The same discipline, beyond healthcare

The challenge. A national multi-unit brand with thousands of locations was pricing from the center. One price list, applied everywhere, adjusted mostly by intuition and competitor-watching. Leadership suspected it was leaving money on the table in strong markets and losing traffic in weak ones, but suspicion is not a number, and a wrong move at national scale is expensive and slow to walk back. They also had a menu restructure under consideration, a change to sizes and portions, with no way to see how it would move mix and margin short of rolling it out and finding out.

The approach. We modeled price elasticity at the location level across the footprint, item by item, so the question stopped being “should we raise prices” and became “where, on which items, and by how much before demand responds.” Every candidate price move could then be tested against expected volume response before it reached a register. We built the menu restructure as a simulation on top of the same behavioral data: when sizes and portions change, where do customers trade up, trade down, or walk, and what is the net effect on margin once substitution is accounted for. The same framework extended to add-on bundles, tuned to grow the check rather than discount demand, and to pricing for entry into new international markets where there was no local history to lean on.

What it enabled. Pricing moved from a single national judgment to a set of location-level decisions, each one defensible with its own elasticity curve rather than a corporate average. The menu restructure could be evaluated as scenarios before anything was printed, so the options that looked clever but fell apart under real substitution were caught in the model instead of in the market. The common thread is the one that runs through all of our work: a pricing or menu decision you can interrogate, tested before you commit to it, rather than a bet dressed up as a strategy.

This study describes work for a consumer brand that has asked not to be named. No figures are published without the client’s clearance.

The challenge. A global consumer manufacturer could see that a fast-growing class of therapy was beginning to reshape demand in their category, but they could not see it clearly or early. By the time the effect showed up in point-of-sale, it was already priced into the past. The forces at work were visible first in healthcare data, in prescription patterns, but that data sat outside the company’s world and outside their forecasting team’s reach.

The approach. We built geographic prescription intelligence from pharmacy claims to quantify how adoption of the therapy was shifting demand in the client’s category, at a geographic and time granularity a forecasting model could actually use. The deliverable was not a slide. It was a structured, regularly updated dataset plus the methodology behind it, built so the client’s own data science team could fold it into their demand model and own the integration rather than depend on us to run it. We stayed alongside them through that integration, which is where a handed-over dataset usually either earns its place or quietly gets dropped.

What it enabled. The signal became one of the stronger non-promotional demand drivers in the client’s model, a forward-looking input they had not previously had and that their competitors, working only from point-of-sale, still did not. More broadly, it proved a bridge we think is underused: healthcare data can answer consumer business questions, when a clinical trend is moving a consumer category and someone knows how to connect the two. The methodology transferred cleanly to the client’s team, which was the point. Evidence you cannot operate is not much use, in any category.

This study describes work for a global manufacturer that has asked not to be named. Analysis ran on de-identified pharmacy claims. No figures are published without the client’s clearance.