The year-five test: how we scope for longevity
Most digital work is scoped for the demo. It has to look right at handoff, survive the launch announcement, and get the invoice paid. What happens in month eighteen, when the plugin needs a paid upgrade or the founder who understood the integration leaves, is somebody else’s problem.
We scope differently because our business model forces us to. Most of our clients stay five-plus years, and we host and maintain what we build. Every corner cut in scoping becomes our own operational debt. So before anything ships, it passes what we internally call the year-five test: a short list of questions about what this system looks like when it is old.
The questions
Who operates this in year five? If the answer requires a specific person, the scope is wrong. Systems need to survive staff turnover on both sides.
What breaks when the vendor changes? Every third-party dependency gets a written exit assumption: what we do if pricing doubles, the API changes, or the product shuts down.
What does growth do to it? A catalog of 40 products becomes 400. A list of 800 subscribers becomes 20,000. We scope the data model for the year-five volume, not the launch volume.
Can the client leave us? Uncomfortable question, deliberate answer: yes. Clean exports, documented credentials, no proprietary lock-in. Clients who stay because leaving is painful are hostages, not clients.
What is the monthly cost of keeping it alive? Licenses, hosting, renewals, and maintenance hours get totaled and stated before the contract is signed, not discovered afterward.
What the test kills
The year-five test kills a lot of tempting scope. The clever custom feature that only one developer can maintain. The bleeding-edge framework with no migration path. The automation that saves twenty minutes a week but adds a vendor, a webhook, and a failure mode. If a component cannot justify its year-five carrying cost, it does not ship.
It also kills a certain kind of pricing. Scoping for longevity is more expensive up front, and we publish that honestly: fixed-scope projects start at $8k. The businesses that balk at that number are usually comparing us to work scoped for the demo. The comparison is not wrong; it is just a different product.
The evidence
Eight years in, the numbers are the argument: 50-plus projects delivered, 99.9 percent average uptime, and a client roster where the typical relationship is measured in years, not launches. Systems we shipped in 2019 and 2020 are still in production, still maintained, still earning their monthly keep.
A launch is a day. A system is a decade. Scope for the decade and the launch takes care of itself.
None of this is exotic engineering. It is the ordinary discipline of assuming the thing you build will still be running when nobody remembers building it. In our experience, that assumption is correct more often than the industry wants to admit.