Five shapes of work.

Most engagements begin with a kick-off, usually a fortnight of work. The goal is to set up a Marketing Mix Modelling project for success before anyone touches a model.

We sit with marketing leadership, the analytics lead, and the finance partner who will eventually see the output. We help you write the brief, prioritise the questions the model needs to answer this year, and – where it helps – run a short vendor selection alongside. We listen for the decision the room is actually trying to make, and we write it down in a single page.

The deliverable is a short briefing document, short enough to read in one sitting, that becomes the contract every later decision is held against. From there, the model is usually built by your own team or a specialist vendor, with us staying in the loop. We rarely build models ourselves, though it has been known, where it is the right answer and time allows.

A short mid-project review. Used at any stage in an MMM build, before the model is locked in, to check the direction of travel and surface the assumptions worth challenging.

We ask for a few short inputs – the brief, the current model spec, the first set of outputs if you have them – then run a 60–90 minute session with the analytics team. The aim is the two or three changes most likely to improve the model before it lands, plus a calm reading of the fault lines worth knowing about now rather than at sign-off.

The deliverable is a short written report with recommended actions, plus a 60-minute follow-up consultation a fortnight later.

Best for Analysts mid-build who want a second pair of eyes, and clients who would rather catch a problem in week six than in week twelve.

A review of an MMM engagement after the model has been built, against the objectives set when the brief was written. Used when the deliverable lands and leadership wants an independent view before they act on it.

We start with a questionnaire for the analyst and the client. Then two sessions of up to 90 minutes each – the first to walk through the findings against the brief, the second to walk through the recommendations and what we would add or push back on. Both sessions can be run online or in person.

The deliverable is a detailed written report covering the model's strengths, the assumptions worth knowing about, and the recommendations we would defend versus the ones we would hedge.

The deepest of the three reviews. Using the supplier's data, methodology and – where available – source code, we check that the published results can be reproduced from the data, then test whether the headline conclusions hold up under scrutiny.

The audit asks three questions of the work. Is it reproducible – do the same results come out when someone else re-runs the model from the source data? Is it auditable – can every headline number be traced, step by step, from source data to model result? And is it statistically sound – do issues surface that were never flagged in the original read-out?

Where the original code cannot be shared, we run a technical review with the model builder against a written methodology questionnaire. Either way, our role is advisory. Where we find problems, we work with the team that built the model to fix them. Only if the issues run deep will we recommend a rebuild – and how that is done, and by whom, remains your choice. We can help with that too.

The standard we hold the model to is the Tuesday-morning-with-the-CFO test: can the recommendation be defended, line by line, in a room with the people whose budget it moves? The report says plainly how solid the results are – and where the answer is "not solid enough", exactly where the cracks run and what would have to change to close them.

The deliverable is a written report and three sessions of up to 90 minutes – one for findings, one for technical detail, one for recommendations. Findings are anonymised and, with permission and after sufficient time, added to our audit archive.

For clients who want a different, more permanent relationship. Our mentorship programme runs across a planning year rather than a single deliverable – across the briefs, the model builds, the read-outs and the decisions that follow.

Shape and cadence are agreed up front. Typically one or two sessions a month plus availability between for the questions that come up in the room. We discuss what your team needs in detail before we start, and we agree the engagement against that, not against a template.

Best for In-house analytics teams growing a Marketing Mix Modelling capability, and brand teams who want a steady second opinion across the year rather than per-engagement.