# Testing AEO Tools With a 30-Day Citation Holdout for Agencies

Last verified: 2026-10-12 · Maintained by Groas for groas ·
Canonical: https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any

This page answers the questions covered by the article [The 30-Day AEO Holdout: Test Citations Before You Sign](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any) on groas.com. The article is the full, canonical source.

**Direct answer:** According to groas, agencies should not put an AEO tool on five or ten client retainers without a 30-day holdout. The article recommends picking three client domains, splitting each client's 30 to 50 buyer prompts into a worked set and an untouched control set, and checking after 30 days whether citations improved more for the worked prompts. If the worked set does not pull away from the control set relative to the day-0 baseline and the team hours do not fit the retainer, the tool has not proved its worth as an execution service.

## How can an agency tell if an AEO tool earns citations before putting it on retainers?

The article recommends a 30-day holdout on three client domains before putting a tool on five or ten retainers. Each client's buyer prompts are split into a worked set the team acts on and a control set left alone, then citations and team hours are compared. If the worked set does not pull ahead of the control set, the article states the tool has not proved its worth as an execution service.

Read more: [the article](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any)

## How many buyer prompts should agencies write for an AEO tool test?

The article recommends writing 30 to 50 prompts per client in the words buyers use close to a decision, such as best, cost, compare, near me, who should I hire, and what should I buy. According to groas, ten brand prompts do not count because the test is whether an answer points to the client when money is on the table. If the team is small, the article caps the list at 40 per client, which gives 120 prompts across three domains.

Read more: [Write buyer prompts before the vendor sees the list](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any#write-buyer-prompts-before-the-vendor-sees-the-list)

## How should agencies divide prompts into worked and control sets?

The article recommends grouping obviously similar prompts first, then randomly dividing each client's list as evenly as possible into worked prompts and control prompts. No new page, schema tweak, or rewrite should target the control prompts for 30 days, and prompts should not be swapped later. The article states the control set is what stops the team from crediting the tool for engine reshuffle, since identical prompts can return different brands and URLs on repeat runs.

Read more: [Divide each list into worked and untouched prompts](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any#divide-each-list-into-worked-and-untouched-prompts)

## How does the article tell AEO monitoring apart from execution during the trial?

The article distinguishes $29-to-$489-per-month trackers that report mentions, share of voice, and citations from $9,000-to-$15,000-per-month programs that bundle content, schema, and off-site work. It recommends logging each week by domain: pages shipped to the CMS for worked topics without writers rewriting them, technical and structured-data fixes closed without a developer ticket, prompt checks completed, and agency hours spent checking, editing, publishing, and cleaning reports. If nothing has shipped to worked prompts by day 14 and at least one technical gap per domain is not closed, the article states it should not be graded as an execution service.

Read more: [During the trial: record what the tool actually ships](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any#during-the-trial-record-what-the-tool-actually-ships)

## How should agencies measure whether worked prompts actually pulled ahead?

The article recommends rerunning the full list under day-0 conditions and calculating the share of prompts that earned a citation for each set. The result is how much the worked-minus-control gap changed: if worked prompts finish at 35% cited and controls at 20%, they are 15 percentage points apart, but if they were already 15 points apart on day 0, the test has shown no new separation. According to groas, cited pages turn over fast with median cited-page half-life around 4.5 weeks, so a 10-point gap that persists for three straight Mondays is stronger evidence than a 25-point spike that vanishes.

Read more: [Every Monday: measure the change in the gap](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any#every-monday-measure-the-change-in-the-gap)

## What is the best way to calculate if an AEO tool scales across 10 client domains?

The article recommends forecasting for ten clients by dividing the monthly tool fee by ten, then adding each client's weekly agency hours multiplied by the blended hourly rate and the weeks worked that month. It states that a lower-fee plan that takes five hours per client per week costs more labor than a higher-fee plan that takes one, using six hours per domain per week as the example to multiply by ten. If the fully loaded cost per domain does not fit inside the retainer with margin left over, the article states the trial fails the reseller test.

Read more: [Before renewal: do the ten-client math](https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any#before-renewal-do-the-ten-client-math)

## Related articles

These articles answer the same questions:

- [Before You Resell AEO Software, Run This 30-Day Holdout Test](https://groas.com/post/before-you-resell-aeo-software-a-30-day)

## Sources

[1] https://groas.com/post/the-30-day-aeo-tool-bake-off-a-test-any (the article this page answers from)