Why LLMs can never replace good marketing agency intermediaries in the pitch process

Some brands are turning to artificial intelligence  (LLMs) to help identify their next marketing agency, but are they discovering the best partners for their business – or simply the agencies that are best at being discovered?

Over the past 18 months, we’ve noticed an interesting shift in the agency search and selection landscape. Some brands appear to be taking the decision to manage agency pitches themselves, rather than appointing an independent intermediary to guide the process.

This seems particularly prevalent amongst mid-sized businesses and the national subsidiaries of regional or global brands. With marketing budgets under pressure, and procurement teams being asked to demonstrate savings, the cost of engaging a specialist pitch consultant can seem difficult to justify internally.

We understand the temptation. After all, how hard can it be to find a good marketing agency?  And now there’s an apparently simple solution: ask ChatGPT, Claude, Gemini or another Large Language Model (LLM) to identify the best agencies for your brief.  Within seconds, you can have a shortlist of agencies, complete with descriptions of their capabilities, apparent sector expertise, notable clients and even reasons why they might be a good fit for your business.  It’s fast, inexpensive and remarkably convincing.

But is it actually identifying the right agencies?

As an independent marketing consultancy with more than 20 years’ experience helping brands find, evaluate and appoint agency partners, we believe there’s a fundamental difference between generating a list of agencies and identifying the right agency for a particular business challenge.

And it’s a difference that marketing leaders and procurement specialists should understand before entrusting an important commercial decision to AI.

We’re not suggesting that LLMs have no role in agency search.  AI can be an extremely useful research tool, helping brands explore the agency landscape, identify unfamiliar businesses and quickly summarise publicly available information.

But using an LLM to generate an agency shortlist introduces several risks that are easily overlooked.

The usual suspects: the risk of homogeneous recommendations

Ask an LLM to recommend the best creative, media, digital or integrated marketing agencies and there’s a reasonable chance you’ll receive a list dominated by familiar names.

That’s not necessarily because those agencies are the best suited to your brief.

It’s because established, frequently discussed agencies often have substantial digital footprints: extensive websites, award entries, published case studies, media coverage, industry rankings and thought leadership.

These provide an abundance of information for AI systems to draw upon.  The result can be a self-reinforcing cycle in which agencies with the greatest visibility are repeatedly recommended, while smaller independent agencies, specialists and newly formed businesses struggle to feature.

Of course, well-known agencies can be excellent partners. But being well-known is not the same as being well-suited.  The risk is that brands end up evaluating the agencies that AI knows most about, rather than those best equipped to solve their business challenges.

And if every brand starts using similar AI tools to build its consideration list, we could see an increasingly homogeneous approach to agency selection — potentially at the expense of innovation, competition and fresh thinking.

Visibility is not the same as capability

Following on from that theme there’s another complication: agencies are becoming increasingly sophisticated at influencing how they appear in AI-generated recommendations.

Just as search engine optimisation (SEO) helped businesses improve their visibility on Google, generative engine optimisation (GEO) is emerging as a discipline focused on increasing visibility within AI-generated answers.

But visibility doesn’t necessarily demonstrate that an agency has the right talent, operating model, strategic capability or experience for a particular client, e.g. agencies that invest more time in client delivery than self-promotion, operate under strict client confidentiality arrangements or have only recently been established.

This raises an uncomfortable question for brands, are you selecting an agency because it is particularly good at marketing your business, or because it is particularly good at marketing itself to an algorithm?

An experienced intermediary brings a different perspective: knowledge of agencies beyond their public-facing credentials, informed by market intelligence, direct conversations, industry relationships and practical experience.

That can uncover highly relevant agencies that would never have appeared on an AI-generated shortlist.

The human factors that AI cannot reliably assess

Successful agency relationships are about much more than technical capability.  Cultural compatibility, chemistry between teams, leadership quality, creative originality, strategic curiosity and the ability to challenge a client’s thinking can all have a profound impact on the success of a partnership.

These qualities are difficult to capture in a website, case study or agency credentials presentation.  They’re even harder for an LLM to evaluate reliably.

An experienced intermediary can ask the right questions of the right people, understands from experience the true nature of an agency’s contribution, understand how its teams operate and is able to assess the likelihood of a productive working relationship.  Finding an agency with the right credentials is one thing.  Finding an agency your business can succeed with is quite another.

Yesterday’s information, today’s recommendation

The agency market is constantly evolving.  Leadership teams change.  Senior talent moves.  Agencies merge, acquire specialist capabilities, restructure their businesses, gain new clients and lose existing ones.  An agency that was an excellent fit two years ago may look very different today.

Yet an LLM can combine information from multiple sources of different ages without clearly explaining how current or reliable each piece of evidence is.  This creates a risk that brands make decisions based on outdated capabilities, historical client relationships or expertise that is no longer present within the business.

The reverse is equally important.  A newly formed agency, or an established business that has recently invested in new capabilities, may be significantly more relevant than its historical online profile suggests.

Experienced agency search specialists maintain an active understanding of the marketplace. They know which businesses are changing, where talent is moving, which capabilities are emerging and which agencies warrant closer investigation.

In a dynamic market, that up-to-date intelligence can make a significant difference to the quality of a shortlist.

The danger of false confidence

Perhaps the greatest risk of relying on LLMs is the confidence with which they present their recommendations.  AI-generated responses can be exceptionally persuasive. They organise information neatly, provide plausible explanations and often appear to offer an objective assessment of the options.  But a compelling answer is not necessarily a well-evidenced one.

LLMs can make factual errors, overstate an agency’s experience, conflate different businesses or draw conclusions from incomplete information.

More fundamentally, an AI-generated shortlist is not the product of a comprehensive assessment of the agency market against a carefully developed set of client requirements.  It’s a response generated from the information available to the model, influenced by the wording of the prompt, its underlying systems and the sources it can access.  Different LLMs may therefore produce very different recommendations to essentially the same question.  So which list should a marketing leader trust?

The false economy of managing an agency pitch without specialist support

Of course, the argument for using AI often sits within a broader decision: whether to appoint an independent pitch consultant at all.  When budgets are constrained, it can seem sensible to avoid external consultancy fees and manage the agency search internally.  But the cost of a pitch consultant should be considered alongside the much greater potential cost of making the wrong agency appointment.

Selecting an agency is not simply a purchasing exercise. It is a strategic decision that can affect marketing effectiveness, operational efficiency, brand performance and commercial outcomes for years.

For marketing procurement professionals, that combination of market knowledge, structured evaluation and commercial rigour provided by a skilled intermediary should be viewed as a means of managing risk and improving value, rather than simply another cost – and in our experience investment in our fees can often be amortised through savings and efficiencies identified through the process.

So, should AI have a role in agency selection?

Used intelligently, LLMs can help brands explore the agency marketplace, understand different service models, identify potential specialists and formulate questions for further investigation.  They can be a valuable starting point for research.

 At Observatory International, we believe the future of agency search and selection lies in combining the efficiency of new technologies with the depth of market knowledge, independence and human judgement that successful agency appointments demand.

Because the real objective of an agency review isn’t to produce a shortlist as quickly or cheaply as possible.  It’s to find the partner most capable of helping your business succeed.  And that’s a decision worth getting right.

Our proven agency search and selection process is thorough and robust – take a look here to find out more details or contact us.