An AEO Programme Budget You Can Defend
Fund three things in order and defer the rest: technical reachability (crawlable pages, working robots rules, content present without JavaScript), a measurement baseline you can compare against later, and content on the buying questions you would want to be cited for. Defer tooling, agency retainers and content at volume until the first three exist, because none of them improve an outcome you cannot yet see. The number nobody can honestly give you is expected return — there is no reliable published conversion rate for AI citations, the traffic is partly invisible, and anyone quoting you a figure is quoting a guess. Budget it as a reachability and content investment that also serves search, which is defensible on its own terms.
We sell in this category, so treat the recommendation to spend less at the start as coming from an interested party who is arguing against their own quarter.
The order matters more than the total
| Priority | What | Why first |
|---|---|---|
| 1 | Technical reachability | Nothing else works if pages cannot be retrieved |
| 2 | A measurement baseline | Without it, later spend cannot be evaluated |
| 3 | Content on buying questions | The thing that can actually be cited |
| 4 | Tooling | Automates measurement you are already doing |
| 5 | Content at volume | Only after the first three show something |
| 6 | Agency retainer | Only with a defined, checkable scope |
What the first three cost
Reachability is usually days of engineering, not a budget line — fixing client-side-only rendering, a robots rule, a missing sitemap entry. The measurement baseline is a fixed prompt set and a few hours a month of sampling, which a spreadsheet handles. Content is the real cost and it is the one that also serves search, which is why it survives scrutiny when the AEO framing does not.
The number you will be asked for
Expected return. There is no honest figure: click-through from AI answers is partly unmeasurable, citation-to-customer rates are not reliably published by anyone, and the channel is young enough that last year's numbers describe a different system. What you can offer instead is a cost, a baseline, and a review date — and saying so is more credible than a projection that will not survive its first quarter.
What to cut first if the budget shrinks
Tooling, then volume, then the retainer. Keep reachability and the baseline: the first is cheap and structural, the second is what makes any future decision evaluable. A programme with no baseline spends for a year and cannot say what happened, which is the most common way this budget gets cancelled.
FAQ
Should we hire an agency for this? — Only against a scope you can check: specific pages, specific fixes, specific measurements with a stated method. Retainers for general AEO improvement are unevaluable by construction, and that is a problem regardless of the agency's quality.
Do we need a tool at all? — Not to start. A fixed prompt list and a spreadsheet measures the same thing for a small competitive set. Tooling buys consistency, history and time back — real value, and not a different answer.
How long before we can tell if it worked? — At least a quarter, and longer for anything content-led. Expect the first two months to produce no legible signal, and plan the review for a date where there might be one.
Put this to work on your own website.
Fenn finds what your customers ask, drafts the articles and site fixes, and measures what ChatGPT, Claude, Gemini, Perplexity and Grok say about you — with every change waiting for your approval.