---
title: "The AEO Sales Deck That Promises to Get Your Client Cited"
description: "A straight-faced tour of the AEO agency pitch deck: knowledge graphs, prompt dashboards, auto-publishing, edge proxies, and the citation guarantee no vendor…"
url: "https://groas.com/post/the-ai-search-visibility-deck-a-straight"
image: "https://groas.com/media/blog/1b089b5ee13adde8f9eb143fe791450ed796a465f209062831c5419209341dc4.png"
published: "2026-09-24T05:11:25.399Z"
modified: "2026-09-30T23:20:52.459Z"
---

[Google Ads Best Practices](https://groas.com/category/google-ads-best-practices) · September 24, 2026 · 9 min read

# The AEO Sales Deck That Promises to Get Your Client Cited

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

Updated September 30, 2026

![Cover image for: The AEO Sales Deck That Promises to Get Your Client Cited](https://groas.com/media/blog/1b089b5ee13adde8f9eb143fe791450ed796a465f209062831c5419209341dc4.png)

In this article

1. [Slide 1: We Manage Your Knowledge Graph](#slide-1-we-manage-your-knowledge-graph)
2. [Slide 2: The Dashboard Tracks Prompts, Not Causes](#slide-2-the-dashboard-tracks-prompts-not-causes)
3. [Slide 3: Publishing Without Friction, Including Editorial Friction](#slide-3-publishing-without-friction-including-editorial-friction)
4. [Slide 4: Go Live in Twelve Minutes, Explain It to Engineering Later](#slide-4-go-live-in-twelve-minutes-explain-it-to-engineering-later)
5. [Slide 5: Unlimited, Within the Limit](#slide-5-unlimited-within-the-limit)
6. [The Question Reserved for the Last Minute](#the-question-reserved-for-the-last-minute)
7. [Three Questions to Ask Before Signing](#three-questions-to-ask-before-signing)

Please turn to the title slide. Three of your largest retainers forwarded you the same kind of screenshot on Sunday: ChatGPT answered “best mid-market inventory management software” with five competitors and no mention of their company. Each email asked some version of _Why aren’t we here?_

It is now Monday morning. Your search lead is muttering about vector weights, your account managers have a quarterly business review coming up, and nobody wants to say, “We don’t know why that answer chose those names.” Fortunately, you have booked this demonstration. Our twenty-minute presentation will show you how to sell Answer Engine Optimization without troubling your team with the mechanics of an answer engine.

Agencies searching for the **best AEO software for agencies** often think they need a way to improve client visibility. For the purposes of this deck, they need something much easier to procure: a convincing explanation of why the improvement is already underway.

## Slide 1: We Manage Your Knowledge Graph

Project twelve glowing spheres connected by neon filaments against a charcoal background. Label three _Entity Resolution_, _Semantic Clustering_, and _Multi-Modal Grounding_. When the client leans forward, deliver the line in a calm voice: **“Our proprietary platform directly manages and optimizes your enterprise knowledge graph across all frontier LLMs.”**

Do not pause. Someone may ask whose knowledge graph you mean: Google’s, Wikidata, the client’s internal schema, or the JSON-LD file your platform added to a staging page. Click to the next animation. The goal is not to explain data ingestion. The goal is to make whatever happened sound too expensive to cancel.

Behind the glowing spheres sits a five-line Schema.org script containing the company’s address and a few basic attributes. Structured markup can help a crawler understand a page. It does not give your vendor a private control panel for what ChatGPT or Gemini recommends when a buyer asks an unbranded question. But the diagram has done its work: nobody in the room wants to be the person who asks whether an RDF triple is the same thing as a database table.

