Prove it, operate it, hand it over.
Nobody signs a managed service contract with a firm they have not worked with. So the sequence starts small, each step decides whether the next one happens, and the last step is the one we are judged on.
The four stages
Each stage ends in a decision you make, not a renewal we assume. You can stop after any of them and keep what has been produced.
Scope one question
We agree a single integrity question, the dataset that could answer it, and what a useful answer would look like. If we cannot write that down in a page, the engagement is not ready to start.
Usually two or three conversations. No fee.
Prove it
Six to eight weeks against real data, ending in findings your investigators can act on — or a clear answer that there is nothing there worth chasing, which is also worth knowing.
Fixed fee, fixed window, defined question.
Operate it
The capability moves into continuous operation: scheduled collection, scoring, analyst triage and referrals into your existing workflow, with the assurance record behind each one.
Base subscription, with part of the fee moving with agreed measures.
Hand it over
Documentation, training and transfer to your team, against a plan written in the first week rather than the last. You can take the capability in-house, move it to another supplier, or stop.
Planned from the start, priced as part of the subscription.
What each stage asks and returns
| Stage | What we need from you | What you get | The decision at the end |
|---|---|---|---|
| Scope | An hour or two from someone who knows the program | A written question, a named dataset, an agreed measure of success | Whether a proof of concept is worth running at all |
| Prove | Data access, or nothing at all where public registers will do | Findings, the method behind them, and what it would take to operate | Whether the findings justify standing the capability up |
| Operate | A named contact, a route for referrals, and outcomes fed back | Ranked referrals, the assurance record, monthly measures | Renew, change the measures, or begin handover |
| Hand over | People to train, and somewhere for the capability to live | Documentation, training, models and the operating runbook | Run it yourselves, or keep us for the parts you would rather not |
What handover actually means
Most suppliers leave the exit vague, because a capability that cannot be moved is a contract that cannot be lost. We would rather write it down at the start, while nobody is annoyed.
Handover means your team can run the capability without us: the models and their documented rubrics, the collection schedule, the triage process, the assurance record and the runbook that ties them together. Training is delivered to the people who will do the work, not to a project board.
It is also in our commercial interest to make it work. Handover milestones sit in the same basket of measures the performance component is settled against, so we are paid in part for making ourselves replaceable.
If you would rather keep us on afterwards for the parts your team does not want to own, that is a conversation, not an assumption baked into the contract.
When we would tell you not to
A proof of concept that cannot succeed wastes your money and our reputation. We will say so before taking the work if:
- The data that would answer the question does not exist, or cannot be shared inside the window.
- The question is really a policy question, and better detection will not settle it.
- Someone already inside the agency can do this, and what they actually need is access or air cover rather than a supplier.
- The pattern is already known and the constraint is investigation capacity. Finding more of it does not help if nobody can work the cases.