Good agency evaluation has never really been about who’s got the glossiest case study film, the nicest awards shelf or the sharpest pitch deck. The best agency selection criteria go deeper than that. They tell you which agency will still be useful when the brief gets murky, the timeline turns idiotic and the relationship has to survive contact with reality.
There’s a particular kind of agency meeting that almost everybody in marketing has sat through at some point. The deck is immaculate. The case studies are glossy without being vulgar about it. The client list is stacked with names designed to make the room feel reassured. The people presenting are clever, composed and just self-aware enough to seem human. Every answer lands. Every slide says, in its own polished way, relax, professionals are here.
And yet a small part of the brain keeps muttering in the background. Fine. But what are these people actually going to be like when the brief is undercooked, the sign-off process is a swamp, the timeline is laughable and somebody senior barges in halfway through asking for something “braver” that won’t upset anyone?
That, really, is the whole game.
Because agency evaluation in 2026 isn’t supposed to be about who performs best in a room for ninety minutes. It isn’t about who’s got the slickest reel or the most polished story about transformation, disruption or cultural relevance. It’s about who’ll still be useful when reality arrives in all its usual forms: mixed signals, conflicting opinions, delayed feedback, budget nerves, legal caveats, internal politics and deadlines that suggest active contempt for the laws of time.
That’s why brands need to evaluate agencies beyond awards and case studies. Not because awards are meaningless, and not because case studies don’t matter. They do matter. Good work should count. Recognition should count. Strong thinking should count. But they’re only fragments of the picture. They tell a brand what an agency has done at its best. They don’t tell it enough about what it’ll be like to work with them repeatedly, under pressure, inside actual organisational constraints.
And that’s where a lot of brands still get seduced by the wrong things. They buy the pitch version of an agency. That version is often extremely attractive. It’s also highly managed. A pitch is a performance. That doesn’t make it fake. It does mean it’s partial. It’s the agency at its most coherent, attentive, polished and deliberately impressive. An appointment, by contrast, is a decision to build a working relationship. Those are two very different things, and confusing them is how brands end up six months later trapped in a cycle of unclear scopes, mounting irritations and meetings where everyone says “alignment” as though it’s a sacred concept rather than a sign that something has quietly gone wrong.
Agency Selection Criteria That Actually Matters
Grey Global
Most agency selection processes still overvalue what’s easiest to admire. Awards are easy to admire. Famous client logos are easy to admire. Elegant decks are easy to admire. Senior people with excellent posture and strong opinions about strategy are especially easy to admire.
What’s harder to admire, because it doesn’t sparkle in the same way, is usually what determines whether the relationship will actually work. How good is the agency at listening when the brief isn’t clear? How disciplined is the delivery model? How stable is the team? How sane are the commercials once the job’s live? How does the agency behave when challenged? Does it clarify difficult problems or merely restyle them? Can it make a messy organisation more effective, or will it simply add another layer of process, language and expense?
That’s the less glamorous stuff, but it’s the stuff that decides whether the appointment becomes useful or exhausting.
The most sensible agency selection criteria tend to sit across four broad areas: capability, behaviour, operating model and proof.
Capability is the obvious one, but even here brands need to be more suspicious than they often are. Great portfolio work doesn’t exist in a vacuum. It was made under a very particular set of conditions, with a particular client, a particular budget, a particular level of trust and a particular timeline. The question isn’t only “is this good?” It’s “is this kind of good repeatable for us?”
That’s the uncomfortable truth lurking beneath a lot of agency glamour. Some brilliant work reflects a genuine system of quality. Some of it reflects an unusually perfect set of circumstances that may never exist again.
Category experience can help, but it’s often overrated in a slightly lazy way. Structural relevance tends to matter more. Has the agency solved this kind of business problem before? Has it handled this level of complexity, scrutiny, ambiguity or stakeholder drag? That’s usually more revealing than whether it once worked for a vaguely similar brand.
Behaviour matters just as much. In fact, it’s often the thing that decides whether the relationship stays healthy once the pitch glow wears off. How does the agency listen? Can it challenge without sounding smug? Can it explain without showing off? Does it make complicated things clearer or just more expensive-looking? What happens when someone disagrees with it? Does it stay calm and useful, or does it start radiating the energy of a person who believes they’re the only adult left in the room?
These things matter more than most brands admit. Plenty of agency relationships fail not because the work is terrible, but because the behaviour around the work becomes exhausting. Technically strong people can still be impossible to partner with if every question feels like an insult and every feedback round becomes a miniature constitutional crisis.
Then there’s the operating model, which is the bit many brands ignore until the contract is signed and the damage is already under way. A good agency should be able to explain, clearly and without interpretive dance, how the work will run. How is scope set? What happens when scope changes? How’s reporting handled? What’s the approval rhythm? Who’s actually on the team? What kind of senior oversight is real rather than decorative? How are decisions escalated? What does the agency need from the client in order to make the model work properly?
