The decision

How much to spend refurbishing each room across a fast-growing estate, when the standard specification was around £10k a room.

What they believed

That every room needed the same refurbishment to protect guest scores.

What we found

Guests' scores depended on a few room features, and those mattered differently across the estate. We modelled guest scores against room features, prices paid, stay frequency and the range of hotels guests stayed in, which isolated the causal drivers of how guests rated a room. That allowed three lighter specifications in place of one.

What they did

Trialled the lighter specifications, checked guest scores held across regions, then rolled them out estate-wide. Refurbishment cost per room roughly halved.

What we learned

This treated each refurbishment as a one-off. It couldn't tell us how the effect builds and fades over the years after. That's what our later work on investment does.