What the first 90 days of AI search work actually look like
"When will this start working?" is the first question an early-stage B2B founder asks about AI search, and it is a fair one. There is no guaranteed date, but there is a usual shape to the first 90 days, and knowing it is the difference between judging the work correctly and pulling the plug too early.
What happens in the first 90 days of AI search work?
The first month is structure: schema, a named author, and question-form pages. The second month is indexing and the first citations in AI answers. By the third month, hot inbound usually becomes a repeatable number rather than a one-off event, though the pace always depends on the market and existing competition, never guaranteed.
Month one: structure before visibility
Nothing is visible to a buyer in month one, and that is by design. The work is entity signals and page structure: Organization and Person schema naming a real author, question-form headings with a direct answer underneath, FAQPage markup with visible answers, and a canonical tag on every page so search engines and AI answer engines agree on which version of a page is the real one. Skipping this month to chase early rankings is the single most common reason AI search work stalls later.
- Weeks 1 to 2. Audit existing pages for missing schema, duplicate or missing canonicals, and any page without a named author.
- Weeks 2 to 4. Publish or rebuild the pages that answer the buyer's actual questions, each with one direct answer and one FAQ block.
Month two: indexing, entity signals, and the first citations
New and corrected pages get crawled and indexed on their own schedule, not on request. This is the month a founder is most likely to check daily and see nothing, which is normal. Behind that silence, entity signals are accumulating: the same author name, the same organization details, and consistent structured data across every page build the trust an AI answer engine checks before it names a company in a response. Hot inbound usually starts showing up somewhere in this window, as early traffic and the first citations in AI answers.
Month three: inbound becomes a repeatable number
By month three, the pages published in month one have had time to be indexed, trusted, and in some cases cited. What was a handful of early visits in month two usually becomes a number that repeats week over week. This is also the point where it is fair to judge the work: a page with no indexing, no impressions, and no citations after three months usually has a structural problem, not a timing problem, and is worth auditing against the fundamentals from month one before assuming the channel does not work.
AI search timeline vs cold outbound timeline
| Cold outbound | AI search (hot inbound) | |
|---|---|---|
| First meetings | Usually week one | Usually month two |
| What drives the timing | Sending volume and list quality | Indexing speed and entity signals |
| What happens if you stop | Meetings stop within days | Traffic keeps compounding from past work |
| Best for | Pipeline from day one | A compounding asset that keeps producing |
See what AEO and GEO actually are for the structure referenced above, and how to measure AI search visibility for the exact signals to check at each month.
Key takeaways
- Month one is structure and entity signals, with nothing visible to a buyer yet
- Month two is indexing, and usually where the first inbound and citations appear
- Month three is where inbound usually becomes a repeatable, week-over-week number
- A slow start after three months usually traces back to missing structure, not the channel itself
- None of this is guaranteed on a fixed date, which is why cold outbound runs from week one alongside it
Frequently asked questions
What happens in the first 90 days of AI search work?
The first month is structure: schema, a named author, and question-form pages. The second month is indexing and the first citations in AI answers. By the third month, hot inbound usually becomes a repeatable number rather than a one-off event, though the pace always depends on the market and existing competition, never guaranteed.
When does inbound actually start showing up?
Usually by the second month, once new pages are indexed and the entity signals around them have had time to accumulate. Some queries move faster if competition is thin; others take longer in a crowded category. There is no fixed date, only a usual range.
Does AI search work faster than classic SEO?
Not faster, but it can surface sooner in the answer itself rather than waiting for a page-one ranking. An AI answer engine can cite a well-structured new page once it is indexed and trusted, while a competitive page-one Google ranking for the same query can take considerably longer.
What if nothing shows up by month three?
Check indexing first: a page that is not indexed cannot be cited or ranked. After that, check the structure, the question-form heading, the direct answer, the schema, and the entity signals on the pages published so far. A slow start usually traces back to one of those, not to the channel itself.
Can an Indian SaaS company selling to US buyers expect the same timeline?
Yes. Search engines and AI answer engines evaluate page structure and entity signals, not the company's registered address, so an Indian SaaS company selling to US buyers works through the same three months on the same system, with reporting in its own time zone.
Does cold outbound keep running while AI search ramps up?
Yes. AI search only reaches a buyer who is already asking a question, and it takes months to build. Cold outbound reaches the much larger group who has not asked yet, and it can put a meeting on the calendar from week one, which is why Montazzo runs both from day one.
Where this fits in a full-stack pipeline
These 90 days describe the hot inbound half of the system, and they only reach a buyer who is already asking the question. Cold outbound reaches everyone else, the much larger group who has not searched yet, and it can put a real meeting on the calendar starting in week one, while the entity signals and structure above are still building toward their own first citations. Montazzo runs both with one team, so the same pages this article describes are the proof the outbound side references in its own messages. See cold email follow-up sequences that get replies for the outbound half of the same system.
Want both channels built by one team?
Montazzo runs hot inbound and cold outbound together for early-stage B2B companies, one team, one report. No SDR team required and live in three weeks.
Let's chat