METIS LABS
There is one question that will tell you more about a software supplier than any case study, and almost nobody asks it: what would have to be true for you to tell me to stop? Here is why we answer it in writing before the first sprint, and what committing to that answer costs us.
Ask it out loud in a sales meeting and watch what happens. Most suppliers have never considered it, and the reason is not dishonesty. It is that nothing in the arrangement requires anyone to. There is a monthly fee, a roadmap with a next quarter on it, and a backlog that never empties because backlogs never do. Continuing is the path of least resistance for both sides. You keep a team that already understands your system, they keep a predictable month, and the question of whether this quarter is still worth its price never quite comes up.
That is worth sitting with, because it means the decision to continue is almost never actually made. It is defaulted into, twelve times a year, by two parties who each have a small reason not to raise it. And a supplier who cannot describe the conditions under which they would lose your business on purpose is not really selling you a project. They are selling you a department, at agency rates, with no agreed reason to ever close it.
So we write the ending down before the beginning. Not a termination clause, which is a legal instrument for the day things go wrong, but a plain description of what finished looks like on the day things go right.
Two endings worth planning for
The first is the clean one, and it is rarer than it should be. There was a problem, it had a number attached, the software moved the number, and the number is holding. Our export work for a Dutch accountancy firm is the tidy version: one stated problem, a self-serve pipeline replacing a manual spreadsheet process, export errors down from five to eight per cent to near zero across twenty clients, six weeks start to finish. There was no honest way to spend a seventh week on it. The right ending was obvious because the target had been named before anyone opened an editor, and you can read the whole thing on the case page.
The second ending is the one people avoid, because it arrives while everything is going well. Diminishing returns on good work is much harder to see than failure. Long engagements are perfectly legitimate. We have been building Brandhub for around twenty-two months, and every one of those months has been justified by what the next block of work was worth. Duration is not the test. The test is whether the next three months buy more than they cost, asked honestly, at a moment when the answer might be no.
That is the uncomfortable part. A project can be healthy, well run, on schedule, staffed by people you like, and still be the wrong place for your next fifty thousand euros. Nothing about the project will tell you that. The team will be busy, the demos will look good, and each invoice will look reasonable next to the one before it. The only thing that surfaces it is a date in the calendar where stopping is genuinely on the table.
We have written before about picking that figure in Set the number before you write the code. The ending is what the number is eventually for. A metric that can only ever justify more work is not a metric. It is a marketing device with a decimal point.
What we write down before the first sprint
Four things, in plain language, in a document you keep:
- The number, and what it looks like when it is met and holding. Not just the target, but how long it has to stay there before we call it done. A figure that spikes for a fortnight and settles back has not been achieved, and a figure that has held for two quarters does not need us standing next to it.
- A review date where stopping is a real option. This is not a renewal conversation. A renewal conversation starts from the assumption of continuing and asks how much. This one asks whether, and it is allowed to end with us recommending you spend the money elsewhere.
- What finished actually includes. Documentation written for a stranger, a handover session with whoever inherits the system, and the thing already running on infrastructure you own. If ending requires a transition project, it was never a plan. It was a hope.
- Who takes it over, named early. Your own team, another supplier, or us on a much smaller footing. Deciding this while everyone is calm costs almost nothing. Deciding it in the week somebody wants out costs a great deal.
None of this is complicated, and none of it is clever. What makes it rare is that it commits the supplier to scheduling a conversation they would rather not have.
What it costs us
The obvious cost is revenue that was working. A studio of our size plans hiring against expected months, and an engagement that ends on purpose in March leaves a hole in the year that has to be filled with new work. Filling it takes selling, and time spent selling is time not spent building. The traditional arrangement, where projects taper into indefinite maintenance, exists partly because it solves a genuine problem for the supplier. We have given that solution up, and we should not pretend it was painless.
There is a subtler cost, which is that it removes our easiest conversation. A roadmap is a pleasant thing to present. There is always another feature, and clients rarely say no to a good-looking one. Committing to ask whether the next quarter is worth its price means putting a difficult question in front of someone who was not asking it, sometimes about work we would enjoy doing.
And the failure mode we watch for in ourselves is not greed, which is easy to spot in other people and easy enough to spot in yourself. It is drift. It is entirely possible to stay six months too long while believing every single month was justified, because each month looked reasonable on its own and nobody stood back far enough to see the shape. That is exactly why the date goes in the calendar at the start, when we have nothing invested in the answer, rather than being raised whenever it happens to occur to somebody.
Ask the question before you sign
You can use this without hiring anyone, and it costs you one sentence. Before you sign, ask your supplier to describe the conditions under which they would tell you to stop paying them. Then ask them to write it down.
Watch which way it goes. A supplier who builds well will have an answer, may look relieved to be asked, and will probably improve on your version of it. A supplier who answers with long-term partnership, or a maturity roadmap, or a conversation about scaling the engagement, has told you something specific: the arrangement has no end state, so the decision to continue will never be made, only defaulted into. That is not automatically a reason to walk away. It is a reason to write the ending yourself and put it in the contract, because nobody else in the room has any incentive to.
If you would rather talk it through than write it alone, open a conversation and tell us what you are trying to fix and what it is currently costing you. We would rather agree the exit now than discover in a year that neither of us ever chose one.