A First AEO/GEO Client Calls for a Pilot and a 12-Point Check

Plan a 90-day pilot, create a buyer prompt set, baseline each engine, then address the gaps. That means grounding the work in SEO fundamentals, gathering real buyer questions, recording what each AI engine says today, and building a work plan around what is missing. The 12-point readiness check below shows how prepared you and the client are before you sign anything.

Checklist

ChatGPT, Perplexity and Gemini explain how to take on a first AEO/GEO client

We asked the three engines how to take on a first AEO/GEO client. All three grounded the work in SEO fundamentals, but each chose a different starting point. ChatGPT began with how to sell it: a foundation and pilot based on a prompt set and baseline testing. Perplexity began with commercial targets and a tracked prompt set, followed by an audit, and cited four outside sites: collystring.com, tofubofu.com, smartsites.com and thrad.ai. Gemini began with an AI visibility audit, was the only one to identify specific crawlers (OpenAI GPTBot and PerplexityBot), placed the greatest emphasis on third-party presence, and cited no sources.

A pilot gives the client a scope they can judge

A client asking for AEO/GEO 'now' usually wants their brand named when a buyer asks ChatGPT for a recommendation. That is not something you can guarantee, and ChatGPT explicitly warns against promising it. You can promise a defined body of work and a clear before-and-after comparison.

Build the scope around deliverables within your control. Report progress on a fixed set of questions, using the same measurement approach at the beginning and end.

1. Commercial targets define the scopeFor which products, buyer types and regions should the brand earn recommendations? Anything outside that list falls outside the pilot's scope.
2. The prompt set stays fixedGather the questions buyers actually ask, then freeze the list to keep the end-of-pilot comparison fair.
3. Baseline testing records the starting pointRun every prompt in each engine and record whether the brand appears, who appears instead, and which sources are cited.
4. The audit sets prioritiesAssess content, crawler access, structured data and third-party presence against the gaps found in the baseline.
5. Delivery and re-measurement show changesComplete the prioritized work, rerun the same prompts on a schedule, and report changes for each engine.

What AI engines say right now

We asked ChatGPT, Perplexity, Gemini “Client wants AEO/GEO now, never done it before. how do you actually go about it?” on 2026-11-05. Here is what came back.

The prompt set comes before site changes

First-timers often rush the prompt set, even though it determines whether the rest of the engagement is meaningful. A keyword list is not a prompt set. Buyers give engines complete questions with context, including their team size, budget and the tool they are leaving.

Gather questions from sales call notes, support tickets, demo requests and lost-deal reasons. Group them by type, since different types tend to call for different fixes. Here is an example for a hypothetical payroll SaaS client we will call PayrollCo:

ChatGPT's proposed range of 50 to 150 questions is a reasonable guide. With fewer, one unusual answer can distort the picture. With far more, a first engagement becomes a data project.

• Category: "What's the best payroll tool for a 20-person agency?"
• Problem: "How do I pay contractors in three states without hiring an accountant?"
• Comparison: "Is PayrollCo or its biggest rival easier for a team with lots of contractors?"
• Alternative: "What can I use instead of my bank's payroll service?"

The baseline captures what each engine says today

Test every prompt in every engine your client's buyers actually use. Record three details for each answer: whether it names the brand, which competitors it names, and which sources it cites. Sources matter just as much as names because they show where the engine learns about the category. A basic sheet with one row for each prompt and engine is enough:

Be prepared for a bleak baseline. Of 61 sites GapCite has tested through its free scan, 48 (79%) were not named by Perplexity or Gemini for any of the buyer questions tested. A blank baseline is normal and provides a clear starting point for measuring change.

Answers vary across engines, too. When we asked about first AEO/GEO clients, Perplexity cited four outside sites, while Gemini cited none. Report results for each engine rather than combining them into one score.

A self-audit shows whether this client is ready for a pilot

Go through this check with the prospect before sending a proposal. Count the yeses. Every no identifies either work the pilot's first month must cover or something the client must resolve before you begin.

The gaps become a work plan

After the baseline and audit, the work plan largely takes shape on its own. Each gap suggests a particular fix, and most fixes are familiar SEO and PR tasks serving a new target.

Set priorities by commercial value, as Perplexity recommended. A practical rule: address gaps in questions close to a purchase decision, such as comparisons and alternatives, before tackling broad educational questions.

Worked example: the sample log above shows that PayrollCo does not appear for "What's the best payroll tool for a 20-person agency?" in either engine, while Perplexity cites two review sites and an accounting blog. This points to three tasks. Publish a page about payroll for agencies of 10 to 50 people, answering the question in the first paragraph and including PayrollCo's actual agency pricing and setup steps. Complete the listings on both review sites. Pitch the accounting blog a contractor-payroll case study featuring an agency customer.

FAQ