The Agency's Original Sin

If you’ve worked in or with agencies for any length of time, you know the shape of this without me describing it. Sales sells a dream. Delivery inherits the reality. And the distance between the two is where careers get ground down, clients get disappointed, and good people burn out swearing they’ll never do it again.

I used to think this was a problem of bad actors: a particular slick salesperson, a particular over-eager shop. I don’t think that anymore. I’ve seen it in too many places, worn too many different logos, to keep believing it’s a personnel issue. It’s structural. It’s built into the incentives. I’ve started calling it the agency’s original sin, because it’s right there at the foundation, and almost nobody designs it out.

”Just say yes, we’ll figure it out later”

Here’s the mechanism. In the room, when the prospect asks “can you do X?”, the pressure is enormous to say yes. Yes closes the deal. Yes hits the number. And the quiet, industry-wide philosophy is: just agree, we’ll figure out how to actually do it later. Later is someone else’s problem. Later is the delivery team’s problem.

So the contract gets signed on a promise that nobody has checked against reality, and the delivery team opens the file to discover they’ve been committed to something ambitious, underscoped, and due sooner than is sane. They didn’t make the promise. They just have to keep it, or eat the fallback when it can’t be kept: the overtime, the scope fights, the awkward call where someone has to walk the client back from what they were sold.

This isn’t one villain. It’s a system that pays for the promise and not the delivery, and systems that do that will produce over-promising forever, no matter how nice the individual people are.

Follow the commission

The reason it’s so sticky comes down to how the money moves. In a lot of shops, the salesperson gets paid (commission booked, deal done) at or near signing. Which means the person who made the promise has been fully compensated before anyone knows whether the promise was any good.

Think about what that does to the feedback loop. It severs it. The most important information in the whole business (did we actually deliver what we sold, and at what true cost?) never makes it back to the person whose behavior it should shape. The salesperson has already moved on to the next deal, made whole, structurally insulated from the consequences of the last one. You’ve built a machine that rewards the promise and quarantines the person who made it from the fallout. Of course it over-promises. You paid it to.

“You’ve built a machine that rewards the promise and quarantines the person who made it from the fallout.”

If you want to fix over-promising, you have to reconnect that loop. Tie some real portion of the sales reward to what actually gets delivered: to the project landing on scope, to the client’s genuine satisfaction, to the true margin after delivery ate whatever it ate. The moment the person selling has skin in whether the thing can be built, the promises get more honest overnight. Not because anyone got more virtuous. Because the incentive finally points the right way.

The scapegoat at the end of the line

I want to tell one generalized story about who pays for this, because it changed how I see it.

A talented creative was brought onto a project to produce written work. The client was unhappy with the output and started describing it, dismissively, as low-quality machine-generated filler: “slop.” The reputation stuck to the creative. It looked, from the outside, like a talent problem: this person just isn’t good enough.

It was nothing of the sort. When you traced it back, the real failure was upstream and invisible. The client had never provided the technical grounding, the point of view, the raw substance the writing was supposed to be built on. You cannot produce sharp, specific, expert content out of nothing. The creative was set up to fail by a broken process (no inputs, no brief, no source of truth) and then handed the blame when the process produced exactly what a broken process produces.

That’s the original sin playing out at the far end of the line. Somebody sold “we’ll handle your content,” nobody scoped what that actually required from the client, the gap got papered over, and when it finally showed, the person with the least power in the chain wore it. The talent was never the issue. The absence of a real process was.

What a healthy shop actually looks like

If the sin is paying for the promise and shielding the promiser, the redemption is a shop where the whole organization has skin in whether the promise was any good. I’ve come to believe the healthiest agencies share a few unglamorous traits, and none of them are about talent: they’re about incentives and nerve.

The incentives are aligned front to back: some real portion of what the salesperson earns depends on the project actually landing well (on scope, on a genuinely satisfied client, on the true margin after delivery took its bite). When the person selling has a stake in whether the thing can be built, the whole conversation in the sales room changes, quietly and permanently, from “how do I close this” to “what can we actually deliver and defend.” You don’t have to police honesty when you’ve paid for it.

The nerve shows up as the willingness to say what you can’t do. It sounds backwards, but telling a prospect “that part isn’t a fit for us, here’s who’s better at it” is one of the strongest trust signals in the business, because everyone else in the room is nodding yes to everything. Honest scoping (including honest no) reads as confidence, and it spares you the doomed engagement where you win the logo and lose the year. The counterintuitive growth lever for a service business isn’t taking more work; it’s turning down the wrong work, so your delivery stays excellent and your reputation compounds into the kind that sends clients to you pre-sold. A shop that overpromises grows fast and churns furiously. A shop whose word is worth exactly what it costs grows slower and never stops, because in a business built on trust, being the one people can believe is the entire moat.

Requirements are not bureaucracy

The antidote to all of this is unglamorous, and that’s exactly why it gets skipped: write things down before you promise them.

The projects I’ve watched drift the longest were the ones with no real requirements: a big engagement wandering for a year and a half, no clear owner, no definition of done, everyone busy and nothing shipping. It felt, in the moment, like agility. It was just fog. Requirements, scoping, a written definition of success: these read as bureaucratic drag to people who want to move fast, but they’re the opposite. They’re what let you move fast without the crash at the end, because they force the honest conversation about what’s actually being promised to happen before the ink dries instead of after.

I don’t think the agency model is doomed. I think it’s carrying an original sin it refuses to name. Pay for the promise and shield the promiser, and you’ll over-promise forever. Reconnect the loop (reward delivery, scope honestly, put the truth in writing) and you get something increasingly rare in this business: an agency whose word is worth what it costs.