What TransUnion's AI study reveals about brand citations

TransUnion's research points to a simple conclusion: most large brands do not have an AI strategy problem first. They have a measurement problem. Marketers are ready to spend more on AI, but many still cannot see clearly how AI platforms shape performance, where the data gaps sit, or how to connect AI exposure to business outcomes. That matters for SEO because AI search visibility lives inside the same blind spot. If your team pitches GEO as a new line item with its own jargon, you risk underselling it. If you pitch it as the search-facing part of a broader measurement gap, executives are more likely to understand why it matters. citeturn1view0turn0view0

GEO is the practice of improving how a brand gets surfaced, described, and cited inside AI-generated answers. The study itself was not built as a GEO report. But the pressure it exposes maps directly to AI Search: bigger budgets, low visibility, weak attribution, and fragmented data. That is exactly why this research deserves attention far beyond paid media and marketing ops. citeturn1view0turn0view0

• What does the study actually tell us?
• Why do walled gardens matter for AI citations?
• Why is confidence rising faster than readiness?
• BotRank's Take
• How should SEO teams change the GEO pitch?
• What should teams measure before asking for more budget?
• FAQ

What does the study actually tell us?

It tells us that AI ambition is growing faster than AI readiness. In the TransUnion-commissioned survey of 100 senior marketing and technology leaders at major U.S. brands, 89% said investment in AI-enabled marketing will increase over the next 12 to 24 months, and 64% said they are confident they will hit their AI goals. But only 42% rated their organization's people readiness as high, only 36% said the same for data and process readiness, and fewer than half, 48%, said they have enough visibility into platform-level AI to optimize with confidence. citeturn1view0

That gap matters because confidence without instrumentation usually produces noisy reporting. A team may know that AI is helping with content production, bidding, personalization, or workflow speed. It may not know which platform contributed what, whether the effect is durable, or whether the gain is actually incremental. In plain English, the board sees enthusiasm while the operators still see fog. This is the same reason many brands are now trying to measure their AI visibility instead of treating AI answers like a novelty. citeturn1view0turn0view0

The numbers are also useful because they prevent a lazy reading of the market. This is not a story about marketers rejecting AI. It is the opposite. Investment appetite is strong. The problem is that readiness is uneven and observability is weak. For enterprise teams, that is usually a data and governance issue before it is a tooling issue. citeturn1view0

• Budget is moving: 89% expect AI investment to grow.
• Confidence is real: 64% expect to reach AI goals.
• Readiness is lagging: only 42% feel ready on people, and 36% on data and process.
• Visibility is thin: only 48% say they can see enough at the platform level to optimize confidently.

A practical example helps. Imagine a retail brand running paid social, connected TV, marketplace media, classic search, and AI-assisted creative production. Leadership sees faster campaign output and rising spend efficiency, so the instinct is to double down. But if each platform reports success in its own way, the team still cannot answer the harder question: which signals made the brand more discoverable, persuasive, or memorable across the whole journey? That is the measurement gap the study exposes. citeturn1view0

Why do walled gardens matter so much for AI citations?

Because AI search creates another opaque layer in a journey that is already hard to measure. TransUnion found that 69% of respondents see blind spots inside walled gardens as a limit on evaluating AI effectiveness, and 70% say cross-channel blind spots make it difficult to understand AI's impact across the customer journey. In the related analysis of those findings, Matt Spiegel argued that AI search should be treated as a natural extension of that same measurement problem, not as a separate universe. citeturn1view0turn0view0

A walled garden is a platform environment where critical data, attribution logic, and user-level visibility stay mostly inside the platform. That definition matters for GEO. When someone discovers a brand through an AI answer, the brand often cannot fully see how the answer was assembled, which source carried the most weight, or what competing brands were considered and discarded. You are measuring from the outside, just as you do with retail media, paid social, or closed recommendation systems. citeturn0view0

This is where many GEO conversations go wrong. Teams often ask, "Which AI citation tool should we buy?" before asking, "What exactly can we observe, compare, and validate across models?" That sounds subtle, but it changes the budget conversation. A citation tracker can tell you that your brand appeared. It cannot, on its own, solve disconnected source data, fragmented analytics, or channel-level opacity.