**Presenter note:** If you cannot name the data you changed or where it appears, keep the animation moving.

![A presenter points at a corporate slide filled with glowing, labeled network nodes.](https://groas.com/media/blog/30bf4d3a08c974ad8c36180fe817b7ab395f14ccb2975f745fc3ec0b2ffcde12.png)

## Slide 2: The Dashboard Tracks Prompts, Not Causes

Now reveal the feature every **AEO tool for agencies** knows how to sell: a prompt tracker. Fill the screen with violet and teal lines measuring _Share of Model_ or an _AI Visibility Index_. The platform sends twenty selected prompts to an LLM repeatedly, checks whether the client’s name or domain appears in the answers, and plots the percentage. It is a tidy picture of what those checks returned.

On Tuesday the line says 42%. On Thursday it says 18%, and an amber alert appears. Did the model change? Did it retrieve different pages? Did a competitor get picked up in a discussion thread? The chart cannot tell you. **Observing an answer is not the same as knowing what caused it**, let alone changing the conditions that produced it.

This distinction is best handled in the presentation with another chart. The client asks what you did this month about AI Overviews; you export fifty pages of graphs. Monitoring makes for a reassuring PDF. Diagnosing a missing mention may require looking at crawlability, indexation, the pages an engine can reach, and the sources it appears to trust. That work is less photogenic, so it belongs outside the deck.

**Presenter note:** When the line falls, call it an insight. Do not call it a fix.

## Slide 3: Publishing Without Friction, Including Editorial Friction

A prospect will eventually ask which tools help a business get featured in Google AI Overviews rather than merely monitor mentions. This is your cue to show an integration diagram: WordPress, Webflow, Shopify, and HubSpot, all connected by clean arrows to your platform. The promise is irresistible. The software detects gaps, drafts forty articles overnight, and publishes them directly to the client’s site. No waiting for a senior strategist. No waiting for anyone, really.

Here is the workflow you need not narrate. The tool takes prompts where competitors appeared, turns each into a template request for a comprehensive guide, and sends the results through a webhook. Forty-eight hours later, the client’s carefully maintained site carries hundreds of articles with titles such as _A Definitive Guide to Industrial Supply Chains in 2026_. They share sentence structures, contain fabricated customer anecdotes, and conclude that supply chain management matters in modern commerce.

![Linocut-style illustration of a robotic conveyor belt depositing stacks of identical papers onto an executive desk.](https://groas.com/media/blog/0ff8fa61c1ab5cc47cc0e71b6ca359801baaee37d193cc3d899b68cfde1303a6.png)

For an agency trying to grow without hiring, the arithmetic is lovely: a $3,500 monthly retainer for “algorithmic entity expansion” against a $450 software bill. The less lovely part arrives when the Chief Marketing Officer finds an auto-published comparison recommending a direct competitor and misstating the client’s refund policy. You now have an 8:00 AM call about why the company website says the company no longer serves enterprise accounts.

**Presenter note:** “Publishes directly to your CMS” describes a capability. It does not describe an editorial standard. Leave that distinction off the slide.

## Slide 4: Go Live in Twelve Minutes, Explain It to Engineering Later

The fourth slide is for the account director who has waited six months for a client’s engineering team to install one tracking pixel. Display a three-step flowchart: point a DNS CNAME at the vendor’s edge network, identify requests from bots such as GPTBot or ClaudeBot, and serve those requests pre-rendered semantic markdown. Underneath, print the clincher: **“Zero developer resources required. Go live in twelve minutes without touching your production codebase.”**

The slide makes a routing change sound like a settings change. An unvetted proxy in front of pages that matter to the business creates a new dependency. If its worker fails, those pages can fail with it. If the arrangement affects checkout traffic, a bad deployment becomes more than an AEO problem. Your client’s VP of Engineering may have opinions about learning this from an alert rather than a design review.

![Vintage-style plumbing schematic showing an oversized bypass valve labeled PROXY leaking onto a server rack.](https://groas.com/media/blog/34d69c31348daf3f66998b55ceb19296cf7b7a1cc7e350c0fd45ec444be76b62.png)

Even when the worker runs, ask what it serves. Pre-rendering content so a crawler can read what a visitor can read is one proposition. Feeding a bot an inflated, keyword-stuffed version that differs from the human-facing page is another. Stale prices, ghost URLs, and conflicting product claims do not become authoritative because they arrived from an edge server. The deck calls this _AI-ready delivery_. Engineering may use shorter words.

