The Psychology Behind Why Personalised Recommendations Work

There's a reason personalised recommendations show up everywhere from streaming services to online shopping, and it isn't just that companies find it profitable. There's genuine psychology behind why a relevant suggestion changes behaviour more effectively than a generic one, and none of that psychology stops applying just because the context shifts from a screen at home to a table at a restaurant.

Choice overload is real, and menus trigger it constantly

A menu with forty items asks a hungry guest to evaluate forty options before making a decision, which is more cognitive work than it sounds like when you're not the one designing the menu. Research on choice overload consistently shows that too many options, without any structure to narrow them, leads to slower decisions, more second-guessing, and sometimes to a guest picking nothing new at all and defaulting to whatever feels safest. A menu that quietly narrows the field toward a few likely-relevant options reduces that burden without removing any choice at all, since everything else is still there if a guest wants to look further.

Relevance builds trust faster than a generic pitch

• A suggestion that matches something a guest has already shown they like feels considered rather than random
• A guest is more willing to try something unfamiliar when it's framed as similar to what they already enjoy
• Relevant suggestions reduce the fear of ordering something disappointing, which is a real barrier to trying new dishes
• A pattern of good suggestions over multiple visits builds a kind of trust that a single great meal alone doesn't fully establish

This is closely related to why word-of-mouth recommendations from a friend carry more weight than an anonymous online review. The friend knows your taste, so their suggestion feels calibrated to you specifically. A personalised menu suggestion works on a similar principle, even though it's a system rather than a person doing the calibrating.

The anchoring effect of what appears first

What a guest sees first tends to set a mental anchor for the rest of their browsing, a well-documented effect in decision research. If the first dish a guest sees is well-matched to their taste, it shapes how favourably they view the whole category, even the parts they scroll to afterward. This is part of why the order dishes appear in matters as much as which dishes are included at all, and why a static, identical order for every guest wastes an opportunity that costs nothing extra to capture.

The verdict prompt as a psychology tool, not just a data source

Asking a guest for a quick thumbs up or down on a dish right after it arrives does two things at once. It gives the system a clean signal to learn from, and it gives the guest a moment to feel heard about their own experience, which matters more than it sounds. People are more likely to trust a system that visibly asks for their input and appears to act on it than one that silently infers preferences without ever checking in. That's part of why roughly 41% of tables leave a dish-level verdict when asked directly, compared with roughly 3% who'd take the time to write an actual review unprompted. The quick prompt captures a signal that would otherwise be lost entirely.

Why familiarity and novelty need to be balanced carefully

Psychology research on recommendation systems generally draws a distinction between suggestions that feel safely familiar and ones that feel like genuine discovery, and the best systems don't lean entirely on either. A menu that only ever suggests what a guest has already ordered starts to feel repetitive and stops adding value beyond convenience. A menu that only ever pushes something new risks feeling presumptuous, guessing at a guest's taste without enough basis. The stronger approach mixes the two, occasionally surfacing something adjacent to what a guest already likes rather than either an exact repeat or a total stretch, which keeps the experience feeling attentive rather than either stale or random.

TasteGraph uses exactly this kind of relevance to shape what a guest sees first in each category, based on what they've told the menu or what they tend to order, which is part of why restaurants running it have seen an example 11.4% lift in average order value from guests engaging with dishes that actually suit them.