Take a B2B software company that sees its name appear in ChatGPT for a few commercial prompts but almost never in Gemini or Perplexity. Without source-level analysis, the team might assume it has a content deficit. In reality, the issue might be that one model relies more on pages where the brand is well represented, while another relies on discussion forums, third-party reviews, or stale documentation. The symptom looks like poor visibility. The underlying cause is measurement and source fragmentation. That is why the idea that AI visibility is an operations problem is more than a slogan. It is an accurate diagnosis of how brands get misread by AI systems.

The study did not test prompt sets across LLMs directly. That is an important limit. It is stronger as a lens on enterprise readiness than as a direct map of AI Search mechanics. But the overlap is still hard to ignore: if marketers already struggle to measure AI across closed platforms and disconnected channels, they will struggle even more when the answer layer itself becomes another black box. That is a fair inference from the data, not a stretch. citeturn1view0turn0view0

Why is confidence rising faster than readiness?

The short answer is that investment is easier than infrastructure. Buying AI tools, launching pilots, and showcasing time savings happen quickly. Building trusted data, identity resolution, cross-team governance, and measurement standards does not. TransUnion's findings point in exactly that direction, with readiness scores for people, process, and data trailing far behind investment intent and goal confidence. citeturn1view0

The study also shows why executives can feel satisfied too early. Sixty-five percent of respondents said they primarily measure AI success through time and cost savings, while less than half use more advanced methods such as marketing mix modeling, multi-touch attribution, and incrementality testing. Efficiency metrics are not useless. They are often the fastest proof that a tool is doing something. But they are a weak proxy for whether AI is changing discovery, brand preference, or commercial outcomes. citeturn1view0

For AI search, this distinction is critical. A brand can be cited more often and still fail to create value. Maybe the citations appear only on low-intent prompts. Maybe the brand gets named but framed poorly. Maybe the cited pages do not actually support conversion because they are outdated, off-message, or technically weak. That is why visibility alone is not enough. You need to know what kind of visibility you earned, where it appeared, and whether it influenced a business result.

Here is the nuance many teams miss: this enterprise framing works especially well for large brands with layered martech stacks and fragmented reporting. It is slightly less decisive for a small company with a simple funnel and a narrow buying journey. Even then, though, the principle still holds. If you cannot explain where your AI mentions come from and whether they matter, you do not yet have a mature GEO program.

BotRank's Take

Our view is simple: the study matters because it shifts GEO out of the "interesting SEO experiment" category and into the "measurable business visibility" category. That is where it belongs. If AI search is one more opaque decision surface, brands need a way to monitor how they appear across models, which prompts trigger inclusion, which competitors show up instead, and which sources shape the answer.

This is where AI Visibility becomes useful in a very practical way. It lets teams build reusable prompts, run them across multiple LLMs, track visibility over time, and inspect the entities, sentiment, and keywords attached to their brand. Paired with Source Analysis, teams can also review the pages cited behind AI answers and check whether those pages genuinely mention the brand or simply influence the response indirectly. That matters because a brand mention without context can look like progress while hiding a representation problem.

The point is not to create another dashboard for its own sake. The point is to make AI visibility observable enough that content, SEO, and brand teams can act on it together.

How should SEO teams change the way they pitch GEO?

They should stop pitching GEO as a niche citation-chasing project and start pitching it as the search-facing layer of AI measurement. That recommendation follows directly from the study's logic and from the accompanying analysis: the real obstacle is not lack of excitement about AI, but lack of visibility into how AI performs across platforms and channels. citeturn1view0turn0view0

A weak pitch sounds like this: "We need a new GEO tool because AI search is getting popular." A stronger pitch sounds like this: "We already invest in AI-enabled marketing, but we still have limited visibility into how AI platforms shape discovery and brand recommendation. GEO measurement closes the search-specific part of that gap." The second version connects to a problem leadership already recognizes.