**Presenter note:** Before anyone touches DNS, establish which requests the proxy handles, who can roll it back, and whether the bot-facing content matches the site.

## Slide 5: Unlimited, Within the Limit

Finish the product tour with the pricing slide. “Agency Unlimited: $499/month for unlimited clients and uncapped tracking.” Add green checkmarks for white-label reports, custom domains, and weekly digests. With twelve retainers to serve, the agency owner can already see the margin. Fifteen accounts into a $499 bill looks better still.

Now open the terms. In this deck’s version of _unlimited_, the workspace has a shared pool of 250 prompt executions a month. Track ten queries on each of three surfaces for three clients, once a day, and you use 90 executions per run. **The third daily run exceeds the monthly pool.** The remaining clients have not made it into the calculation. At that point the platform either charges for additional checks or refreshes less often. Our fictional pricing table puts overages at $0.15 a check, discreetly placed beneath the large word _unlimited_.

As we discussed in our look at [flat-rate multi-domain AEO pricing](https://groas.com/post/flat-rate-aeo-pricing-per-domain), the useful number is not just the headline fee. It is what the plan lets you do across the domains and prompts you actually need to monitor.

![An endless receipt tape covered in tiny asterisks and surcharges spills from a brass cash register.](https://groas.com/media/blog/ae81d085bfd695f4cc11414bc287978946f9f3638386f697d79581ed027b18ac.png)

**Presenter note:** If a buyer asks how many checks their client roster requires, offer a custom quote. The slide has completed its assignment.

## The Question Reserved for the Last Minute

The sales rep now opens the floor. API webhooks are welcome. SSO integration is welcome. Agency partner discounts are especially welcome. The question not printed in the speaker notes is: **Can this software guarantee that my client will be cited in an AI answer?**

No. A dashboard can sample answers. A publishing tool can put pages online. A proxy can change how certain requests receive those pages. None of those actions gives a vendor control over every answer an engine returns. If the pitch has quietly moved from “we can improve the conditions for visibility” to “we can guarantee placement,” you are hearing the old first-page-ranking promise in new vocabulary.

There is still work worth doing. An engine needs to reach and understand the relevant pages. A client’s claims need to be accurate, consistent, and useful enough to appear beyond its own website. If core specifications hide behind a page a crawler cannot read, another violet line chart will not expose them. If outside sources do not support the story the client tells about itself, a private “knowledge graph” will not supply that support by decree.

Put that mechanism on slide one, and the guaranteed-citation claim has nowhere to sit.

## Three Questions to Ask Before Signing

Keep the deck open long enough to ask three operational questions. They are less glamorous than _Multi-Modal Grounding_, which is why they help.

1. **Does the platform make technical fixes, or only report problems?** If it flags a comparison page that a crawler cannot parse but leaves the rendering work to your client’s unavailable developers, you have bought a chore generator. Find out who makes the change and who checks the result.
2. **What happens after a client’s mentions fall?** An alert tells you where to look. Ask how the vendor investigates the drop and what, if anything, it can change. “The dashboard notifies your team” is an answer about notification.
3. **What does the contract say about outcomes?** If the rep admits that third-party answer engines vary, good. Ask how that admission fits with the deterministic ranking promised twenty minutes earlier. You are allowed to have them explain both sentences in the same room.

Monitoring is comparatively easy to package: send prompts, record outputs, draw a line, add the agency logo. Execution is harder to fit into a screenshot. Someone has to identify what can be changed, make the change, and check whether the business got anything beyond another report. A $499 dashboard may help an agency observe the problem. It is not, by itself, a way to solve it.

That is why the alternative to this deck is not a prettier deck. An agency can put people on the technical and content work, or it can use an autonomous engine such as [groas](https://groas.com/) to execute search and content optimization continuously under defined guardrails, including work on [earned search](https://groas.com/earned-search) visibility. Either way, the useful promise concerns the work performed and how it is measured. It is not a reserved seat in ChatGPT’s next answer.

