Thinking · how-we-work
Why we stopped selling support hours
Every support retainer dies the same way: a timesheet argument between a developer and an owner who each think the other is being unreasonable. Both are right. The unit is wrong.
01The argument hours always start
A field on a report shows the wrong total. The owner rings: “that was only a small thing.” The developer knows it took three hours. Someone is about to feel ripped off — and it doesn’t matter who.
That isn’t a personality problem. It’s a unit problem. Hours measure what it cost us. You only care what it did for you. Two people looking at the same job through different units will disagree about it forever.
So we stopped. Kindra doesn’t sell hours on anything we look after.
02The principle that replaced them
Broken
It doesn’t do what we built it to do. A bug, something that stopped working, a promise that isn’t delivering.
Fixed. Free. Unlimited. No argument.
New
It doesn’t exist yet. A new field, screen, report or integration.
A job — your monthly changes, or a quote. You’re told which first.
Notice what that does to the old argument. The question is never “is this a lot of work?” — that’s about our effort, so it’s a negotiation. The question is “did this ever work?” — that’s about the system, so it’s just a fact.
03What a Change actually is
If new work isn’t hours, it has to be something. We call it a Change:
This is one Change
- Add a field to a screen and put it on the report
- Change how a calculation rounds
- A new export, or a new filter
- Email someone when a job closes
- Add a permission to a role
This gets quoted
- Rebuilding how data is stored
- Changing how money is calculated
- Anything touching several parts at once
- A brand-new integration
We say the second list out loud up front — not in month two. A fixed allowance that quietly absorbs a rebuild is a promise nobody can keep.
04Three rules that make it honest
- Counted when it ships, not when you ask. Four goes to get it right is our problem. You never pay for rework. Nothing is “small” or “large” any more — it’s shipped or it isn’t.
- Scoped in writing before it starts. You get one paragraph and one line that says what done looks like. You reply yes. No yes, no start. Anything outside the paragraph is a new Change.
- Unused changes roll over. One month, up to double your allowance. A quiet month isn’t a wasted one — and nobody banks six months and lands a project on us in January.
05Why this matters more than it sounds
Software rarely breaks loudly. It breaks quietly.
An equipment hire business we look after had a system that couldn’t undo a mistake. Two wrong clicks before dawn, the month approved minutes later, and one machine went out invoiced at zero. Nobody was asleep at the wheel. There was simply nothing watching.
The things that stop it — hosting we control, monitoring, daily backups with the restore actually tested, patches as they land — cost almost nothing per client to keep doing. So they’re in every plan, unmetered, with fixes alongside them, unlimited.
Hours could never say that. A retainer of ten hours a month says nothing about whether the backups restore. A Care Plan does.
06What it looks like in practice
Three plans, laddered plainly. Every one carries the same three guarantees: you own the code, it’s backed up daily and we test the restore, and you can leave on thirty days’ notice and we hand you the lot.
Plan 1Keep It RunningCare only. New work quoted.
Plan 2 · most clientsRunning & ImprovingTwo changes a month. Four-hour response.
Plan 3PartnerFive changes. Same-day, emergency line, a seat at the table.
The full offer, the story and every question people ask us are on the Care Plans page. If you’d rather just know what state your own system is in, send it to us and we’ll tell you straight.
If this is the leak you recognised, half an hour with us finds out what it actually costs.
