Notes · The work · July 23, 2026

AI should be making your agency deliver more, not just earn more.

A couple of years ago, it took me a team of five and a handful of agencies to run a brand's growth. Paid, creative, email, the lot. Now I run the same thing on my own.

Same output. Often better.

That's not a brag, and it's not because I got five times smarter overnight. It's the tools. The doing got cheaper. The judgment, the bit that actually moves the number, is still the hard part, and that hasn't changed. But the doing? A fraction of what it was.

Here's the thing nobody's really saying out loud: if that's true for me, it's true for your agency too.

Full disclosure.

Before I go further, I've got a horse in this race. I do fractional work now, so it's squarely in my interest to tell you that you might not need your agency. Fine. Take what follows with that in mind. But I'd say it even if I had nothing to sell, because I've watched it play out from the inside, on both sides of the invoice.

I'm not here to tell you to fire anyone. Good agencies are worth every penny, and some are evolving faster than the brands they work for. The question isn't whether agencies are good or bad. It's simpler, and a bit more uncomfortable than that.

Are you getting the upside?

Where the saving goes.

AI has made everyone faster. Agencies included. The campaign that used to take a team a week takes a good one an afternoon. That efficiency is real, and it's sitting inside every agency in the world right now. The only question is what they do with it.

The good ones take that speed and hand it back to you. More tests run, more creative in market, faster iteration, sharper targeting, all for the same fee. You feel it in the results.

The comfortable ones pocket it. Same retainer, same monthly deliverables, a lot less graft behind the scenes, and a tidier margin at their end. The senior person you're actually paying for barely touches the account, because a junior with the right tools now clears the old bar in half the time. You'd never know. The invoice looks exactly like it did in 2023.

Both are using AI. Only one is sharing it.

The test.

So here's how you tell which one you've got. It isn't clever. It's just honest.

Look at your output and your results over the last year or two. If they've genuinely improved, stay. Your agency is evolving with you and they're not taking you for a ride. That's rarer than you'd think, and worth holding onto.

If they're flat, same results, same spend, same story as eighteen months ago, then something's off. Either they haven't leaned into the tools, or they've leaned in and kept the gains for themselves. Neither is a reason to stay.

The trap isn't the agency. It's the habit.

Most brands don't stay with an underperforming agency because they've weighed it up. They stay because switching is a faff. The account manager is lovely. There's a contract, a rhythm, a shared history, and nobody fancies the awkward conversation. Habit dressed up as loyalty.

I've written before about the order founders get burned in: agency, then marketer, then fractional, each hire fixing the last gap and opening a new one. This is the quieter version of the same story. You don't have to get burned to be overpaying. You just have to stop asking whether the thing you signed two years ago is still the best version available.

What actually changed.

The reason this matters now, and didn't a few years ago, is the alternative.

Leaving an agency used to mean replacing it. Three hires, a new stack, months of ramp. For most brands that math never worked, so they stayed. Now, more often than not, leaving can mean one person. One operator, AI doing the heavy lifting, eyes on the whole thing, instead of the work split across four vendors who each own a slice and none of whom own the number.

That isn't right for everyone, and I won't pretend it is. Some brands genuinely need the scale and the specialist depth an agency brings. One person, however well equipped, can't be a full creative studio and a media team and a lifecycle shop all at once. If that's you, and they're delivering, keep them and don't look back.

But a lot of brands are paying agency prices for work one good operator could now do on their own, and they've never once stopped to check. Not because the answer's daunting. Just habit.

So ask the question.

Look at what you're paying. Look at what's actually improved. Be honest about the gap between the two.

AI should be making your agency deliver more, not just earn more.

If it isn't, that's worth knowing. Whatever you decide to do with the answer.

← Back to Notes