How Zebora mapped the AI presence of one of advertising's most complex industry organisations
A six-week AI brand audit for the 4As, building a bespoke 542-prompt framework to measure how ChatGPT represents an organisation spanning ~800 competitive landscapes, and where its visibility gaps sit against its growth strategy.
4As wanted to understand how AI perceived one of the US advertising industry's most complex membership organisations, as well as whether that perception supported its ambition to reach more smaller, independent agencies who may not have the 4As on their radar.
Zebora spent three weeks understanding the 4As business and translating what it provides into 56 independent reporting dimensions, commercially weighted around its priorities and measured through 542 bespoke prompts.
The headline results looked strong at first. The 4As ranked #1, ahead of its primary competitor set, with LLMs clearly recognising it as a trusted industry authority. But deeper analysis showed something the overall numbers had hidden: ChatGPT struggled to articulate why 4As was different, while visibility was particularly weak among the smaller independent agencies it wanted to reach.
The work gave 4As a full 360° view of its AI presence, a prioritised diagnosis of what to do and why, plus a shared framework that stakeholders across the organisation could work from together.
Client context: why the 4As measured its AI presence now
The 4As (The American Association of Advertising Agencies) represents and supports advertising agencies across the US, providing services spanning industry insight, talent, training, best-practice guidance, advocacy, community and networking.
A new CEO and a changing market
Following a leadership change under new CEO Justin Thomas-Copeland, 4As was looking at how the organisation needed to evolve for the next generation of agencies.
Generative AI was changing how people discovered information, evaluated organisations and made decisions. As an organisation responsible for helping US advertising agencies navigate change in marketing, 4As recognised it needed to understand the implications early, starting with its own brand.
The commercial priority
There was also a specific commercial priority. 4As wanted to increase its relevance among smaller, independent agencies and independent agency groups that might not historically have considered the organisation an obvious partner.
The measurement gap
4As understood SEO, content and digital measurement, but it wasn't directly measuring its presence in AI. Its unusually broad proposition also meant a conventional set of generic brand prompts was unlikely to represent the organisation accurately.
Zebora was commissioned to conduct a six-week AI brand audit and strategic diagnosis across November and December 2025, initially focused on ChatGPT.
Why is the 4As so hard to measure in AI answers?
4As isn't a conventional single-category business. An agency looking for legal or policy guidance creates one competitive landscape (law firms, for example). Someone looking for talent development creates another (recruiters, jobs boards). Industry benchmarking, new-business support, training, events and networking each introduce different alternatives again.
The eventual analysis found around 800 different competitive organisations appearing alongside 4As across the full body of prompts.
A conventional GEO approach based on a small set of generic prompts could tell 4As whether it was being mentioned. It couldn't represent the different needs of different agencies, where 4As could add value for each, the relative importance of those needs to the organisation, or the very different competitive landscapes surrounding them.
There was also understandable scepticism internally about whether AI visibility measurement could represent an organisation this complex at all. So for Zebora, the first challenge wasn't running prompts. It was working out what should be measured in the first place.
How Zebora went deeper: building a framework for a complex organisation
Zebora spent the first three weeks working closely with 4As to build a prompt framework capable of representing the organisation as a whole.
We started with the supply side: understanding what 4As actually provides across its services, audiences and areas of expertise. We then mapped this against demand-side intent, looking at what advertising and marketing agencies were actually looking for and the language they naturally used to look for it. This let us translate the internal structure of 4As into a taxonomy built around real agency demand, rather than simply turning the organisation's own terminology into prompts.
The framework covered six areas: Master Category, Services, Features, USPs, Use Cases and Customer Segments. Across these, we defined 56 independent reporting dimensions that best represented the business, each measured using around ten different demand-side prompts. That produced a final taxonomy of 542 bespoke prompts, categorised, weighted and designed to mirror how agencies actually search.
Not every dimension was treated equally. Working with 4As, we applied a commercial weighting system so the overall measurement reflected the organisation's strategic priorities. Community & Networking, for example, was weighted 2.5× more heavily than Advocacy & Policy Guidance, so a visibility gap in a strategically important area counted for more than an equivalent gap somewhere lower priority.
Because every dimension could be analysed independently, we could move past “how visible is 4As” and start asking where it was visible, for what, among whom, against which competitors and where that mattered most.
What Zebora uncovered: an authority AI couldn't differentiate
At first glance, 4As was performing well. ChatGPT recognised it as an established, authoritative organisation within advertising, ranked it #1 competitively ahead of ANA, AMA and IAB, with a 74/100 sentiment score. But overall visibility sat at only 49/100.
