Data running costs · 2-minute video
What does it really cost to keep a dashboard, pipeline or AI model running?
Much more than the bill. The cloud or licence cost is the cheap part. Add the team's time on incidents, questions and changes, and you can decide what to keep, watch or retire all year, not once a year in a clean-up.
First step of the €1,200 workshop. Invoiced after. Full refund if it wasn't useful.
or test it on your own numbers first ↓2 min 16 s · Captions on. Nothing loads until you press play.
The clean-up that grows back
Every so often someone gets fed up. Too much noise, nothing can be found. So there's a clean-up: dashboards, pipelines, an old AI model, the copy left behind by the last migration. It all goes, and for a while it feels solved.
Then the questions start. Where did my report go? Who decided that? Quick requests pile up and, one by one, things get rebuilt. A few months later there's more than before, and nobody can say if the cut was worth it.
It's like clearing out the fridge once a year. It feels good, but it doesn't change what you buy. And in every argument, “we have too many dashboards” loses to “but it's only €50 a month.”
Put the full cost next to what it returns
Most companies already track part of this: the cloud bill, usage numbers, a list of owners. Those show what a product consumes. Not what it costs the people around it, or whether it still earns its place.
The €50 is the cheap part. Add processing, storage, incidents and support, and the team's time taken from what it could build next. Then every product gets one of three answers.
Keep
It earns its place
What it returns is more than what it costs, people included.
Watch
It's still growing
New products need time. Give it a date to look again, not a free pass.
Retire
It's mature and keeps losing
It costs more than it returns, or hardly anyone it was built for still uses it. Let it go, and tell the people who use it first.
Keep, watch or retire: decided all year as part of normal work, not once a year in a clean-up.
Keep, watch or retire: check your products
Start with the product someone defends as “only a small bill”. Add the team hours it takes, and see what it really costs and where it lands.
Runs in your browser. Nothing you type is sent, saved or tracked.
Salary plus overheads. €80 is a common starting guess.
Optional. A rough count of similar dashboards, pipelines or models.
Your result appears here.
Enter the monthly bill, the team hours and whether it is still growing. Start with the product someone defends as “only a small bill”.
Do this across your whole portfolio
A 30-minute call. We agree whether the session is worth it and who joins. A data lead can join if you want one.
The working session, on your numbers. Every dashboard, pipeline and model on the list, its full cost with people included, and keep, watch or retire next to each. Nothing gets switched off in the room.
What you leave with. A clear name for what's costing you, one number you can defend to the board, and a first list of what to stop, with the reason next to each.
In one e-commerce company, working around data was costing €930,000 a year in people's time and tools. €611,000 of it could be stopped. It took three weeks to measure, and the first cancellations were made before the engagement ended.
€1,200, invoiced after the session. Full refund if it wasn't useful. The fee is credited toward the Data Capital Audit if you continue.
Thirty minutes, no slides. If the session isn't worth it, you'll hear that on the call.
What happens in the workshopWhat data products really cost: common questions
What is included in the cost of running a dashboard?
The bill, plus processing, storage, incidents and support, plus the team's time on fixes and questions. Cost tools show the first part well. The team's time is usually most of it.
When should you retire a dashboard or data product?
When it's mature and keeps costing more than it returns, or hardly anyone it was built for still uses it. Retiring starts with telling the people who use it.
We did a clean-up last year. Why would this stick?
This isn't a bigger clean-up. A clean-up removes things without changing how they get added, so they grow back. A full cost next to each product makes the decision part of normal work. More on this: Did our data cost cut actually save money?
Can't we do this in a spreadsheet?
You can. The hard part from inside is pricing colleagues' time and saying “stop” to a product someone built. No stake in any of it makes that easier.
Bring the list. We'll put a full cost next to every product on it.
First step of the €1,200 workshop. Invoiced after. Full refund if it wasn't useful.
Want the whole-business view? What is your data really costing the business?