Tell us the data decision you're trying to get right.
Not a demo request. Not a call about our capabilities. Describe the choice in front of you and who owns it, and we will tell you in writing whether there is a forecast worth building behind it.
What happens after you hit send.
Four steps, and you can stop at any of them. Nothing here is billable until there is a signed scope, and most conversations end at step three because the honest answer was no.
The four steps after you send a brief
Acknowledgement
A named principal replies within one business day — not an auto-responder and not a scheduling link. If your question can be answered in a paragraph, that is what you get.
A 20-minute scoping call
Three questions: which decision is at stake, who owns it today, and what data already exists. No deck, no capability tour, no second call to repeat the first.
A written one-page view
The opportunity as we see it: the decision, the accuracy bar it would have to clear, a rough size of the prize and what would have to be true. Free, and yours to keep.
A proposal, only if warranted
Scope, team, price and dates — usually a two-week Diagnose phase first. If there is no real engagement here, we say so and point you at whoever is better placed.
The questions people ask first.
The smallest thing we sell is the two-week Diagnose phase — a decision audit that ends in a written view of your three highest-value forecasting opportunities. It is deliberately small so you can find out whether we are useful without committing a budget cycle to it.
Full build engagements typically run 12 to 20 weeks. Every published project reaches a working model inside a day of compute. If your problem is genuinely a two-day question, say so and we will answer it on the call rather than write a proposal for it.
Almost always. Most teams that get in touch already have analysts, engineers and a warehouse that mostly works. We are not a staffing shop and we are not here to rebuild what functions — we bring the modelling depth and the outside view, and your team keeps the keys.
The Embed phase exists for exactly this reason: adoption, training and handover, with a measurement plan that survives our departure. If your team cannot run the model after we leave, we have not finished.
We will sign your NDA before the scoping call if you prefer, and we have a standard mutual one if you would rather use ours. Send the brief without confidential figures until it is in place — a description of the decision is enough for us to be useful.
Everything on this site is published openly, because none of it is client work — it is analysis on public datasets you can download yourself. That is why you will read a dataset name rather than a logo.
Delivery is remote-first, in fortnightly increments, with a standing review your team can actually attend. It is faster and it costs you less in travel you are ultimately paying for.
The exceptions are worth the travel: Diagnose workshops, steering sessions and the handover at the end of Embed are better in a room. On-site days are arranged around your team rather than ours, so they are a train ride, not a red-eye.
A Diagnose phase can usually begin two to four weeks after a signed scope. Larger builds depend on what else is in flight and are sometimes queued a little longer.
If you are working to a board date or a planning cycle, put it in the brief. We would rather tell you now that we cannot hit it than discover it in week six.
Read the work before you talk to anyone.
Everything we would say on a first call is already written down. Every project carries its own method and its limitations, and every figure in them is computed from a public dataset rather than typed.