A lot of bad agency relationships are really just bad operating models wearing human faces. Everyone’s broadly competent. Everyone’s even quite pleasant. But the structure is wrong. The pricing model creates constant tension. The approval process is glacial. The client expects endless flexibility without cost. The agency expects decisiveness from an organisation that can’t decide what biscuits to order for a meeting without a steering group. Eventually all that structural friction starts getting interpreted as personality.
And finally there’s proof. Case studies are useful, but they’re curated objects. They’re built to persuade. Brands should want more than output and atmosphere. They should want some evidence of impact and a credible account of how the agency thinks about effectiveness. Useful agencies can usually talk about outcomes in grown-up terms. They can connect creative and strategic decisions to business effects. They can describe what success looks like before the work starts. They can say what they’d do if the results weren’t good enough. They can discuss short-term wins and long-term brand value without pretending those are enemies.
That’s what brands should be buying. Not just output. Judgment.
How Brands Approach Agency Evaluation in Reality
HUSH
It’s also worth being honest about something the industry politely dances around. Many agency evaluations are messy because the client organisation is messy.
Marketing wants ambition. Procurement wants comparability. Finance wants certainty. Legal wants risk reduced to particles. Senior leadership wants something “fresh” that also looks comfortingly familiar. Half the criteria are spoken aloud and the other half are left floating around the process like unacknowledged emotional weather.
Under those conditions, agency selection starts becoming less about choosing the best partner and more about surviving internal contradiction without anybody powerful feeling exposed.
That’s where hidden criteria do enormous damage. If the real brief is “find us the safest option we can still describe as bold”, the agencies will feel it. If the real brief is “save money without making it look like that’s the point”, they’ll feel that too. If the organisation wants innovation that doesn’t change anything meaningful, the process will reward agencies that are good at packaging novelty in stakeholder-safe language.
Once that happens, the evaluation is no longer really about agency quality. It’s about political fitness.
And to be fair, agencies know this. They’ve seen it before. Which is why so many pitches end up responding to the emotional reality of the room rather than the formal logic of the brief.
Pitches themselves don’t help. They aren’t useless, but they are deeply unnatural. They reward confidence, story control, fluency and room-reading. Those are all useful traits. But they’re still only a slice of what a brand’s actually buying. A pitch tells you how well an agency can perform under observation. It doesn’t tell you enough about how it delivers month after month when the work gets fiddly, the enthusiasm fades a bit and the stakeholders start behaving like stakeholders.
That’s one reason industry bodies have spent years trying to improve pitching practices and long-term relationship design. The IPA’s Partnering for Growth report still frames successful relationships around Shared Foundation, Organisation, Understanding and Learning, which is a far more useful lens than simply asking who “won the room.”
Smarter brands keep shortlists disciplined, ask practical questions early and use formats that reveal more of the working dynamic. They know the pitch room is evidence, not truth.
What Makes a Strong Agency Partnership?
TDC PR
A strong agency partnership is rarely the one that seems most dazzling in week one. It’s usually the one that becomes more valuable as time goes on.
That happens when there’s a clear shared foundation. Both sides know what the relationship is for. The goals are clear. The commercial structure is realistic. The way of working suits the actual organisation rather than some fantasy version of it. Roles are understood. Accountability is visible. Feedback becomes precise instead of political. Learning is built into the rhythm.
Again, this is why the IPA’s S.O.U.L. model still holds up. It sets out four building blocks for sustainable client-agency relationships: Shared Foundation, Organisation, Understanding and Learning. The IPA’s current report still describes these as the characteristics and contexts that create long-term, mutually sustainable relationships.
Shared Foundation matters because a relationship without a clear purpose quickly becomes a supplier arrangement with better branding. If both sides don’t know what success looks like, the agency can’t really be strategic. It can only keep producing activity and hope it gets mistaken for progress.
Organisation matters because structure is what keeps pressure from turning personal. Good relationships often feel “easy” not because everyone’s magically compatible, but because the roles, scopes, review rhythms and decisions are sensibly organised.
Understanding matters because trust isn’t built out of vague goodwill. It’s built out of communication that stays honest under pressure. Agencies and clients need to be able to challenge one another without the whole thing turning brittle.
Learning matters because the best partnerships don’t just deliver. They improve. They absorb feedback. They review performance. They get smarter. They develop a clearer sense of what works in that particular relationship instead of repeating the same avoidable mistakes until everybody starts blaming “the market” or “the pace”.
Good partnerships also have a quality that’s easy to feel and surprisingly hard to fake. They reduce friction. Not all difficulty, because that would be childish. But they reduce unnecessary friction. Meetings become clearer. Problems get named properly. Decisions move faster. Disagreement feels productive rather than theatrical. The work improves because the relationship can actually absorb pressure without starting to crack.
