---
title: "5 Cheap AEO Promises I Wouldn’t Buy on a Small Budget"
description: "Cheap AEO packages turn countable changes into claims of progress. Here are five promises I would reject, and the test I would demand instead."
url: "https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/25bf3e36-f52b-47ac-a983-41813965fd2f.png"
published: "2026-11-05T05:48:54.319Z"
modified: "2026-11-05T05:48:54.372Z"
---

November 5, 2026 · 10 min read

# 5 Cheap AEO Promises I Wouldn’t Buy on a Small Budget

[](https://groas.com/author/alexander-perelman)

[Alexander Perelman](https://groas.com/author/alexander-perelman)Personally Tested

Head Of Product @ groas

[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A lavishly wrapped gift box with a coral bow and a $590/mo tag opens to reveal only shredded paper and a tiny chart—impressive packaging, empty results.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/25bf3e36-f52b-47ac-a983-41813965fd2f.png)

In this article

1. [Myth 1: Add FAQ blocks and schema and the citations will follow](#myth-1-add-faq-blocks-and-schema-and-the-citations-will-follow)
2. [Myth 2: More ChatGPT mentions prove the service is paying for itself](#myth-2-more-chatgpt-mentions-prove-the-service-is-paying-for-itself)
3. [Myth 3: AI answers already made normal SEO the wrong bet](#myth-3-ai-answers-already-made-normal-seo-the-wrong-bet)
4. [Myth 4: A lab test found the winning tactic, so it will work for you](#myth-4-a-lab-test-found-the-winning-tactic-so-it-will-work-for-you)
5. [Myth 5: GEO is branding, so stop asking how we will measure it](#myth-5-geo-is-branding-so-stop-asking-how-we-will-measure-it)

A founder forwarded me a [$590-a-month AEO offer](https://ampdigital.marketing/) promising FAQ fixes, schema and a weekly ChatGPT mention report. I told him to pass. Every deliverable was easy to count; none, as presented, showed how the work would bring in a customer. That is the cheap-AEO trick: sell visible activity as commercial progress, then call the gap between the two “branding.”

## Myth 1: Add FAQ blocks and schema and the citations will follow

I understand why this one sells. Every ranking AEO article seems to list FAQs and schema as quick citation wins, and the big [158,099-answer study](https://aihospitalityalliance.com/) gives sellers numbers to point at. Across different websites, pages with FAQ blocks were cited 18.1% more, summary boxes 20.9% more and schema 7.5% more. Put those figures beside a before-and-after screenshot of your page, and a $590 retainer starts to look like an investment.

I used to tell clients a version of the same thing about featured snippets in 2023. I was wrong in the same way. Strong sites already tend to format pages well, and strong sites already get picked up. The formatting rides along with that strength. **It does not create the strength.**

When the researchers held the website constant and compared pages on the same domain, the lift vanished. FAQ blocks came in at negative 3.4% and summary boxes at positive 6.9%; neither result was statistically significant. Retrieval explained 32% of citation variation against 2.8% for all page features combined, a 12x gap. A separate [Ahrefs test of 1,885 pages](https://geoaura.world/) added schema and compared the results with 4,000 controls. AI Overview citations fell 4.6%, while AI Mode rose 2.4% and ChatGPT rose 2.2%. That is noise, not a lever.

None of this means an FAQ is forbidden. A clear answer can help a reader, and a vendor can certainly ship a schema change. It means the invoice should not treat either change as evidence that the page now has a better shot at being retrieved. Ask which retrieval problem the work addresses and how the seller will check whether it changed. If they cannot name the problem, **they are billing you for decoration.**

![Ink cartoon of a marketer polishing FAQ blocks while a robot cites a competitor](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/a4051758-5ade-4b70-a7a3-593717c27b8f.png)

## Myth 2: More ChatGPT mentions prove the service is paying for itself

A weekly mention chart can climb if you choose enough prompts. That is why vendors love it. The chart is real enough; the conclusion printed under it is the problem. To see the gap, keep the five things it often collapses into one number separate:

1. **Mention:** your name appears in an answer.
2. **Citation:** the answer links to you.
3. **Referral visit:** someone clicks through.
4. **Qualified enquiry:** that person asks for a price for a job you actually do.
5. **Revenue:** they pay.

Each step can lose people. Counting the first as proof of the fifth is how a $349-a-month retainer looks busy while your pipeline stays flat. One practitioner measurement guide makes the distinction plainly: track the five separately, and [do not book the full value of an unclosed opportunity as revenue](https://panovistamarketing.com/) because one AI referral asked for a quote. I would rather see a small, honest referral number than a large mention number pretending to be sales.

