The Best Customer Relationships Outlive the Project.

01It nearly always starts small

A module. A migration nobody wanted. Two engineers for a quarter. A proof of concept with a deadline attached to a board meeting. The work itself is rarely the interesting part; what is being tested is how we behave — whether the estimate holds, whether the status report matches reality, whether the person who sold it is still reachable in month three.

We have learned to treat these as exactly what they are: an audition in which the technical outcome is only half the mark. Delivering well is table stakes. Being straightforward about what is going on while delivering is what people remember.

02What actually earns the second call

We have asked, more than once, and the answers are consistent enough to be uncomfortable. Nobody says 'the architecture'. What they say is that we told them about the problem in the week it appeared rather than the week it became unavoidable. That the same people were still there. That when we were wrong, we said so first.

• 01Bad news travels at the same speed as good news, or faster.
• 02The estimate that is inconvenient is the one that gets given.
• 03The people who did the discovery are the people who do the work.
• 04When something is outside our competence, we say so and name who is better placed.
• 05Nothing is billed that could not be explained line by line to the person paying for it.

Anybody can be trusted on a good quarter. The relationship is decided by what happens in the bad one.

03The long middle, where the work changes shape

Something happens after the second or third engagement that no contract can specify. The questions arrive earlier — before the requirement is written, while it is still a worry rather than a specification. We start being asked what we think rather than what we would charge. Occasionally we are asked to argue against something we would otherwise be paid to build.

That is the point at which the relationship becomes genuinely valuable in both directions. We stop rebuilding context every time and start bringing memory: why that integration was written the way it was, which team owns the awkward data, what was tried in 2021 and why it did not survive. That institutional memory is worth more than any individual deliverable, and it cannot be bought — only accumulated.

04Delivery wins a project. Trust builds a partnership.

The practical consequence is that we optimise for the second conversation rather than the first invoice. It means occasionally recommending less than we could sell, finishing something properly when a faster finish was available, and handing over documentation good enough that the client could, in principle, continue without us.

That last one is the test we hold ourselves to. Every engagement should end with the customer more capable than it began — able to run, extend and argue with what we built. The customers who could leave and choose not to are the only kind of retention worth having.