A marketing director picks a new creative agency roughly once every two or three years. That’s it. Compare that to procurement teams buying software or logistics providers, who run structured evaluations constantly, and you start to see the problem: most people making a genuinely high-stakes decision are doing it for only the second or third time in their career, and without algorithms.
So they fall back on what they know. A recommendation from a colleague. An agency they remember from an awards show. Whoever pitched well last time, regardless of whether that agency was actually right for this brief. None of this is negligence. It’s just what happens when you’re choosing blind in a market too large to actually study.
And it is large. The global agency landscape is fragmented across hundreds of thousands of agencies worldwide, spanning creative, media, digital, PR and dozens of specialist categories. No single marketer, no matter how experienced, has meaningfully surveyed that landscape. They’ve surveyed the twelve agencies they’ve personally encountered.
Key Takeaways
- Marketing directors often rely on personal experience to choose agencies, leading to uninformed decisions.
- AI aids procurement by providing data-driven shortlists instead of relying solely on familiarity.
- Tools like TrinityP3’s Agency Search platform identify agency partners based on specific capabilities and experience.
- Combining data-driven insights with human expertise improves decision-making in vendor selection.
- This shift towards data-led sourcing applies to various industries, not just marketing, addressing infrequent, high-stakes selections.
Table of contents
Data Is Quietly Replacing Gut Feel with Algorithms
This is where AI algorithms have started doing something genuinely useful in procurement and vendor selection, agency hiring included. Not the flashy stuff. The unglamorous work of pattern-matching capability against requirement, at a scale no human shortlist ever could.
The mechanics are fairly simple. Feed a system category data, performance history and capability profiles across a large agency pool, and it can surface a shortlist based on actual fit rather than familiarity. An agency that’s quietly excellent in FMCG packaging but has never won an award gets found. An agency riding reputation from one good campaign five years ago gets weighed against what it’s actually delivered since.
This isn’t unique to marketing. Procurement teams across industries are running the same experiment: replace instinct-led sourcing with data-led sourcing, and see what changes. In agency selection specifically, the potential is significant: fewer wasted pitches and fewer appointments that fall apart eighteen months in because nobody checked whether the agency had actually done this kind of work before.
What This Looks Like in Practice for Algorithms
Marketing consultancies have started building tools around exactly this problem. TrinityP3, for instance, offers an AI-powered Agency Search platform that helps marketers identify potential agency partners based on capabilities, category experience, and other fit criteria, rather than relying on awards or “who you know.”
That narrows the field. What happens next still tends to involve people.
A shortlist, however well-built, is still just a list. Someone has to sit across the table, read the room, and work out whether this particular team will actually deliver when a campaign goes sideways at 4pm on a Friday. That’s not a data problem. It never was.
Why the Combination Has Real Potential
Data alone gets you a well-matched shortlist. Expertise alone gets you sound judgement, applied to whatever options happened to land in front of it. Bring the two together, and each has the chance to sharpen the other: the shortlist gives judgement something worth acting on, and judgement turns a good shortlist into the right decision.
That’s the real shift in vendor selection right now, and it isn’t limited to marketing. Procurement teams evaluating law firms, engineering contractors and software implementation partners face the same structural problem: infrequent, high-stakes decisions made against a market too large for any one person to study. A consultant’s pattern recognition, built over years of watching deals succeed and fail, is only as useful as the pool of options it gets applied to. Widen that pool with real data, and the judgement itself gets more valuable, not less.
Where This Goes Next
None of this is confined to advertising agencies. Wherever the buyer only makes the decision once every few years and the supplier market runs into the thousands, the old model of picking from memory instead of algorithms starts to look less like tradition and more like a gap waiting to be closed.
What’s changing isn’t that expertise matters less. It’s that expertise is finally being pointed at the right options before the pitch process even starts. For an industry that has run on reputation and referral for the best part of a century, that’s not a small shift. It’s just a quiet one.