The leak is not just a matter of bookkeeping. Across 100 sites, [ChatGPT sent 0.62% of the traffic Google organic sent](https://searchengineland.com/ai-search-branding-channel-498618). Another dataset found organic delivered 39 times the traffic of AI and about 30 times the conversions. And 632 controlled tests could not tie AI visibility to later branded or direct visits. Those findings do not make every mention worthless. They do make “mentions went up” a weak answer to “did this pay for itself?”

Before paying for the chart, I would write down the baseline: referral visits, qualified calls and closed jobs from AI answers over the last 60 days. Then I would write down the number that makes the retainer worth renewing. A vendor can argue for a different objective, but they have to name it before the work begins. Otherwise, **you are buying the chart, not the customers.**

## Myth 3: AI answers already made normal SEO the wrong bet

This pitch usually arrives as a chart: AI traffic climbing fast, Google flat. Why keep paying for pages and links? The chart may be real at internet scale and still tell you little about where to put a small budget. A growing slice is not automatically the slice that books your jobs.

The [Ahrefs benchmark across more than 118,000 sites](https://ahrefs.com/) from June to October put AI assistants at 0.46% of measured visits, against 27.55% from search engines, 31.08% direct and 16.77% paid. September alone had all AI combined near 0.43% of visits. I saw a similar panic around AI Overviews in paid search last year. Much of the overlap was informational, while the [real damage came from structural account problems](https://groas.com/post/ai-overviews-not-killing-google-ads-ctr-real-culprits). It is tempting to blame the new thing on the results page before checking the account or the work that already brings buyers in.

Those shares are aggregates across every niche Ahrefs could measure. They do not forecast a plumber in Austin or a Shopify store selling replacement filters. Your mix depends on who asks buying questions and where those questions get answered. But the proposed response still has to make sense: pulling money from search work that produces calls to chase a channel with little measured traffic reverses the burden of proof. **Keep funding the search work that books jobs until your own analytics show AI referrals doing that work instead.**

## Myth 4: A lab test found the winning tactic, so it will work for you

Every cheap deck seems to have a benchmark where one rewrite style beats everything else. The strongest recent example is _[LIVE-GEO](https://anythingengineoptimization.com/)_. It published rewrites as live pages and tested them on real engines rather than scoring text in a sandbox. Its CORE rewrite took the top spot 66.3% to 76.3% of the time on ChatGPT, Gemini and Grok across 80 fictional brands. That is a result worth reading. It is not a work order for your shop.

Read the next number before signing anything: the engines agreed on the same top product only 48.8% to 66.3% of the time. Same pages, same questions, different winners depending on which model answered. A tactic that wins on one engine and loses on another is not a law. It is a lab result involving made-up brands and product pages, in a preprint that has not been peer-reviewed. Fictional brands also lack the reviews, history and competitors around a real business. Those differences matter when you try to carry the result into your own market.

What the deck sells here is borrowed certainty. I would ask the vendor to show three buying prompts where my brand is absent today, the change they would make on my pages, and how they would check those prompts again in 30 days. That does not require them to guarantee a win. It requires them to make a claim that can survive contact with my actual business. **Demand observation on your brand, your pages and your buyers’ prompts.** If the answer is another benchmark PDF, keep the budget.

## Myth 5: GEO is branding, so stop asking how we will measure it

This is the one I find hardest to kill because half of it is true. AI answers can shape awareness and trust before anyone clicks. Judging GEO only on last-click leads can miss that influence, much as judging TV on coupon codes could miss what happened before the purchase. The [branding-versus-performance debate](https://searchengineland.com/ai-search-branding-channel-498618) makes that case carefully. I have some sympathy for it.

What I do not have sympathy for is using “branding” as permission to report mentions forever. If awareness is the objective, say which prompts and audiences matter and what change in prompted recall, branded search or assisted pipeline would justify keeping the work. Pick a measure that can inform a decision. A mention total without that decision attached is still a mention total, however expensive the slide design looks.

The same distinction applies to guarantees. Models, retrieval and policies change outside any agency’s control, which is why [agency guides flag guaranteed citations and number-one-in-ChatGPT promises](https://wcvdigital.com/). A seller can promise pages shipped, fixes made and reports sent. They cannot promise the date your shop starts getting cited. If a proposal offers guaranteed citations, I would keep my card in my pocket and ask what the team actually controls.

Say you spend $2,000 a month and someone offers a managed AEO add-on for $590. I would put three questions on one page before discussing the price:

1. **Work:** What ships in the next 30 days, on which URLs, to address which retrieval gap?
2. **Baseline:** Where do we stand today on the relevant prompts, mentions, referral visits and qualified calls?
3. **Decision:** What will the next report tell us to do: renew, change course or stop?

