Check Client Logos Before Signing That GEO Contract

You Need to Verify How Client Logos Are Actually Being Used

According to a publicly shared account of suspected fraud, A stumbled across a promotional post from an AI marketing agency on LinkedIn. On the agency’s homepage, the logo of the financial group A works for was displayed as a “client case study.” A manages vendor relationships at the group, so she knows the company’s business partners inside and out — and this was a firm she’d never heard of. When she asked which affiliate the agency had worked with and whether it had permission to use the logo, instead of answering, the company blocked her.

You, dear reader, may have gotten a pitch for “GEO¹​ agency services” yourself lately — the kind of sales pitch promising that your brand will show up in ChatGPT’s or Perplexity’s answers. But why is it that right now, logo theft like this and “guaranteed exposure” claims are surfacing all at once?

GEO is a real strategy. But the moment words like “guarantee,” “domination,” or “only one spot left” get attached to it, that’s a signal to check whether you’re looking at exaggerated sales talk. Today, let’s lay out the criteria for telling the two apart.

What’s actually happening right now

There’s a firsthand account from someone in the marketing industry. It described a GEO agency that displayed major corporate logos on its homepage like a portfolio — and when asked whether the collaborations were real, gave no answer. According to the post, some of the logos were later removed. The author put it bluntly: “They used GEO, the hottest new trend, as a shield to build a plausible-looking front that only fools the uninformed.”

This isn’t an isolated case. A well-known Korean marketing community’s newsletter published a similar tip. The sales pitches follow a recurring pattern: flaunting expertise with lines like “our team is from top-tier universities and has mastered the AI logic” or “we have connections with key personnel at Naver,” while pressuring prospects into upfront payment and signing with urgency tactics like “prices are going up soon” or “there’s only one slot left.”

The most devious part is how results get reported. Some agencies reportedly type a specific brand name repeatedly into a generative AI, then open a new chat window, screenshot the resulting brand mention, and submit it as a “monthly performance report.” The problem is that generative AI answers vary enormously depending on the user’s phrasing, conversation context, and personalization settings. A single screenshot simply can’t prove sustained improvement in visibility. You need to check the actual query, account settings, model, and timing of measurement together.

Content ownership needs checking too. There are reported cases where an agency builds content on a domain it newly registers and owns itself — and if the client doesn’t complete the full 2-year contract, that domain and all its content simply disappears, or has to be bought back separately. Even after paying for the service, ownership and post-contract usage rights can vary entirely depending on what’s actually in the contract.

There’s a reason this tactic works so well specifically in GEO. In an early-stage market, “who have you worked with” — that is, references — become the currency of trust. But since GEO results are genuinely hard to verify directly, a single major-brand logo on a homepage generates far more trust than it deserves. So rather than assuming a real contractual relationship just because a logo is displayed, you need to check the actual evidence behind it.

Why Now

GEO is the effort to improve how content gets cited or how brands get mentioned in AI answers. But the existence of this demand doesn’t mean every agency’s performance promises hold up to scrutiny.

A single search box is reshaping the food chainThe ruler of an ecosystem is restructuring its own ecosysteminlevel9.com

SparkToro analyzed Similarweb’s US mobile and desktop web panel data from January to April 2026 and found that 68.01% of Google searches ended without a click.²​ This isn’t a global search figure — it covers the US only. AI answers aren’t the only factor at play; instant-answer features, search interface design, and usage habits could all be contributing. Google announced in May 2026 that AI Mode had surpassed 1 billion monthly users worldwide. That’s the backdrop for why companies now care so much about how they appear in AI search.

Naturally, this makes companies anxious. If people ask AI instead of typing into a search box, then how their brand shows up inside that AI’s answer becomes something they need to manage. The demand is real. Which is exactly why the sales pitches for GEO agencies keep multiplying too.

