Proof of value sprint

Stage 02 of the KARPINITY Method, sold on its own: in 2 to 4 weeks, a working prototype is evaluated against the metric in your business case, with an evaluation written before it was built.

Duration: 2-4 weeks
Price: On request, scoped to your use case
What you receive: A working prototype scoped to the metric, not a polished product.
An evaluation report: measured movement, or none, against the stage 01 baseline.
A plain recommendation: go to production engineering, no, or adjust and re-test one named variable.
What this doesn’t include: Production engineering: integrations, monitoring, access controls or the governance pack. That is stage 03, and it starts only after a go.
A predetermined result. The sprint runs before a production budget is committed, so an honest no is cheap.
A pilot with live users, or a rollout. Those are stages 04 and 05.
Operating the prototype. It is built to answer a question, not to run in production.
Who you need involved: The metric owner from stage 01: two to three hours a week, mostly checkpoint reviews.
Someone technical enough to check the evaluation method, not just read the result: three to four hours across the sprint.
Access to the data confirmed in stage 01 or in a data readiness review, arranged through whoever owns it.

What it is

A fixed-scope engagement that builds a working prototype and tests it against the metric named in your business case. The sprint tells you whether the idea moves your number, with an evaluation as evidence.

It is stage 02 of the KARPINITY Method. The method allows 3 to 6 weeks for the stage. The sprint fits 2 to 4 because it covers one metric and data that stage 01 or a data readiness review has already confirmed.

Of the three entry offers, this one needs the most time from your side: your team checks the evaluation as closely as ours does.

The success metric, defined up front

The evaluation is written before the prototype is built, and frozen before the first result is read. In week 1 we agree with you:

the metric and its baseline from stage 01;
the data the prototype runs against, and the cases held back for evaluation;
the pass threshold: the movement that counts as a go, and the movement below which it is a no;
how a result is graded, and by whom.

Nobody changes the threshold after the numbers come in.

Week by week

Week 1

Finalise the evaluation design and freeze it.
Set up the working environment and confirm access to the sources.

Weeks 2 and 3

Build the prototype, scoped to the metric.
Interim checkpoints, so the final evaluation holds no surprises.

Week 4 (for the four-week scope; a shorter sprint compresses build and evaluation)

Run the evaluation against the baseline.
Deliver the report and a go, no or adjust recommendation, walked through in person.

What you decide at the end

Go. The prototype met the threshold. The next step is stage 03, engineering the system: a production-grade build with the security, data-protection and human-oversight work that has to exist before live users and personal data meet it.
No. The prototype did not move the metric. You stop here, after a few weeks instead of a few quarters.
Adjust. The result points at one variable. We scope a focused follow-up sprint to re-test it.

Questions

What is the difference between a proof of value and a proof of concept?

A proof of concept shows that something can be built. A proof of value shows whether it changes the metric you care about, measured against a baseline and a threshold agreed beforehand.

What happens if the answer is no?

You receive the evaluation report and a plain recommendation to stop. A no costs a fraction of a failed production build.

Who decides whether it counts as a go?

The threshold is agreed with your metric owner in week 1, before any build work. The report applies it. The decision to proceed stays with you.

Why is the sprint 2 to 4 weeks when the method says 3 to 6?

The sprint covers one metric and confirmed data. If the result is “adjust”, a follow-up sprint re-tests the one variable that needs it.

Can we use our own data?

Yes, and it makes the better test. The data is confirmed first in stage 01 or in a data readiness review. Where it contains personal data, we apply the controls on the principles page before any model sees it.

What does it cost?

Price is on request, scoped to the metric and the data.

What happens next

A go leads to stage 03. A no ends the engagement with an answer. An adjust leads to one more bounded sprint. Start a project with the metric you want to move.