A proposal that cannot answer those questions is not cheap. It is expensive confusion at a low monthly price. Moving to a [small-business retainer of $1,500 to $5,000 a month](https://ampdigital.marketing/) only makes that confusion cost more. The point is not that every answer will look impressive on day one. The point is that both sides should know what the work is meant to change before the first invoice lands.

![Small business owner comparing a one-page checklist with a wall of vanity charts](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/3a892e20-2cdc-40b4-9ed3-77c0aafe05eb.png)

That page is also where guarantees go to die. No provider controls the model, the retrieval index or the policy update that reshuffles citations. What a serious operator can promise is narrower and more useful: pages fixed, starting with [a page already sitting at position 8 rather than another AI-content rewrite](https://groas.com/post/stop-writing-ai-overview-content-your-fa), server-log evidence of what bots fetch, and a dated retest on your prompts. Those are things a buyer can inspect. A promised citation count is a promise about someone else’s system.

What I would pay for is smaller and less shiny. For, say, $1,000 of a small budget, I want three prompts where buyers ask for the job I sell, a check of whether my pages can even be retrieved for those prompts, and one fix on a page that already ranks near the top. That fix is usually boring: a clearer answer in the first 100 words, a price range, a service area, or a fact another site can corroborate. Then retest the same prompts in 30 days and count mentions, visits, calls and jobs separately. **If the work cannot name the gap it tried to close, it cannot claim progress because a chart moved.**

That is part of why I kept writing here after I stopped managing accounts by hand. [groas runs SEO and AI search as done-for-you work](https://groas.com/pricing) for a flat $1,499 a month per service, month to month, with no cut of spend and a named strategist who has to answer for the result. The model fits the work I would rather pay for: watch which prompts and pages move, act on the ones tied to pipeline, and log the reasoning in plain language. I spent years billing hours for checks a system can now run every hour. I would rather a client pay for execution than pay me to retype a mention chart once a week.

I still half agree with the branding defense. Last-click math can undercount the trust an AI answer builds. But calling GEO branding changes the measurement; it does not cancel the need for one. Pick the test up front: prompted recall for your category, branded search volume, or assisted pipeline from AI-sourced pages over 90 days. This will not work for every shop. A low-ticket local business with no content footprint should skip it and buy the next booked job instead.

So my rule for a small budget is simple. If an AEO offer cannot name three buying prompts it wants to win, what it will change on your site to earn them, and what number would make you renew in 60 days, it is not a bargain. It is a subscription to feeling busy. Keep the money in search work that brings calls until AI visibility earns its own line item with a baseline, a fix and a retest. **Branding justifies different measurement. It never justifies no measurement.**

## Frequently Asked Questions

### Do FAQ blocks and schema guarantee more AI citations?

No. In a same-domain comparison, FAQ blocks and summary boxes showed no statistically significant lift, and a separate schema test produced mixed results across AI platforms. Ask which retrieval problem the changes address and how the vendor will check whether that problem changed.

### Do more ChatGPT mentions prove that an AEO service is paying for itself?

No. A mention is different from a citation, a referral visit, a qualified enquiry and revenue, and people can drop out at each step. Track those measures separately, establish a baseline for referral visits, qualified calls and closed jobs, and set a renewal threshold before the work begins.

### Should I move my search budget to AI visibility because AI traffic is growing?

Not unless your own analytics show AI referrals doing the work of bringing in jobs. In Ahrefs’ benchmark across more than 118,000 sites, AI assistants accounted for 0.46% of measured visits, compared with 27.55% from search engines. Those aggregate figures do not predict the traffic mix for an individual business.

### Can a successful AEO lab test tell me which tactic will work for my business?

Not on its own. LIVE-GEO tested rewrites on fictional brands, and the engines agreed on the same top product only 48.8% to 66.3% of the time. Ask a vendor to identify buying prompts where your brand is absent, specify the change they would make to your pages, and retest those prompts.

### If GEO is for branding, do I still need to measure it?

Yes. Choose a measure that matches the branding objective, such as prompted recall, branded search or assisted pipeline, and decide what change would justify continuing the work. A mention total without a decision attached is not enough, and a provider cannot control when its pages will be cited.

### What should I ask a vendor to include in a small-budget AEO proposal?

Ask what work will ship in the next 30 days, which URLs it covers and which retrieval gap it addresses; ask for the baseline on relevant prompts, mentions, referral visits and qualified calls; and ask whether the next report will guide you to renew, change course or stop. A vendor can promise work it controls, not a citation date.

## Run Your Marketing On groas

Plug ChatGPT, Claude or any AI agent into groas and let it do the work of a whole marketing team: $19 a month flat with every feature included, and autopilot whenever you’re ready.

