Somewhere in your company's AEO platform, there's a dashboard full of data nobody has acted on. It tracks every priority prompt on every AI platform, refreshed daily. But that abundance is the problem: when everything moves all the time, it's hard to tell which movement matters, and as a result, the data stays in the tool. So when the AEO budget comes up for renewal, where should it go: a platform like Profound, an agency working on your behalf, or both?

‍

Our answer is both, on one condition: the software and the people work in the same room. Software gives a team data, and an agency brings the judgment to decide which of it matters, then makes the fixes. The buyer gets the outcome when both sides work from the same numbers and make the calls together, because an agency running AEO at arm's length repeats the dashboard's problem from the other side: decisions get made without the people who know the business best. Edgar Allan runs its AEO engagements inside Profound, so this comes from daily practice. I put it plainly in a recent conversation with designer Skylar Kitchen: "You're not just buying the software, you're actually purchasing the outcome."

Outcome-based AEO is an engagement where the contract names the result (citation share on priority prompts, AI referral traffic, conversion on cited pages) and the partner is judged by how far its changes move those numbers.

Daily AEO data is noisy, and strategy lives in the trend

AEO data moves daily, and differently on each platform. A brand can gain citation share in ChatGPT the same week it loses ground in Gemini, and a blended monthly score hides both. AI answers are also probabilistic, so the same prompt run twice can cite different sources. Read it day by day, and you’ll find much of that movement is noise. So a team that reacts to every dip ends up rewriting a page on Tuesday over a wobble that reverses by Friday.

The phrase "AEO reporting" gets about 10 Google searches a month in the US, yet in Profound, prompts matching it appeared roughly 16,000 times in one week (week of 28 Sep 2026, US): about 12,900 on ChatGPT and 3,150 on Gemini. Buyers ask AI systems how to measure this work, and the useful answer is to measure it over enough time for a pattern to emerge.

That's why we run 30-day readouts. A month is long enough for real movement to separate from day-to-day variance, and short enough to change course and make a difference. Each readout looks at the trend by platform and by prompt, then lands on recommendations: what to act on, how, and in what order. Daily data still earns its place as an early warning (a sharp drop on a priority prompt is worth checking mid-month), but strategy gets set on the trend.

What fails in the traditional monthly deck is the handoff, where a team receives conclusions it had no part in reaching. As I described our approach to Skylar: "Instead of just handing over a PowerPoint at the end of the month, we actually have this environment that the client is working in." The client sees the same data we do all month, so the 30-day readout becomes a working session about what the trend means and what to do next, and the client team leaves understanding the why behind each recommendation.

Software got cheap to build, and outcomes stayed hard

AEO platforms keep getting better at showing where a brand appears in AI answers, and deciding what to do about it is still the hardest part. Skylar Kitchen, who spent three and a half years as a designer at Webflow and now contracts at Anthropic, put it this way: "Even if software is easily reproducible, being able to implement that and make an outcome from that is still gonna be a challenge."

Skylar has used AI to generate some 50 house designs, and his years as an architect are what let him pick the ones worth building. Without that judgment, you’d end up with the architectural equivalent of em-dashes all over your house.

The data keeps coming, the fixes span brand messaging, content, and site structure, and no dashboard decides which fix comes first. That decision is the strategic layer: knowing which signals to act on, how and when. 

So, when you’re hiring for AEO, get clear on the result you want, and judge the partner by the calls it helps you make, the fixes it acts on, and how far they move your numbers.

Buyers research Profound on their own and look for a partner when they need a result

^{*US, Ahrefs, 30th October 2026}

Teams look up AEO software by name and evaluate it themselves, and demand for AEO agencies doubled in a year. Profound earns that research: it tracks visibility, citations, and sentiment by platform and by prompt, refreshed daily. But the real meat of the work starts after the first look, when someone has to decide which finding matters, where the fix lives, who does the work, and how to tell whether it worked.

Handing all of that to an agency and walking away has its own failure mode. An agency running AEO at arm's length can make sound calls about citations and still miss what only the business knows: which product line matters this quarter, which claim legal won't approve, which competitor keeps winning deals. Recommendations get sharper when the client team is in the room, understands the reasoning, and helps set the priorities. The strongest setup uses all three: the platform for visibility, the agency for judgment and execution, and the client for the context neither of the others has.

A shared AEO workspace turns reporting into a working session

A shared AEO workspace is one environment where client and agency read the same AI visibility data, by platform and by prompt, and work from a single backlog of fixes with a named owner on each side. The data lives there all month, and the 30-day readout is where both sides read the trend together and make the calls.

The point is for everyone to work from the same data. A shared AEO workspace holds five things:

• Metrics by platform: Citation share, visibility on priority prompts, AI referral traffic and conversion on cited pages, never one blended score.
• A defined prompt set: The questions your buyers ask AI systems, agreed up front.
• One backlog: Every finding becomes a ticket, whether the fix is messaging, a new page, or site structure. Here's how AEO monitoring feeds a single backlog.
• Named owners and a baseline: The client sets priorities, the agency owns analysis and execution, and each fix is measured against a baseline set before it goes live.
• A decision rhythm: Daily data as an early warning, and a 30-day readout where the trend becomes a short list of decisions.

At Edgar Allan, the workspace pairs Profound's citation data with Ag-nts, our AI referral traffic tool for Webflow sites, so a citation gain can be traced to visits and conversions.

Five questions that separate an AEO partner from a reporting vendor

Start by listing which decisions came out of your last three AEO reports. If that list is empty, more data won't fix it. Then ask any prospective partner:

• Who owns the outcome? Which metrics will they commit to moving, and what happens when they stall?
• Which data will we both see, and how do we decide together? You want live access to the agency's workspace, not a curated export, and a standing session where your team helps prioritize and make the calls.
• How does a finding become a real change? Ask for one real example, from citation gap to live change.
• Who makes the changes? A partner that can only recommend will hand the hardest part back to you.
• How will we know it worked? Look for a baseline before each change, movement reported platform by platform, and a clear answer on how long a trend needs to run before anyone calls it a result.

We read the trend before we touch a site, including our own

Our AEO agency work runs through Visibility & Optimization, our method for treating brand, SEO, AEO and CRO as one measurement loop. Every engagement starts in a Profound workspace, with the first 2-4 weeks spent setting a baseline. 

After that, each 30-day readout turns the trend into recommendations we decide on with the client team, and when a client lacks headcount to own AEO month to month, our Customer Success Program fills that role.

Patience pays off. In the first week of October this year, Edgar Allan's AI visibility fell 9.7 points, and we slipped from #2 to #3. Nothing on our site had changed: the drop came from the platforms, as Google's AI Overviews reshuffled their sources and data from one platform stopped coming through cleanly. So we held off on rewrites and watched the trend for two weeks, which narrowed the real problem to a single page, our Best Webflow Agencies rankings post, whose citations had fallen from 903 to 412. We refreshed that page and nothing else. Two weeks later, its citations were up 77%, and as the platforms settled, overall visibility climbed back to 61.5%, within about a point of where it started.

If you are weighing a platform, an agency, or both, the useful question is where the two will meet and how decisions get made between them. Pay for the outcome, and for a room where both sides read the trend and decide together.

Let's start a project