Tesco and Sainsburys and Deliveroo and Just Eat have got it wrong. When it comes to choosing food within an app I don’t want “Only awesome food” or “little twists”. I want those things, but through the filter of recommendations – only show me what I am likely to like, and don’t show me the rest.

Six months ago I wrote this original blog post and got a lot of feedback that I wasn’t alone in thinking that most grocery and food delivery apps on the market sucked for consumers. They might do the job. Some even might do the job fairly efficiently, which might work for the business that makes the app. But what about the consumer? Where’s the food app with great UX? With a joyful, exciting, inspiring, addictive customer experience? It still doesn’t exist.

But it could so easily. All it would take is Tesco or Just Eat or any of the offer apps to work up some new features. They don’t have to invent them even. They can steal them.

And the easiest way to make an ok food app into a great food app is to create better recommendation filters. All we have to do is steal from Amazon or Netflix or Spotify.

In Kevin Kelly’s awesome book on digital disruption, The Inevitable, he points out that while people think they make their consumption choices based on what they think and like, it’s much more likely that we like what our friends like or what the ‘herd’ does. In social science, it’s called ‘social proof’ and it explains everything from how we vote, to how a yawn spreads through a room. Put simply: When we’re not sure what to do, we do what others do.

This is particularly useful to focus on in food apps as we live in a world where choice has been selected for. There are 50,000 SKUs in a large supermarket. But psychology experiments suggest that too much choice is a bad thing and leads to choice paralysis. In shopper marketing terms, a brand wants to be part of a shopper’s consideration set, but what they forget is that by having too many brands or SKUs without that set makes it likely that they will revert to the product they always choose, or their friends choose…. or worse still, choosing nothing at all.

So how can we help solve this issue in grocery and food delivery apps? By borrowing from Amazon’s recommendation engine and playing on the tried-and-tested ‘people who bought this also bought…’ filter.

If you want to tweak people’s behaviour, it’s a lot harder to do that by persuading people one at a time in an app. It’s a lot easier to pick up on trends and amplify those. Let’s imagine three kind of filters that the likes of Tesco.com or Deliveroo could employ:

1. “People who bought this, also bought…”

This filter is responsible for one-third of all Amazon sales. If you work on a food app of any kind, you want to build this filter right now. It will increase basket spend, and as food delivery businesses live or die by their average basket spends, it would be insane not to build this.

2. “You might also like…”

Or more correctly: “Based on your past purchase and browsing behaviour, and that of your friends and family and people like you, you might be more likely to buy this…”. “People who bought this, also bought” works based on large volumes of data, while this filter would work by inviting people to login to an app with Facebook and then identifying friends and family to understand network influencers. Do this and you’ll increase the number of new products that people try. “Oh look, my sister is buying that new organic baby food, I might try that for my baby”.

3. “Have you tried…?”

Kevin Kelly points out in The Inevitable that there are more types of recommendation filters to be explored, but we haven’t got to them yet, as technology platforms have so far been very controlling about how they filter information for us. As tech firms starts to give users more control over how we navigate through the noise with filters, there are fringe filters that could tweak people’s choices just a little. How about a filter for things we might have already tried, but previously dismissed because we didn’t like it… but with the right prompt, we might give a second chance. For example, you might say you dislike spouts, but you might be presented with a ‘Have you tried…?” filter that recommends “New sweet sprouts, with orange butter”. Might this kind of filter encourage some of us to reappraise our choices?

Filters work for Amazon and Netflix and Spotify because there’s too much digital stuff to choose from, we can never even consider it all, let alone consume it all. Hundreds of new food brands and thousands of new restaurants open every year. We have a limited attention span to even notice them. We need recommendation filters to help us find more of what we will like, and help us ignore everything else. Netflix knows the value of recommendations and employs 300 people to work on how it recommends new stuff to its subscribers, with a budget of $150 million dollars.

If you want to make a great grocery or food delivery app, start with recommendation filters. Some app features are great for business. Some app features are great for customer experience. Recommendation filters do both. They are tried and tested.

The world still needs a great food app. I want an app that makes my food shopping and my food delivery better, more inspiring while providing me safe choices – things I’ll most likely love. Deliveroo’s current poster campaign is all about inspiring people to try new foods, by showing the breadth of food choices. The copy line: “Only awesome food”. But people don’t want all food that’s awesome. That’s overwhelming. People want food they will like. Recommendation engines will do that… and keep people coming back again and again.

Viv Craske is a freelance digital strategist and innovation consultant and the author of Surviving Digital Disrutpion