The incentive nobody says out loud
A retainer priced on hours rewards the agency for the client never learning the work. That incentive rarely shows up as bad faith. It shows up as documentation that never quite gets written and dashboards only one side can read.
Nobody designs it this way. It emerges from a billing model, and billing models are quietly the strongest force in any agency relationship. If the invoice is denominated in hours, every efficiency is a revenue cut, and the organisation will route around it without anyone ever deciding to.
What dependence actually looks like
It rarely looks like incompetence on the client side. It looks like a marketing team that cannot answer a board question without emailing the agency first. It looks like a reporting dashboard nobody internal could rebuild, ad accounts whose structure is undocumented, and a campaign history that exists only in one account manager’s memory.
Each of those is defensible on its own. Together they are a dependency, and the client usually only discovers it at the moment they try to leave.
What changes when you teach instead
Clients who understand the work brief better, argue more usefully, and stay longer on harder problems. The engagement stops being maintenance and starts being the thing only an outside team can do. That is better work and, over years, better business.
The uncomfortable part is that this also makes the relationship easier to end. That is the point. A client who stays because leaving would be painful is not a reference, and they are not going to hand you the harder work either. A client who stays because the work is good will do both.
The obvious objection
If you teach the client to do the work, surely they stop paying you to do it. In practice they stop paying for the part that was always commoditised, and start paying for the part that is not.
Nobody internal is going to build a measurement framework from scratch, run a regional launch they have never run, or tell the chief marketing officer something they do not want to hear. Those are the engagements worth having, and they are not available to an agency whose main asset is that the client cannot log in.
What this means for pricing
It means moving the invoice off hours. Hours price your time, which makes your own efficiency a loss and your client’s learning a threat. Pricing a defined outcome, or a fixed capability transfer with a stated end, puts both sides on the same side of the efficiency question for the first time.
It also makes the conversation honest. An engagement with an end date and a written definition of what the client’s team will be able to do without you is a harder thing to sell and a far easier thing to renew.
Where this does not apply
Not everything should be transferred, and pretending otherwise is its own kind of dishonesty. Work that is genuinely specialist, genuinely infrequent, or carries real regulatory exposure is usually cheaper and safer to keep outside. No in-house team should be learning to run a once-every-three-years brand migration on the job.
The test is frequency. If the client will do something weekly, they should own it, and an agency still holding it after a year is holding it for the wrong reason. If they will do it once every few years, they should rent the expertise and spend their attention elsewhere. Most arguments about scope would be shorter if that question were asked first.
What clients should ask for
- Documentation written for your team, not a deck written for your board.
- Administrator access to every account and tag, held in your own name.
- A dashboard somebody internal can rebuild from the definitions up.
- A named person on your side who is being deliberately trained, and a date by which they are expected to own the work.
- An explicit answer to what happens if you leave in six months.
An agency comfortable answering all five is worth more than one that is not, and the questions cost nothing to ask.