Where the differentiation broke down
The deeper analysis explained the gap. 4As didn't have an authority problem so much as a differentiation problem. ChatGPT understood broadly what 4As did and regarded it positively, but struggled to give a clear, distinctive answer to “why 4As.” Its breadth was an asset, but it was also fragmenting how clearly its identity came through: the audit found 4As lacked a clear canonical positioning statement, both at master-brand level and for individual use cases such as networking events versus policy advice, with that differentiation often buried across a broad and complex content estate spanning owned and third-party sources.
Performance also varied depending on what was being measured. Visibility was stronger around category and service-level queries, but fell off across use cases and customer segments.
The gap with independent agencies
One of the most important findings was something 4As hadn't previously known. Despite its ambition to increase relevance among smaller and independent agencies, ChatGPT visibility fell to just 25/100 among small independent agencies and 30/100 among independent agency groups. It was noticeably stronger for larger agencies and networks. Zebora had found a measurable disconnect between 4As' growth strategy and how the organisation was actually being represented in AI.
The analysis also exposed gaps across strategically relevant use cases including agency sourcing, new-business development and fee-setting, alongside weaker representation in areas such as advertising-law support and talent development. Some of the most valuable signals were also hidden behind member logins; messaging wasn't always explicit enough about audience, use case and differentiation on the pages LLMs could actually access; and strategically important pages such as the membership sign-up page were consistently missing from AI answer citations.
What the diagnosis changed
The diagnosis shifted the question from “how do we increase our AI visibility” to something more specific: how does 4As make its distinctive value clearer to AI, particularly for the audiences and agency needs that matter most to the organisation.
Action & outcomes
Beyond the numbers, the audit gave 4As something it didn't have before: a single view of how the whole organisation shows up in AI.
4As is a genuinely fragmented business. Events and networking, talent, industry insight and advocacy have historically operated with a fair amount of independence from one another, each with its own priorities and its own view of the member. Building the 542-prompt taxonomy meant mapping all of it (services, audiences, use cases and strategic priorities) into one framework, which turned out to be one of the most valuable outcomes of the project in its own right. For the first time, 4As could see how these different parts of the business were each being understood and represented by AI, as well as how they related to one another.
4As started this project with essentially no visibility into how it appeared in AI at all, so the baseline below isn't just a set of scores. It's the first time the organisation has been able to see itself through this lens.
The 2025 baseline
How to read these scores
Competitive position: 4As' rank against its measured set of rivals. AI visibility: how often it appears across the commercially weighted query set. Sentiment: how positively it is described. AI Brand Health: a composite of how well AI understands the brand, covering its authority, trust and clarity in the models. Citations from owned sources: the share of cited sources that are 4As' own pages. Content readiness: how well 4As' web content supports its discoverability and representation in LLMs. Technical readiness: how well the underlying site technically supports LLM discovery.
Following the audit, Zebora ran sessions with stakeholders across 4As covering how LLMs work and form perceptions of organisations, what the growth of LLMs means for marketing and brand discovery more broadly, what the findings meant specifically for 4As and where AI perception aligned with the organisation's strategy and where it didn't. Feedback was strongly positive, particularly around how the process brought visibility to parts of the organisation that had rarely been looked at together before.
The prioritised action plan
The diagnosis translated into a clear, prioritised set of actions rather than a generic list of GEO fixes:
- 1Establish clearer canonical “why 4As” positioning, at both master-brand and individual use-case level.
- 2Make audience, use-case and value messaging more explicit across owned content.
- 3Strengthen relevance and representation for smaller and independent agencies specifically.
- 4Surface more evidence of 4As' value outside member-only areas.
- 5Prioritise categories to work on based on strategic importance, current visibility, competitive intensity and ease of improvement.
- 6Strengthen key owned signals, particularly around the membership experience.
What happened next
This engagement was scoped as a standalone audit and strategic diagnosis, not a programme of implementation, so there's no before-and-after visibility uplift to report yet. What 4As does have is something it didn't have going in: a clear, evidenced picture of how AI currently represents a business that's genuinely difficult to represent, plus a prioritised plan for what to do about it.
For an organisation as fragmented as 4As, that diagnostic clarity is itself a significant outcome. Different parts of the business could see, in one shared framework, how they were each showing up in AI and how that connected to the rest of the organisation, something that hadn't existed before this project. The 2025 baseline now gives 4As a fixed point to work from, whether that's implementing the recommendations internally or returning to Zebora for the next phase.
Findings are based on Zebora's measurement of 542 commercially weighted prompts on ChatGPT across November and December 2025, spanning 56 independent reporting dimensions and roughly 800 competitor entities. This is a single baseline audit on one AI assistant, a diagnostic snapshot rather than a before-and-after comparison.
See what AI really understands about your business
Your organisation is more complex than a handful of generic prompts. Your AI measurement should be too.