Home · Case area
Menu boards and merchandising screens: the most testable fixture in retail
If you want to learn in-store experimentation, start with a screen. Not because screens are magic — because they are controllable. Every other fixture in a store resists testing: re-printing shelf strips takes weeks, moving a gondola needs a night crew, and once the change is made, reversing it costs as much as making it. A digital menu board flips between variants in seconds, runs different content at lunch and dinner, and can be assigned per store from a laptop. It is the closest thing physical retail has to an A/B testing tool that ships with the fixture.
The original experiment that made this site's name years ago was exactly that — a digital menu board test in a working food hall, asking whether suggestion-led content moves the basket. We're not going to recycle its numbers; data belongs to whoever's store produced it. What's worth passing on is the design, because the same skeleton works for any menu board or merchandising screen test you might run.
What screens let you randomise
Three levers, all cheap to pull. Content: two creative approaches — say, item-and-price versus suggestion-led ("add a cold brew to that") — assigned to matched store groups. Daypart: the same store can serve as its own comparison when content alternates by week between identical time slots, which partly tames between-store noise. Position and count: board above the counter versus board in the queue, one screen versus two — questions of placement rather than pixels, but the screens make the variants practical to run.
Mechanically, all of this depends on the signage platform doing scheduling and per-store assignment without someone driving around with a USB stick. Most modern tools handle it — dayparting and remote scheduling are standard listed capabilities on platforms like Kitcast, whose player apps run on Apple TV and Android — so the limiting factor in a menu board test is almost never hardware or software. It's experimental discipline.
A design that isolates the screen's effect
- One question. "Does suggestion-led content lift attachment rate on cold drinks?" is testable. "Do the new boards work?" is not a question, it's a mood.
- Matched stores, not volunteer stores. The franchisee who is excited to host your test runs a better store than average; that enthusiasm is a confound wearing a friendly face. Pick test and control stores on trading similarity, then inform them — not the other way round.
- Decide the metric before launch. For menu content it's usually category mix or attachment rate, not total sales — totals drown in weather and payday cycles. Write the success threshold down. The pre-registration habit feels bureaucratic and saves every argument later.
- Run past the novelty. Anything new in a store gets looked at for a fortnight. Effects that survive week four are real; effects that decay with the novelty were never about the content.
- Hold everything else still. The classic menu-board-test killer is a price change or promo landing mid-window in half your stores. Coordinate with whoever owns the trade calendar before you start, not after you notice.
What usually surfaces
Without quoting anyone's numbers: the recurring pattern across published QSR and convenience studies is that menu screens move mix more reliably than they move traffic — they are better at steering a decision already underway than at creating one. Which is precisely why the lazy metric (door swings) so often shows nothing while the right metric (what ended up in the basket) shows plenty. Design for the mechanism, and the measurement gets much easier.