In the first 90 days with an AEO agency you should expect a measured baseline, a repaired technical foundation, roughly 20 to 40 published or restructured pages, and your first unlinked mentions in AI answers somewhere between day 30 and day 60. You should not expect revenue. Linked citations trail mentions by two to four weeks, meaningful citation share movement takes about 90 days, and AI referral traffic does not become a measurable channel in analytics until around month six. Any agency promising pipeline inside 90 days is either redefining pipeline or selling you something that will not hold. At retainers that typically run $2,000 to $10,000 a month for mid-market programs and $15,000 and up for enterprise, that is $6,000 to $30,000 spent before you have a number you can take to a board.
What actually happens in days 1 through 30?
Measurement and repair. Nothing published in month one is meant to move a metric; it is meant to make the metrics readable.
A competent engagement opens by recording a baseline visibility score across ChatGPT, Perplexity, Google Gemini, Microsoft Copilot, and Claude. That means running a fixed prompt set, usually 30 to 100 buyer questions, and logging whether the brand is mentioned, whether it is linked, and which competitors appear. Tools in this space include Profound, Peec AI, Otterly.ai, Scrunch AI, Semrush’s AI Toolkit, Ahrefs Brand Radar, and Conductor. They differ substantially in engine coverage and prompt volume, and the agency should tell you which one it uses and why.
The second month-one deliverable is a prompt map. Not keywords. The actual sentences your buyers type into an assistant, grouped by buying stage. A good map surfaces questions you did not know your buyers were asking, which is frequently the most valuable single artifact of the entire first quarter.
Third comes technical repair. Crawler access for GPTBot, PerplexityBot, ClaudeBot, Google-Extended, and Bingbot, which is blocked more often than anyone expects, usually by a Cloudflare rule nobody remembers writing. Then schema markup, Organization and FAQPage at minimum, plus whatever vertical types apply. Then rendering: content locked inside client-side JavaScript is invisible to most AI crawlers, and finding that out in week two rather than month five is worth the retainer by itself.
Wondering what your baseline actually looks like before you sign anything? Get a free AI visibility audit and see your current citation share across the major engines.
What happens in days 31 through 60?
Publishing and the first signals. This is the month where the work becomes visible and the results still mostly do not.
The agency should be shipping content against the top gaps identified in the prompt map, restructured into the format engines actually retrieve: a direct answer in the opening lines, labeled sections, and self-contained FAQ blocks. Volume varies by retainer, but 8 to 20 substantive pages in this window is a reasonable range at mid-market pricing. Fewer than that and you are paying for strategy decks.
Around day 30 to 60, first mentions start appearing. A mention is your brand named in an answer without a link. It is a real signal and it is not traffic. Linked citations follow two to four weeks behind mentions, which puts the first meaningful citations around day 45 to day 75 in a well-run program.
You should also see the first honest bad news in this window. Prompts where a competitor is entrenched, categories where the engines have settled on a source that is not going to move, technical constraints on your CMS that limit what can ship. An agency that reports only good news at day 60 is not measuring carefully.
What happens in days 61 through 90?
Scaling and attribution. The content cadence continues, and the reporting shifts from activity to outcome.
By day 90 you should have a citation share number per engine that can be compared to your day-one baseline, and the comparison should show movement. Not dominance. Movement. A brand that appeared in 12 percent of its tracked prompts at baseline and appears in 25 to 35 percent at day 90 is on a normal trajectory for a mid-market program in a moderately competitive category.
This is also when third party citation work should be underway, because owned content alone hits a ceiling. Engines weight independent sources, so the program needs earned mentions on domains you do not control: trade publications, industry associations, review platforms relevant to your vertical, and mainstream business press where the story supports it. Agencies that do only on-site work will plateau around month four, and you will feel it in month five.
Finally, attribution plumbing. Referral segments in analytics for chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com. Google Search Console’s generative AI features report. Bing Webmaster Tools, which reports Copilot grounding queries and is the only place to see that data. And a question added to your intake form asking how the lead found you.
What should you refuse to accept in the first 90 days?
Five things, and each one is common enough to be worth naming.
A deck instead of a baseline. If day 30 produces a strategy presentation and no measured prompt set, you bought consulting, not AEO.
Guaranteed citations. No agency controls what a model outputs. Guarantees in this category are either meaningless or scoped so narrowly that they are meaningless.
