One of the biggest decisions a founder makes, usually made with less rigour than picking a new hire. Here's the whole thing — who to pick, how to vet them, and the two mistakes that quietly do the most damage.
What worked for one business doesn't automatically mean it'll work for yours. Different category, different margin structure, different starting point — a glowing reference tells you the agency can do good work somewhere, not that they're the right fit for your specific account. Worth listening to references. Worth not treating them as the whole decision.
A logo wall full of recognisable names looks reassuring. It's worth a second look before it reassures you completely. Big-name clients often mean big budgets — and big budgets buy more spend, more data, and more room to test, which genuinely can produce better results. But big names can also mean an agency that's been following the same playbook everyone else in their category follows, regardless of whether it's actually working, because a cautious board or a big client's own internal politics rewards "safe" over "right." The logo tells you they won the business. It doesn't tell you which of these two stories is true.
Two patterns show up often enough, and cost enough, that they deserve more room than a paragraph each:
A 12-month contract sounds sensible when everything's going well. It's a different conversation four months in, if the relationship isn't right, mistakes keep happening, or you're spending more time managing the agency than the agency is saving you. A good agency backs its own work enough to accept a break clause — genuine confidence doesn't need a long lock-in to survive. An agency that resists one is often protecting the contract, not the results.
This isn't about expecting things to go wrong. It's about not being trapped for eight more months if they do.
Agencies will naturally want to move fast, especially if a real account restructure is on the table. Worth resisting the urge to match that pace. Set clear, specific targets before anything gets signed — not vague ones that can be reinterpreted favourably in six months. And challenge every number they show you, especially the forecast. Part 3 of this series is a real example of exactly what happens when that step gets skipped.
15 minutes. No pitch, just a look.
Book a free chat →