 answers and a 20% mention rate, the 95% margin is roughly plus or minus 3.6 points overall and wider per engine.

Week 13: Readout and next-quarter plan

90-DAY AEO READOUT (one page)

1. Baseline vs now: mention rate, citation rate, share of voice (by engine)
   Moves outside the margin of error: [list]
2. By prompt family: biggest gain, biggest remaining gap
3. What shipped: [8 pages, profiles, requests sent, inclusions won]
4. Leading indicators: verified bot hits, indexed pages, new reviews,
   listicle inclusions, AI referral sessions, self-reported AI leads
5. What we will stop doing
6. Next quarter: next offering or industry, 6 pages, off-site targets,
   prompt set v2 changes
7. Pipeline link: opportunities with AI-assisted discovery noted

Effort budget

Illustrative planning figures, not benchmarks. Adjust to your team.

Weekly status template

AEO WEEK [n] STATUS
Planned: [deliverables from the plan]
Shipped: [URLs, fixes, requests sent]
Blocked: [item, who can unblock, by when]
Signals: [bot hits, new reviews, inclusions, AI referral sessions]
Next week: [top three]

What realistic progress looks like

By day 90 the reliable wins are the ones you control: access fixed, pages live, profiles complete, requests out, measurement working. Visibility moves on engines that retrieve live pages can follow once your changes are crawled, which is why the week 3 fixes matter. Third-party proof compounds slowly and keeps paying after the quarter ends. Timelines by engine are discussed in how long AEO takes.

The stakes are rising in Google too. Semrush's study of more than 600,000 keywords found AI Overviews on commercial-intent queries grew 71% between November 2025 and April 2026, so B2B buying queries are increasingly answered before the click.

For a proof point on what sustained work can do, our St. Martinus University program grew visibility by roughly 3,600%.

A worked example

Illustrative example. A 30-person procurement software company with no marketing funnel starts the plan.

• Week 2: baseline shows a 4% mention rate and no citations on category prompts. Validation prompts describe the product as an invoicing tool, which is wrong.
• Week 3: the site is a JavaScript single-page app; the developer enables server-side rendering for the ten money pages.
• Week 4: the LinkedIn and G2 descriptions still reflect a pivot from two years ago. Both are rewritten to match the canonical statement.
• Weeks 6 to 9: four rewrites and four new pages ship, including "procurement software for SAP Ariba users" and an alternatives page.
• Week 8: eight of ten customers agree to a G2 review.
• Week 12: the team compares against the baseline with margins of error and focuses the next quarter on the prompt families that moved.

Because the company had no funnel, the team also added a meeting page and a same-day follow-up rule before week 6. Visibility without a conversion path produces visits, not pipeline. Our KNNX engagement started with no funnel at all and reached $12M in pipeline over 28 months once the full system was in place. Full funnel B2B marketing explains how the stages connect.

Mistakes that stall a 90-day plan

• Changing pages before the baseline. You lose the ability to prove anything.
• Waiting on access. Chase logins in week one, not week four.
• Writing without experts. Generic pages do not get cited. Thirty minutes of SME time per page is the cheapest quality lever you have.
• Starting off-site work late. Reviews and listicle inclusions take longest, so start by week 8.
• Chasing llms.txt and schema first. Google says neither AI text files nor special schema are required for its AI features. Do them after access, entity and pages.
• Reporting vanity numbers. Report mention rate, citation rate and share of voice by engine, with margins of error, plus pipeline signals.

The AEO and GEO playbook includes templates for every week of this plan. If you want the plan run with you, our AEO, GEO and SEO team starts with a free audit that covers weeks 1 to 3.