How discovery works · Evergreen

Thousands of Diner Choices or One Influencer

By Shaz Khan · Published August 28, 2026 · 5-minute read

A food influencer gives you one person, one visit, one order, one night. Halal Food Pass gives you what thousands of diners actually chose when real alternatives were sitting on the screen next to each other. Both are worth something, but they are not the same size of evidence.

What one visit can tell you, and what it cannot

Start with what an influencer does better, because it is real. They eat the food. We do not. They can tell you the lamb was dry, explain why a Hyderabadi biryani is not a Lahori one, and make you want to drive forty minutes on a Tuesday. No ranking system produces that.

Now the limits.

They may have been paid to make the video, and the restaurant may be ordinary. You drive out on the strength of it and get an ordinary meal.

The kitchen may have known they were coming and cooked better than it does on a normal Wednesday. You order the same dish and it is not the same dish.

And their palate may simply not be your palate. You get their palate anyway.

All of that arrives from a sample size of one. One kitchen, one night, one person's taste.

Now put thousands of diners with similar taste in front of the same choice. When they keep landing on the same restaurant, that is far harder to explain away as a good night in the kitchen or a paid post. It is a restaurant that earns attention repeatedly, when nobody is filming.

That is what we are building: exposure earned by consistency, not by one good night's cooking.

What we can see that one person cannot

Every time our site puts restaurants in front of someone, a small decision gets made. This one, not that one. Tonight, at this hour, in this neighbourhood, against these specific alternatives.

One of those decisions tells you nothing. Thousands of them tell you something no reviewer can: when this restaurant and four comparable ones were on screen at the same moment, which one did people actually pick?

To be exact about what that is: we see which restaurants diners choose to look into, not which meals they ate. That is interest, not a verdict, and we treat it as interest.

Two rows go to the influencer and we are not going to pretend otherwise. The rest is why we think we are the better answer to "where should I eat tonight."

From what a crowd chose to what you will choose

Knowing what diners picked at 8pm on a Friday is useful, but it is not yet personal. The bridge between the two is simple enough to say in a line:

People who choose alike, eat alike.

This is what rescues a crowd from being useless. A crowd on its own produces an average, and an average is a Top 10 list, which is the thing worth complaining about in the first place. Split that same crowd into people who behave like each other and the average becomes worth having again, because it is now the average of people who want what you want.

Where we are today: Halal Food Pass learns from aggregated patterns, not from individual profiles, and it does not need to know who you are. As reliable patterns accumulate, discovery can become increasingly personal. That is the direction, and we will say so plainly when it arrives.

What we got wrong

We built this the naive way first.

Our early system reported strong, confident patterns about what diners preferred. Some of them looked dramatic. We were pleased with ourselves.

Then we tested those patterns properly, and most of them disappeared.

They had never been preference. They were Halal Food Pass reading its own layout back to itself. The site had shown certain restaurants more often, collected the taps that followed, and reported its own decisions to us as though they were facts about you.

That is the most useful thing we have ever learned about our own product. It is also why we are slower to make claims than sites that have never checked. If you want the principles behind how discovery actually works, we have written those down separately in how Halal Food Pass decides what to show you.

Why your phone seems to be listening

You have had this happen. You mention something out loud, and an hour later it is on your screen. It feels like the phone was listening.

It almost certainly was not. What happened is less sinister and far more impressive. The pattern of what you do, where you go, what you linger on, and what people like you did next was enough to predict what you were about to want. No microphone required. The data already knew.

That is unsettling when it is pointed at selling you a mattress. Point the same idea somewhere it actually helps you.

It is 7pm on a Thursday. You are tired, you have decided nothing, you are just hungry. You open Halal Food Pass and the first thing on the screen is the thing you were about to think of. Not the one with the best video. The one you were going to want, at the hour you were going to want it, close enough to actually go.

We are not there yet, and we will tell you plainly when we are. But that is what we are building toward, and it is exactly why we are careful about what counts as evidence. A system that predicts what you want has to learn from what people genuinely preferred, not from what a homepage happened to show them. Build it on the wrong signal and you have not built understanding. You have built a very confident machine for recommending whatever was already popular.

So who should you trust

Watch the influencer. They are better than us at making a place feel worth going to.

Then ask the question they cannot answer. Out of everywhere you could eat tonight, is this actually the one? What else was within six minutes? What did people pick when this restaurant had to compete instead of being the only thing on screen?

One person can tell you what they tasted once, months ago, possibly on someone else's money.

We can tell you what thousands of people chose to look into when they had real alternatives, and we can tell you when we got it wrong.

Decide for yourself which is more likely to be right about dinner.

Frequently asked questions

System descriptions on this page reflect Halal Food Pass's discovery architecture as of August 28, 2026, and continue to evolve as the platform grows. Halal information is provided by each restaurant; Halal Food Pass is not a certification body.