See what to expect by Day 30, 60 and 90, from a clear baseline to measurable AI visibility progress.
Ninety days is a reasonable point to ask whether a marketing engagement is working. It’s enough time for an agency to learn the business, identify problems, implement meaningful work and begin measuring what changed. It’s also early enough that nobody should pretend every long-term result has arrived.
That question gets more complicated with Answer Engine Optimization because companies are being sold with very different expectations. Some are promised prominent placement or recommendations across AI search within a few months, while others receive a long list of activities without much explanation of what those activities are supposed to accomplish. Neither approach gives a client a particularly useful way to evaluate whether the engagement is actually working.
I think the first 90 days should be simpler than that. By Day 30, I want a diagnosis and a clear understanding of where we started. By Day 60, I want to see meaningful work based on what we discovered. By Day 90, I want a before-and-after assessment that tells me what changed, what didn’t and what we’ve learned. An agency can’t guarantee what Gemini, Perplexity, Claude, ChatGPT, Google AI Overviews or other AI systems will say about your company on Day 90, but it can absolutely be accountable for what it did during those 90 days.
What Should You Have After the First 90 Days of AEO?
One of the easiest mistakes is evaluating an AEO retainer by activity. Twelve articles were published, 40 pages were optimized, schema was added and a dashboard was created. Those things may all be useful, but completing a list of deliverables doesn’t tell you whether the agency identified and worked on the right problems.
I’d evaluate the first quarter differently. Each stage should answer a different question, and the answers should build on one another. The first month should tell us what’s wrong, the second should show what we’re doing about it and the third should give us enough evidence to evaluate whether we’re heading in the right direction.
By Day 30, You Should Have a Diagnosis
The first month should establish where you are starting by identifying the questions potential customers are actually asking and determining how your business appears when those questions are tested across relevant AI platforms. Which competitors appear instead? Are you mentioned but rarely recommended? Is information about the company inaccurate? Which websites and third-party sources are being cited? Are there entire categories of questions where the company is essentially invisible?
The agency should also be able to explain the likely reasons. Maybe important expertise isn’t documented anywhere online, or perhaps the website doesn’t clearly communicate a service that is important to the business. Third-party information may be inconsistent, competitors may have substantially stronger authority, or there may be technical problems making important information harder to discover. You don’t need a 70-page audit to prove that work happened, but you should have a diagnosis that makes sense to the people running the business.
There should also be an agreed-upon scoreboard. That’s particularly important because AI visibility isn’t as stable as a traditional ranking report. Meltwater’s August 2026 AI Search Visibility Report analyzed approximately 7.3 million citations across eight major AI platforms and found that source preferences can vary based on factors including the prompt, category, location, timing and retrieval system. The researchers specifically point to repeated measurement as more useful than a one-off snapshot.
If an agency tells you three months from now that visibility improved 28%, everyone should already understand exactly what that percentage represents, which questions were measured and what the new result is being compared against.
By Day 60, You Should Be Able to See What Changed
By the second month, the engagement should have moved well beyond diagnosis. You should be able to point to actual things that are better because of what was discovered during the first month, although the specific work should look different from one company to another.
One business may need substantial website improvements, while another may have a strong website but very little evidence of its expertise outside of it. Another may discover that its most important customer questions aren’t answered anywhere, or that inconsistent information across listings and industry sources is making the company harder to understand. That’s why I’d be concerned if every client received essentially the same Day 60 deliverable list. The work should reflect the problem that was actually uncovered.
This is also where the difference between measurement and execution becomes important. MediaSense’s 2026 report on AI search identifies a central challenge for brands: turning AI visibility insights into action rather than simply adding another tool or dashboard. That’s a useful distinction for evaluating an agency as well. By Day 60, the engagement should be changing something, not simply becoming better at describing the problem.
That might mean new articles, but it could just as easily mean original research, stronger service pages, executive commentary, case studies, videos, clearer company information or better explanations of work the organization already does. By this point, I care much less about how many pieces of content were produced than whether the company now has useful information and authority signals online that weren’t there before, and whether there was a clear reason for creating them.