The problem is that GEO is still an early-stage market. SEO³​ has had more than 20 years to build up measurable metrics — search ranking, traffic, click-through rate, conversion rate. GEO, by contrast, varies by AI model, and even the same question can produce a different answer each time. This is called non-determinism⁴​, and it means you can’t guarantee identical exposure for every question and every user. You need to check exactly who the promise applies to, over what period, and under what measurement conditions.

Even in SEO, rankings shift depending on location, timing, and personalization. In GEO, you also have to account for the answers themselves changing. In both cases, what matters is repeated measurement across multiple conditions, compared against a baseline period.

None of this means performance can’t be measured at all. It means agencies and clients need to agree on which questions to test, how many times to repeat them, and how exactly to count citations, mentions, traffic, and conversions. Without clear criteria, it’s hard to tell an inflated report from an honest one.

Real GEO Means Trust, Structure, and Distributed Reach

So what does proper GEO actually look like? I actually laid this out once before, two years ago in June 2024. Back then, I emphasized exactly three things: credibility, readable structure, and distributed reach. Cite reliable statistics and sources clearly, break your writing into digestible sections with subheadings, and push the content out across communities and social media too.

Search Engine Optimization → Search Engine Results Generation Optimization?Generative search engines (Google SGE, Bing Chat, Perplexity, etc.) are spreading rapidly. These engines find multiple web pages in response to a user’s question …haebom.dev

What’s interesting is that academic findings don’t stray far from this either. In the “GEO” paper published by researchers at Princeton and Georgia Tech, content that attached source citations, included statistics, and added expert quotes improved visibility metrics in generative AI responses by as much as 40% under experimental conditions. That said, the effect varied by domain. In other words, there’s no “secret algorithm-cracking technique” hiding inside GEO. Trustworthy content and clear structure are the things worth improving. But this experiment doesn’t guarantee the same effect across every commercial model. It shares the same roots as basic SEO fundamentals—registering your site with Google Webmaster, keeping your sitemap clean, and getting mentioned by credible channels.

Create & capture demand in the AI Search era — Think with GoogleGoogle’s Official Digital Marketing Publication. See how to create and capture demand in the era …business.google.com

One more thing: the distributed-reach point I emphasized two years ago still holds. Generative AI doesn’t just crawl your own homepage—it draws broadly on Reddit, Quora, and community posts too. Getting well-sourced content discovered and cited across relevant channels can genuinely widen your chances of being picked up. But flip that around: this process takes time. And that’s exactly where it becomes clear why anyone promising to “guarantee” something as slow-moving as search optimization “in a short period” should raise a red flag.

With this baseline in mind, you can scrutinize a specific pitch much more concretely. Here are five questions I’d put to anyone offering a GEO agency service.

• What exactly are you guaranteeing? Check whether it’s exposure to all users or measured results for a specific set of queries. You also need terms for what happens if the promise isn’t met.
• Can they explain the scope of work? If they can’t specify “what, on which channels, how” and just keep repeating trendy buzzwords, that’s a red flag.
• Can the references be verified? If all you get is a logo—no point of contact, no contract, no confirmable scope of actual collaboration—that might just be logo-flexing.
• Are they rushing you into a contract? If they won’t give you time to review, citing pricing or limited slots, check the scope of work, refund terms, and cancellation conditions first.
• What proof do they offer of results? A single screenshot of a new chat isn’t enough to judge by. Look for whether they present reproducible metrics, like citation frequency or AI-driven traffic. And be sure to confirm, before signing anything, who actually owns the domain where the content is being accumulated.

Honestly, these five points are just the basics of what’s commonly called performance marketing—marketing built on exposure and data. If a marketing service dealing in exposure and data can’t answer these questions, you’re better off holding off on the contract until you get verifiable answers. To be clear, these red flags alone don’t legally prove fraud.