Next Sunday, another client may send another screenshot with another list that excludes them. You can reply with a PDF showing that _Share of Model_ was up on Wednesday, then wait for the inevitable follow-up about what semantic clustering did for their pipeline. Or you can ask what the engine could reach, what the client actually published, which claims others support, and what your team can change.

Nobody selling AEO software can guarantee a citation, and the vendors worth buying from will tell you that before they show you slide two.

## Frequently Asked Questions

### Does managing a knowledge graph let a vendor control what ChatGPT recommends?

No. Structured markup such as a Schema.org script can help a crawler understand a page, but it does not give a vendor a private control panel over what ChatGPT or Gemini recommends when a buyer asks an unbranded question.

### If my AI visibility score drops, does the dashboard tell me why?

No. A prompt tracker repeatedly sends selected prompts to an LLM and plots whether the client's name appears, so it shows what the checks returned but not the cause. Observing an answer is not the same as knowing what caused it, and diagnosing a drop may require looking at crawlability, indexation, reachable pages, and trusted sources.

### What are the risks of a tool that auto-publishes AI-written articles to my CMS?

Auto-published content can carry fabricated anecdotes, duplicate sentence structures, and factual errors. The article describes an auto-published comparison that recommended a direct competitor and misstated the client's refund policy. Publishing directly to a CMS is a capability, not an editorial standard.

### Is serving pre-rendered content to AI bots through an edge proxy risky?

It can be. An unvetted proxy in front of important pages creates a new dependency, and if its worker fails, those pages can fail with it. There is also a difference between pre-rendering content so a crawler reads what a visitor reads, and feeding a bot an inflated, keyword-stuffed version that differs from the human-facing page.

### What does an 'unlimited' AEO plan usually actually limit?

In the example given, Agency Unlimited at $499 per month includes a shared pool of 250 prompt executions per month. Tracking ten queries on each of three surfaces for three clients, once a day, uses 90 executions per run, so a third daily run exceeds the pool, after which the platform charges overages ($0.15 a check in the example) or refreshes less often.

### Can any AEO software guarantee my client will be cited in AI answers?

No. A dashboard can sample answers, a publishing tool can put pages online, and a proxy can change how requests receive pages, but none of these gives a vendor control over every answer an engine returns. A guaranteed-citation claim is the old first-page-ranking promise in new vocabulary.

### What should I ask an AEO vendor before signing a contract?

Three operational questions are worth asking: whether the platform makes technical fixes or only reports problems, what the vendor does after a client's mentions fall, and what the contract says about outcomes. If a rep admits that third-party answer engines vary, ask how that fits with any deterministic ranking promised earlier.

## Pay For Results, Not For Hours

Businesses buy the outcome, agencies resell it, and groas answers for it either way.