That’s what brands should be looking for when they talk about agency fit. Not whether the room had a nice vibe. Whether the relationship has a realistic chance of becoming robust.
Why Agency Relationships Break Down
Yolk Creative London Limited
Most agency relationships don’t collapse in one dramatic moment. They erode. They get slowly sanded down by confusion, weak habits and badly designed expectations.
Sometimes the brief is poor from the start, which means the agency is trying to solve a problem the client hasn’t properly defined. Sometimes the selection criteria were never genuinely agreed, so every round of feedback becomes a referendum on taste. Sometimes the shortlist was too big, the pitch too theatrical and the final decision too dependent on performance rather than fit.
Sometimes procurement and marketing are pulling in different directions. One side is trying to create commercial efficiency. The other is trying to build strategic value. The agency ends up pinned between them, expected to deliver transformative outcomes through a model built mainly to minimise spend.
And then there are the newer breakdown drivers, which aren’t really that new anymore. Transparency issues. Contractual ambiguity. Data handling questions. AI governance. If the agency is using AI tools while handling client data or campaign assets, governance can’t be treated as a side issue. The ICO’s guidance still says data protection law applies to AI systems that process personal data, and the wider accountability guidance updated in February 2026 emphasises demonstrating how privacy is respected and protected.
On the commercial side, ISBA’s Media Services Framework 2025 explicitly says the updated framework takes transparency to the next level, addresses agency non-compliance in proprietary media and refines the generative AI elements so they’re easier for advertisers to implement. Even if a brand isn’t appointing a media agency, the broader lesson is obvious enough: clarity and transparency are no longer optional signs of virtue. They’re part of what competence looks like now.
Most of the time, then, agency relationships break down because the selection process didn’t properly examine the things most likely to create trouble later. The warning signs were there. They just weren’t the glamorous ones.
Questions Brands Should Ask Before Choosing an Agency
Lemonade Illustration Agency
By the time a brand is seriously considering appointment, the questions should be less about performance and more about reveal.
What problem does the agency think it’s solving? Where does it disagree with the brief? What assumptions is it making? What would it need to know before committing to a route?
Who’s actually going to do the work? Not the charming senior people who materialise for the pitch and then disappear back into the clouds, but the real delivery team. How stable are they? What happens if key people leave? How much senior attention is genuine?
How will the work actually run? How is scope handled? How are approvals structured? How does the agency prefer to work with internal stakeholders? What does it need from procurement, legal, brand and product teams? How does it stop the process becoming a swamp?
How will success be defined? What gets measured? What happens when performance is mixed or weak? How is learning captured? How does the agency balance immediate delivery against longer-term brand effects?
And then, because it’s 2026 and brands really haven’t got much excuse for pretending otherwise, how does the agency handle data, transparency and AI? The ICO still presents its AI guidance as best practice for data protection-compliant AIand as an interpretation of how data protection law applies to AI systems that process personal data.
That means brands should feel perfectly comfortable asking what AI tools are in use, what the policy is on client data, what safeguards exist, what needs approval, what’s contractually clear and how risks are assessed. If the agency treats those questions as fussy or obstructive, that’s useful information. So is the opposite.
None of that is box-ticking. It’s how you find out what you’re actually buying.
A Smarter Way to Evaluate Creative Agencies
Something Big
If there’s one principle worth holding onto, it’s this: run agency evaluation as though you’re buying a working relationship, not admiring a presentation.
Start by deciding whether a pitch is even necessary. Sometimes the real issue isn’t the agency. It’s the brief, the scope, the governance or the client’s own internal dysfunction. Replacing the agency can feel satisfyingly decisive while solving absolutely none of that.
If a search is justified, write criteria that reflect business outcomes and delivery reality. Keep the shortlist disciplined. Use workshops where useful. Probe the operating model. Interrogate proof properly. Treat transparency and governance as core criteria. Build the relationship architecture from day one rather than assuming goodwill and chemistry will improvise one later.
Most of all, stop pretending the best agency is some abstract ideal waiting to be discovered by the cleverest room. The best agency is the one most capable of doing useful, effective, transparent, well-governed work inside your actual organisational conditions, with your actual pressures, at a pace and price you can live with, while staying honest enough to remain valuable once the honeymoon ends.
That may be a less romantic way to choose.
It’s also far more likely to work.
After the Pitch is Over
Brands rarely fail at agency evaluation because they didn’t stare hard enough at the case studies. They fail because they didn’t properly assess the things that predict performance after the pitch is over: briefing discipline, decision clarity, team stability, commercial logic, trust, learning culture, transparency and governance.
Those are the parts of agency selection criteria that don’t make for thrilling theatre, but they’re usually the parts that decide whether a relationship compounds value or slowly curdles into a cycle of rebriefing and polite frustration. And in a market where current industry guidance still puts real weight on sustainable partnership design, transparency and responsible AI use, brands have even less excuse for choosing on surface alone.
