Your biggest commercial calls — price, promotion, range, partner terms, retention — are mostly made on instinct and a dashboard that reports the past. We engineer the system that makes them deliberately, at the right size, and largely on their own. Identifying the business signals that matter, and automating the optimal decision.
Every consumer business already has the raw material — EPOS, CRM, web analytics, panel data — and the means to act: a media agency, CRM systems, stores, a range team, account teams. What’s missing is the part that reliably turns one into the other. The Commercial Decision System sits in the middle: it distils signal from the noise of dashboards, decides the optimal move, and issues it as an instruction your existing teams and tools can run — so the operation trades by design rather than by improvisation.
The dashboards, feeds and models you already own. We distil from them the few things genuinely worth acting on — we don’t replace what’s there.
The part that decides: what is heading where, what moves it, what each option is worth, and the rule to act on. This is what we build.
The agencies, systems and teams that execute. We issue the correct instruction into them — we don’t build that infrastructure.
A dashboard shows you everything. A system tells you what matters.
We determine which movements are genuine signals to act on — a demand shift, a competitor’s price move, an availability problem, a cohort starting to churn, a change in trade terms — and which are noise to leave alone. The output is fewer, better decisions, not more reports.
Inside the layer, every decision passes through the same four steps. Each one maps to a methodology we use in production — nothing here is theoretical.
Our proprietary forecasting library projects the trajectory — of demand, price response, cohort value — rather than reading last period’s number. It is the same production-grade forecasting we build for trading decisions, which is why the rest of the system can be trusted.
How the forecasting library works →Correlation is cheap; cause is the point. This layer models the levers that change the outcome — price, promotional depth, distribution, retention spend — and how much each one moves the needle, so the system reasons about what would happen if, not just what happened.
Methodology →Every option is valued as a distribution of outcomes, not a single number — and sized to the strength of the evidence behind it. Big, well-evidenced edges get backed hard; thin ones get a small, reversible bet with a clear point to stop.
Methodology →The system resolves to a decision rule: the move it makes, every time those conditions recur. The same quality of call gets made whether or not your best analyst is in the room that day — which is what makes the operation scalable rather than heroic.
Methodology →
The system is built to run the ninety-nine per cent of decisions a system should — reliably, repeatably, at a quality the best operator would sign off. The expert’s time goes entirely to the one per cent that decides the year. You supervise it in plain language — interrogate a call, change a constraint, set a guardrail — without touching a spreadsheet. The judgement no system should automate stays with you; everything else stops being your problem.
The system doesn’t hand you another deck to debate. It issues the precise move — reprice this, hold that promotion, flex these partner terms, intervene on this cohort — into the systems you already run. We don’t build that infrastructure; we make it act correctly. That is the difference between knowing what to do and the business actually doing it.
The forecasting library at the core of layer one is real, working IP, with published benchmarks you can inspect. A practice that builds production decision systems gives unusually rigorous commercial advice — because the advice is what the system is made of.
Built for the boldest consumer businesses with complex routes to market.
The diagnostic is the fastest way to find out. If a decision system isn’t right for you, we’ll tell you.