Oswarld’s Lens

I’ll admit I’m writing this with somewhat mixed feelings. Two years ago, when I first introduced GEO on this blog, I was genuinely excited about the possibility that even a small blog with weak domain authority could get a shot. I don’t think that optimism was wrong. But in the meantime, I’ve watched the same concept get used to dress up thin, unsubstantiated sales pitches and reel in clients.

From my experience building GTM strategies, this is a very familiar pattern. The forecast that “technology will change the world” is usually right. But the people selling you when and in what order that change will arrive almost always make it sound faster and more certain than it really is. The big shift—search moving toward AI—is real. But “so pay us upfront, right now” is an entirely different claim.

What these vendors are really exploiting, in my view, is the anxiety of the people responsible for making decisions. The market is visibly shifting, but the metrics for measuring success haven’t been settled yet—and that gap in confidence is exactly what gets mined, with a single screenshot passed off as proof of results. So the countermeasure is simple, too: don’t rush into a contract. Just check the five questions and baseline conditions I laid out earlier. The old-fashioned approach—producing trustworthy content and citing sources clearly—has now become the standard by which you can filter out the hype.

Closing

• The shift from search to AI-generated answers is real. So is the demand for GEO.
• But phrases like “we guarantee exposure,” “we’ve cracked the algorithm,” or “there’s only one spot left” are signals that you need to dig deeper into the evidence and conditions behind them.
• You don’t need to hunt for some secret trick. Just check the basics: credibility, well-structured readability, and verifiable performance metrics.

If you ever find yourself on the receiving end of a GEO agency pitch, try copying the five questions above and putting them straight to the other side. If they can’t answer the core conditions concretely, I’d suggest thinking twice before signing anything.

Have you come across any GEO or AI-marketing pitches recently that made you go, “huh, that sounds off”?

---

💬 Tell us in the comments about any agency pitch that struck you as strange. 📨 If you have a colleague weighing a GEO proposal, forward this to them.

---

References & Further Reading

Primary sources

• Pranjal Aggarwal et al., “GEO: Generative Engine Optimization”, KDD ‘24 (ACM SIGKDD), 2024. : This experiment—testing strategies like citing sources, statistics, and quotes—is the core evidence behind today’s piece. Read it alongside the experimental setup and the differences across fields.
• “Thinking of Hiring a GEO Agency? Hold On a Second!”, i-Boss Q Letter. : A well-organized rundown of real cases in Korea, including fake screenshot reports and domain lock-in tactics. Worth reading before any agency pitch meeting.
• Reporter Jang, “Suspected Fraud Cases in Generative AI Search Optimization (GEO) Agencies”, news report, 2025. : The report containing the logo-theft case that opened today’s piece.

Background

• Google, “SEO Starter Guide”. : GEO’s roots ultimately trace back to SEO fundamentals. You can find everything from webmaster registration to sitemaps here.
• “In 2026, Less than One Third of Google Searches Still Send a Click”, SparkToro, 2026. : The source of that 68% zero-click figure, and why search is shifting toward AI answers.
• Kwangseob Ahn, What Is GEO (Generative Engine Optimization)? (2024) : The piece where I first laid out GEO, two years ago. Read it alongside today’s piece to see how the same concept has been put to different uses.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

Footnotes

1. GEO (Generative Engine Optimization): a strategy for optimizing content so that generative AI systems like ChatGPT and Gemini cite or reference it when producing answers. Unlike SEO, which aims for “ranking high in search results,” GEO’s goal is “being cited in AI answers.” ↩

2. Zero-click search: a search that ends without the user clicking any link. Besides AI answers, this can happen for various reasons, including instant answers or simply ending the search early. ↩

3. SEO (Search Engine Optimization): the long-established practice of optimizing content to rank highly on the results pages of search engines like Naver and Google. GEO’s fundamentals come directly from here. ↩

4. Non-determinism: the property whereby a generative AI’s answer can differ each time, even for the same question. Because of this property, any promise to “guarantee exposure” is fundamentally hard to keep. ↩