Your agency isn't a partner. It's a vendor with a pitch deck.

I've been on both sides of this. Running agencies. Hiring them. Watching founders pay $40k for a website that looks like it cost $40k and does none of the things the business needed. The pattern is depressingly consistent.

What founders think they're buying

A founder hires an agency because they want a co-owner of the outcome. 'Make us look like a real company.' 'Get our pipeline moving.' 'Figure out the brand.' What they want is someone who will stay accountable to whether the business is better in six months.

What agencies optimise for

Agencies optimise for scope. Scope is contract-enforceable. Outcomes are not. The moment an outcome becomes the measure of success, the agency's margin depends on things they can't control — the founder's own follow-through, the market, the competition, the product. No sensible agency lets that define their revenue.

So instead they scope. A website with 14 pages and a blog. A brand refresh with three deliverables. Six months of paid media with a weekly report. The scope is delivered. The outcome is someone else's problem.

This isn't moral failure. It's structural.

I don't think agencies are bad. I ran one. The commercial model only works if scope is the unit. Trying to sell outcomes without controlling them is how agency founders burn out.

The fix isn't to find a better agency. The fix is to stop buying from them like you're buying co-ownership. Buy the scope. Own the outcome yourself, or hire someone whose job is to own the outcome and whose incentive is to end the engagement when the outcome is in the bag.

The two-part structure that actually works

• Agency: scoped delivery of a specific artefact. Fixed price, fixed timeline, fixed definition of done.
• Internal owner (or fractional): runs the outcome. Briefs the agency, measures the result, kills the engagement when the outcome is achieved.
• What doesn't work: hoping the agency plays both roles. It can't.

If the agency is more invested in the next engagement than the current outcome, you're the one with the incentive problem.