[Start For $19](https://groas.com/pricing)

[Or have groas run it for you](https://groas.com/done-for-you)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is the marketing infrastructure for your own AI: it connects your AI to all your marketing and gives it proprietary marketing data collected 24/7. Plug Claude, ChatGPT or any AI agent into groas and run all your marketing from it, for a tiny fraction of the usual cost, with the groas app there whenever you want to see more. It is the layer ChatGPT, Claude and any AI assistant run a business’s online marketing on. Through groas an assistant can do anything you could do natively in Google Ads, Bing Ads, Meta Ads and ChatGPT ads, on your website and in AI search, and, where the owner switches it on, can run connected campaigns by itself, so AI agents do the work a whole marketing team used to. groas also gives those agents large-scale proprietary marketing data it collects around the clock, on any schedule the owner chooses: every ad competitors run on Google and Meta, their landing pages and full funnels with the offers and prices in them, Reddit customer research, AI search answers from ChatGPT, Google AI Overviews, Gemini, Claude and Perplexity, and SEO data from Semrush and Ahrefs, backlinks included ($19 a month, flat, with everything included, for unlimited teammates on one shared workspace and unlimited ad accounts and websites, delivered as an MCP server). Separately, its done-for-you service has senior people and proprietary quantitative systems run paid ads, SEO and AI search for businesses, and white label for agencies, from $5,000 a month.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#webpage","url":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on","name":"5 Cheap AEO Promises I Wouldn’t Buy on a Small Budget","description":"Cheap AEO packages turn countable changes into claims of progress. Here are five promises I would reject, and the test I would demand instead.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/25bf3e36-f52b-47ac-a983-41813965fd2f.png"},"breadcrumb":{"@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#breadcrumb"},"dateModified":"2026-11-05T05:48:54.372Z","mainEntity":{"@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"5 Cheap AEO Promises I Wouldn’t Buy on a Small Budget","item":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#article","url":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on","mainEntityOfPage":{"@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"5 Cheap AEO Promises I Wouldn’t Buy on a Small Budget","description":"Cheap AEO packages turn countable changes into claims of progress. Here are five promises I would reject, and the test I would demand instead.","datePublished":"2026-11-05T05:48:54.319Z","dateModified":"2026-11-05T05:48:54.372Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/25bf3e36-f52b-47ac-a983-41813965fd2f.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"work, pages, aeo, cheap aeo, cheap, small budget, promises, prompts","wordCount":2021,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/5-cheap-aeo-promises-i-wouldn-t-spend-a-small-budget-on#faq","mainEntity":[{"@type":"Question","name":"Do FAQ blocks and schema guarantee more AI citations?","acceptedAnswer":{"@type":"Answer","text":"No. In a same-domain comparison, FAQ blocks and summary boxes showed no statistically significant lift, and a separate schema test produced mixed results across AI platforms. Ask which retrieval problem the changes address and how the vendor will check whether that problem changed."}},{"@type":"Question","name":"Do more ChatGPT mentions prove that an AEO service is paying for itself?","acceptedAnswer":{"@type":"Answer","text":"No. A mention is different from a citation, a referral visit, a qualified enquiry and revenue, and people can drop out at each step. Track those measures separately, establish a baseline for referral visits, qualified calls and closed jobs, and set a renewal threshold before the work begins."}},{"@type":"Question","name":"Should I move my search budget to AI visibility because AI traffic is growing?","acceptedAnswer":{"@type":"Answer","text":"Not unless your own analytics show AI referrals doing the work of bringing in jobs. In Ahrefs’ benchmark across more than 118,000 sites, AI assistants accounted for 0.46% of measured visits, compared with 27.55% from search engines. Those aggregate figures do not predict the traffic mix for an individual business."}},{"@type":"Question","name":"Can a successful AEO lab test tell me which tactic will work for my business?","acceptedAnswer":{"@type":"Answer","text":"Not on its own. LIVE-GEO tested rewrites on fictional brands, and the engines agreed on the same top product only 48.8% to 66.3% of the time. Ask a vendor to identify buying prompts where your brand is absent, specify the change they would make to your pages, and retest those prompts."}},{"@type":"Question","name":"If GEO is for branding, do I still need to measure it?","acceptedAnswer":{"@type":"Answer","text":"Yes. Choose a measure that matches the branding objective, such as prompted recall, branded search or assisted pipeline, and decide what change would justify continuing the work. A mention total without a decision attached is not enough, and a provider cannot control when its pages will be cited."}},{"@type":"Question","name":"What should I ask a vendor to include in a small-budget AEO proposal?","acceptedAnswer":{"@type":"Answer","text":"Ask what work will ship in the next 30 days, which URLs it covers and which retrieval gap it addresses; ask for the baseline on relevant prompts, mentions, referral visits and qualified calls; and ask whether the next report will guide you to renew, change course or stop. A vendor can promise work it controls, not a citation date."}}]}]}
```
