The interviewer asks you to estimate the size of a market from scratch, no data given. It's testing your structure and logic, not whether you know the real number.
Two ways to build the same estimate. Demand side is usually your default, supply side is the pivot.
Start with the buyers.
Start with the sellers.
Demand side is usually the easier one to reason through live, so lead with it. But you can be asked for both. Solve with demand side first, and if the interviewer asks "any other approaches?", that's your cue to bring up supply side.
Once you're on demand side, the population needs splitting one of three ways, based on what's actually driving the behaviour, not habit.
Use only when age is the real driver, like PT sessions, retirement products, toys.
Non-user / casual / frequent. Use when it's habit-driven, like coffee, snacks, subscriptions.
Use when affordability is the filter, like insurance, luxury goods.
One more lens: think in households, not just people. If it's the kind of thing bought once per household rather than once per person, like broadband, a washing machine, a shared subscription, convert population into number of households first (population ÷ average household size), then apply your fraction at the household level instead of the individual level. Easy to miss, but it changes the whole shape of the maths for the right kind of product.
Full walkthrough, demand side, using an age bracket since age drives the behaviour here.
Worth sanity-checking any number like this against something real before you say it out loud. Here it lines up reasonably well with actual UK nappy market reporting, which is the kind of check that separates a defensible estimate from a made-up one.
Worth knowing all four before you walk in, you won't always get to pick which one shows up.