By Day 90, You Should Have a Before-and-After
The third month is where the original baseline becomes valuable. The agency should return to the same commercially important questions, use the same methodology and compare what was happening at the beginning of the engagement with what is happening now.
That comparison should go deeper than a single visibility score. Are you appearing for more commercially important questions? Are AI systems describing the business more accurately? Is your website appearing as a source where it wasn’t before? Are certain pieces of new content being cited? Are competitors still dominating the same areas? Perhaps most importantly, are patterns emerging across different platforms rather than the agency focusing on whichever individual result happens to look best that week?
Those platform differences aren’t trivial. Yext’s 2026 AI citation research analyzed 155.5 million location-grounded citations across 1,623 brands and found significant patterns in where AI systems sourced information. Among the findings, 80% of those citations pointed to sources a brand could influence, particularly its website and listings. Yext’s broader research also shows meaningful differences in citation behavior among AI models, which is one reason a company’s AI visibility shouldn’t be judged by what happens on a single platform.
A business may make progress in one AI environment while remaining largely absent from another, and the evidence contributing to those answers may differ as well. Part of a useful AEO engagement is beginning to understand those patterns instead of reducing AI visibility to whether one particular platform happens to recommend the company.
I’d also look outside the AI visibility dashboard. Identifiable referral traffic from AI platforms may be emerging, but branded search, direct traffic, leads and CRM outcomes can provide additional context. Someone may discover your company in an AI-generated answer, remember the name and search for it later, which means analytics may attribute that visit to search rather than the AI interaction that introduced the business in the first place. No single metric is going to explain everything, so the goal should be to build enough evidence across several measurements to understand whether something meaningful is happening.
The report should include what didn’t work as well. If the agency created several new resources and none appears to have influenced visibility yet, that is useful information. If one platform is responding while another remains flat, that matters too. A useful Day 90 report should help the business decide what to do next, which is difficult if the agency only presents the evidence that makes the engagement look successful.
What If the Numbers Haven’t Moved?
Sometimes the Day 90 answer is going to be that AI visibility hasn’t changed very much yet, and that doesn’t automatically mean the engagement failed. If the agency uncovered significant problems, corrected them, built substantially better resources, clarified confusing information and established measurements that didn’t previously exist, there may be good reasons to continue while watching how those changes are discovered and used.
What matters is whether the agency learned something during the process. Maybe third-party authority is weaker than expected, or perhaps the company appears for informational questions but disappears when someone asks an AI system for a recommendation. New content might be getting discovered without being cited yet, or a competitor may continue to dominate because it has years of credible external evidence that can’t realistically be recreated in one quarter. Those findings should influence what happens during the next 90 days.
There is a meaningful difference between “the results haven’t arrived yet” and “we don’t really know what happened.” An agency that can explain what it learned, what appears to be limiting progress and what it intends to test next is giving the client something useful to evaluate. An agency whose primary explanation is that it published 24 blogs isn’t giving the business nearly as much to work with.
Ninety Days Is Enough Time to Expect Accountability
I wouldn’t judge an AEO agency after 90 days solely on whether your company earned a recommendation from a particular AI platform or whether AI referral traffic suddenly doubled. Those outcomes involve systems the agency doesn’t control, and anyone guaranteeing a specific result across AI search on an exact schedule is promising something they can’t actually command.
I would judge whether the agency understands the business better than it did on Day 1 and whether that understanding produced meaningful work. There should be a clear baseline, a diagnosis the client understands, improvements tied directly to what was discovered and a before-and-after assessment using the measurements everyone agreed on at the beginning. Even when the results are still developing, the client should have a much clearer picture of where the company stands across AI search and what needs to happen next.
Ninety days may not be enough time to prove the full commercial value of an AEO program, but it is more than enough time to determine whether the engagement is becoming more intelligent, more measurable and more useful to the business. That’s ultimately what I would want from the first quarter: not a guaranteed recommendation from any particular AI system, but enough meaningful progress and evidence to make a smart decision about the next one.
Related:
Ask us! What questions do you have about content, strategy, pop culture, lifestyle, wellness, history or more? We may use your question in an upcoming article!
Like MediaFeed’s content? Be sure to follow us.
This article originally appeared on Fireball Agency and was syndicated by MediaFeed.co