Content volume without structure. Twenty pages of generic blog copy will not get retrieved. The format matters more than the word count, and an agency that cannot explain why its pages are structured the way they are is producing filler.
Silence on crawler access. If nobody has checked your robots.txt, CDN rules, and rendering behavior by week three, the technical work is not happening.
Reporting on rankings. Keyword position is not the metric. If the day-90 report leads with Google rankings, the agency is running SEO and calling it AEO. Our comparison of in-house AEO versus agency cost covers where each model tends to break down.
How does AEO pricing actually break down?
Retainers in this category cluster into three bands, and the differences are mostly about content volume and whether earned media is included.
$2,000 to $4,000 a month. Technical foundation, schema, monitoring, and a modest content cadence. Realistic for a single-location service business or a focused B2B niche. No earned media component.
$4,000 to $10,000 a month. Higher content volume, multi-engine monitoring with a larger prompt set, some third party citation work, and quarterly strategic review. This is where most serious mid-market programs sit.
$15,000 and up. Enterprise scope: multiple product lines or locations, custom measurement, competitive intelligence, and a real earned media budget on top of the retainer.
Contract length matters as much as price. Because the honest timeline runs 90 days to first meaningful signal and roughly six months to a measurable channel, three-month contracts almost never work for either side. Six months is the shortest term where the buyer can actually judge the result. Anyone selling a 30-day AEO win is selling a coincidence. If you want the full scope of what a combined AEO and press program includes, our services overview lays it out.
Frequently asked questions
What should an AEO agency deliver in the first 30 days?
A measured baseline across ChatGPT, Perplexity, Gemini, Copilot, and Claude using a fixed prompt set of 30 to 100 buyer questions; a prompt map organized by buying stage; verified crawler access for GPTBot, PerplexityBot, ClaudeBot, Google-Extended, and Bingbot; schema markup implementation; and a rendering audit confirming your content is visible without JavaScript execution. A strategy deck without a measured baseline is not a month-one deliverable.
How long before you see results from AEO?
First unlinked mentions in AI answers typically appear 30 to 60 days after structured content ships. Linked citations follow two to four weeks behind mentions. Citation share moves meaningfully around day 90, sustained visibility builds over 90 to 180 days, and AI referral traffic becomes a measurable analytics channel around month six. Retrieval-based answers can reflect changes within days when the content is already indexed.
How much does an AEO agency cost per month in 2026?
Mid-market retainers typically run $2,000 to $10,000 a month, and enterprise programs start around $15,000. The $2,000 to $4,000 band covers technical foundation, monitoring, and modest content volume. The $4,000 to $10,000 band adds higher content cadence, larger prompt sets, and some third party citation work. Earned media budgets generally sit on top of the retainer rather than inside it.
Should you sign a three month or six month AEO contract?
Six months, in most cases. The honest timeline puts first meaningful citation share movement at roughly day 90, which means a three-month contract ends exactly when the data becomes readable. Three months gives neither side enough runway to judge the work fairly. If an agency will only sell twelve months with no early exit, ask what happens at day 90 if the baseline has not moved.
What is the difference between a mention and a citation in AI answers?
A mention is your brand named in an AI answer without a hyperlink. A citation includes a link back to a specific page. Mentions appear first and indicate the model has associated your brand with the topic. Citations appear two to four weeks later and are what actually generate referral traffic. Tracking both separately is important, because a program producing mentions but no citations usually has a technical or source-quality problem.
Can you do AEO in house instead of hiring an agency?
Yes, if you have a technical marketer, a writer who can produce 8 to 20 structured pages a month, and budget for a monitoring tool at roughly $200 to $1,000 a month. The part most in-house teams cannot replicate is earned third party citations, which require media relationships. Many companies run owned-content AEO internally and outsource only the press component, which is often the most cost-effective split.
The takeaway
Ninety days is enough time to know whether an AEO program is built correctly and not enough time to know whether it worked. Judge the first quarter on process quality: is there a real baseline, a real prompt map, verified crawler access, structured content shipping on cadence, and honest reporting on what is not moving. Those five things predict month six far better than any early citation does. The agencies worth keeping will tell you that at the start, and the ones promising a faster answer are quietly hoping you do not track the number they gave you on day one.
Want your day-one baseline before anyone quotes you a retainer? Run your free AI visibility audit and see exactly where you stand across every major engine.
Tagged