[See If You Qualify](https://groas.typeform.com/to/xC1bQNUT)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","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/the-ai-search-visibility-deck-a-straight#webpage","url":"https://groas.com/post/the-ai-search-visibility-deck-a-straight","name":"The AEO Sales Deck That Promises to Get Your Client Cited","description":"A straight-faced tour of the AEO agency pitch deck: knowledge graphs, prompt dashboards, auto-publishing, edge proxies, and the citation guarantee no vendor…","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://groas.com/media/blog/1b089b5ee13adde8f9eb143fe791450ed796a465f209062831c5419209341dc4.png"},"breadcrumb":{"@id":"https://groas.com/post/the-ai-search-visibility-deck-a-straight#breadcrumb"},"dateModified":"2026-09-30T23:20:52.459Z","mainEntity":{"@id":"https://groas.com/post/the-ai-search-visibility-deck-a-straight#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/the-ai-search-visibility-deck-a-straight#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":"Google Ads Best Practices","item":"https://groas.com/category/google-ads-best-practices"},{"@type":"ListItem","position":4,"name":"The AEO Sales Deck That Promises to Get Your Client Cited","item":"https://groas.com/post/the-ai-search-visibility-deck-a-straight"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/the-ai-search-visibility-deck-a-straight#article","url":"https://groas.com/post/the-ai-search-visibility-deck-a-straight","mainEntityOfPage":{"@id":"https://groas.com/post/the-ai-search-visibility-deck-a-straight#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"The AEO Sales Deck That Promises to Get Your Client Cited","description":"A straight-faced tour of the AEO agency pitch deck: knowledge graphs, prompt dashboards, auto-publishing, edge proxies, and the citation guarantee no vendor can make.","datePublished":"2026-09-24T05:11:25.399Z","dateModified":"2026-09-30T23:20:52.459Z","image":{"@type":"ImageObject","url":"https://groas.com/media/blog/1b089b5ee13adde8f9eb143fe791450ed796a465f209062831c5419209341dc4.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"articleSection":"Google Ads Best Practices","keywords":"Google Ads, Google Ads Best Practices","wordCount":1892,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"Ex Goldman Sachs and Ex 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/the-ai-search-visibility-deck-a-straight#faq","mainEntity":[{"@type":"Question","name":"Does managing a knowledge graph let a vendor control what ChatGPT recommends?","acceptedAnswer":{"@type":"Answer","text":"No. Structured markup such as a Schema.org script can help a crawler understand a page, but it does not give a vendor a private control panel over what ChatGPT or Gemini recommends when a buyer asks an unbranded question."}},{"@type":"Question","name":"If my AI visibility score drops, does the dashboard tell me why?","acceptedAnswer":{"@type":"Answer","text":"No. A prompt tracker repeatedly sends selected prompts to an LLM and plots whether the client's name appears, so it shows what the checks returned but not the cause. Observing an answer is not the same as knowing what caused it, and diagnosing a drop may require looking at crawlability, indexation, reachable pages, and trusted sources."}},{"@type":"Question","name":"What are the risks of a tool that auto-publishes AI-written articles to my CMS?","acceptedAnswer":{"@type":"Answer","text":"Auto-published content can carry fabricated anecdotes, duplicate sentence structures, and factual errors. The article describes an auto-published comparison that recommended a direct competitor and misstated the client's refund policy. Publishing directly to a CMS is a capability, not an editorial standard."}},{"@type":"Question","name":"Is serving pre-rendered content to AI bots through an edge proxy risky?","acceptedAnswer":{"@type":"Answer","text":"It can be. An unvetted proxy in front of important pages creates a new dependency, and if its worker fails, those pages can fail with it. There is also a difference between pre-rendering content so a crawler reads what a visitor reads, and feeding a bot an inflated, keyword-stuffed version that differs from the human-facing page."}},{"@type":"Question","name":"What does an 'unlimited' AEO plan usually actually limit?","acceptedAnswer":{"@type":"Answer","text":"In the example given, Agency Unlimited at $499 per month includes a shared pool of 250 prompt executions per month. Tracking ten queries on each of three surfaces for three clients, once a day, uses 90 executions per run, so a third daily run exceeds the pool, after which the platform charges overages ($0.15 a check in the example) or refreshes less often."}},{"@type":"Question","name":"Can any AEO software guarantee my client will be cited in AI answers?","acceptedAnswer":{"@type":"Answer","text":"No. A dashboard can sample answers, a publishing tool can put pages online, and a proxy can change how requests receive pages, but none of these gives a vendor control over every answer an engine returns. A guaranteed-citation claim is the old first-page-ranking promise in new vocabulary."}},{"@type":"Question","name":"What should I ask an AEO vendor before signing a contract?","acceptedAnswer":{"@type":"Answer","text":"Three operational questions are worth asking: whether the platform makes technical fixes or only reports problems, what the vendor does after a client's mentions fall, and what the contract says about outcomes. If a rep admits that third-party answer engines vary, ask how that fits with any deterministic ranking promised earlier."}}]}]}
```