That reframing changes what comes next. Instead of treating success as "more citations," teams can define success in layers:

• Coverage: Are we present on the prompts that matter?
• Representation: Are we described accurately and competitively?
• Attribution: Which sources and pages influence those answers?
• Readiness: Do the cited pages deserve to be cited?
• Impact: Can we connect visibility changes to qualified traffic, leads, or assisted conversions?

For example, a financial services brand might discover that its comparison pages are frequently cited, but its compliance pages and trust content rarely surface. A raw citation count would miss the strategic issue. A better GEO pitch would show that the brand is visible for category-level prompts but underrepresented on trust-heavy prompts where users decide whether to engage.

There is also a sequencing lesson in the TransUnion findings. Before buying another dashboard, audit whether your data foundation is connected enough to support one. If the team cannot align prompts, pages, campaign periods, and business outcomes, the new tool will report activity but struggle to prove impact. That does not mean you delay forever. It means you pair tracking with operating discipline, clear ownership, and a prioritized backlog. In practice, that is where GEO recommendations and technical audits become more useful than another disconnected report. citeturn1view0turn0view0

What should teams measure before they ask for more budget?

They should measure the parts of AI visibility that move a decision from anecdote to evidence. The goal is not to build a perfect model on day one. The goal is to prove that AI search can be tracked consistently enough to deserve budget and operational attention.

A good starting scorecard includes five layers:

• Prompt coverage: the share of important prompts where your brand appears.
• Model spread: whether visibility is concentrated in one assistant or consistent across several.
• Source quality: which domains and pages influence answers, and whether they reflect your current positioning.
• Sentiment and framing: whether AI describes your brand as trusted, expensive, risky, innovative, niche, or interchangeable.
• Outcome linkage: whether changes in AI visibility line up with traffic quality, conversions, or downstream sales signals.

If you want a practical benchmark list, BotRank has already broken down the GEO metrics that matter in AI search. That kind of framework helps teams move away from vanity reporting and toward decision-grade reporting.

One concrete workflow is to start with ten commercial prompts, ten comparison prompts, and ten reputation prompts. Track them across multiple LLMs every month. Then review not just whether you appeared, but which pages were cited, how competitors were framed, and whether your strongest conversion pages were present in the answer chain. That process will not give you perfect attribution. It will give you pattern recognition, which is enough to prioritize action.

The next step is operational. If a cited page is outdated, improve it. If the wrong third-party pages keep shaping the answer, respond with better evidence and clearer entity signals. If your brand is inconsistently described, align language across product, help, PR, and partner surfaces. This is where GEO becomes execution, not commentary.

And this is the main strategic lesson from the TransUnion research: more AI spend without better measurement creates the illusion of progress. Better visibility without better interpretation creates the illusion of control. Mature teams need both. citeturn1view0turn0view0

FAQ

Is this study really about GEO?

Not directly. It is a survey about AI-enabled marketing readiness, but its measurement gaps map closely to the same observability problems SEO and GEO teams face in AI search. citeturn1view0turn0view0

Does more AI investment automatically lead to better brand citations?

No. The survey shows investment appetite is strong, but platform visibility and readiness remain limited, which means spend can rise faster than measurement quality. citeturn1view0

Why are walled gardens such a problem for AI visibility?

Because they limit what marketers can see about how performance is created. AI search adds another opaque layer, so brands often have to infer influence from the outside instead of observing it directly. citeturn1view0turn0view0

What is the smartest first step for an SEO team?

Start with measurement, not mythology. Build a repeatable prompt set, track visibility across models, inspect the sources behind answers, and tie those patterns to real business pages before asking for a larger GEO budget.

The takeaway is straightforward: do not pitch AI search visibility like a shiny new channel. Pitch it like what it is becoming for serious brands: a measurable layer of market visibility that sits on top of data quality, source credibility, and cross-team execution. If you want to make that layer visible instead of guessable, BotRank is the natural next step.