To Eat List is a social network built around the most essential IRL social thing in human culture: food. To Eat List is for sharing your favorite restaurants with all your friends and followers, and getting restaurant recommendations from people whose taste you trust.
The Need
Traditional "restaurant review" apps are plagued with bots, trolls, paid reviews, and angry rants from strangers. To Eat List needed to be as much the antidote to this as possible.
To Eat List is based on a simple premise–the best recommendations are from other users you know and trust, not randos on the internet. To Eat List would focus first on floating recommendations from people you follow, then using overall popularity as a fall-back in places where your friends haven't added favorites yet.
It would also do away with reviews in favor of a binary recommendation structure. Would you recommend this place to a friend, yes or no? If yes, onto your favorites list it goes! If no, that's cool – no need to spread negativity, just remove it from your list and move on. This structure would help minimize "gaming" the results, like you see with traditional reviews and average star ratings systems.
Challenges
- A complete U.S. restaurant database was needed from day 1, so a first-party database was out of the question
- Need to send push notifications based on location data, like a reminder to favorite or remove a restaurant from your list just after you've left
- Required alternative discovery methods besides just your friends
- Sought an economical, yet robust and accurate location database
- Sharing To Eat List restaurant profiles outside the app required a way to drive downloads without gating the content
Our Role
After painstakingly reviewing and testing location databases and APIs from Google, Facebook, FourSquare, and more, the most reliable economical solution for an API-enabled, third-party location database for To Eat List was Google Places. This provided a robust, nationwide, accurate restaurant listings from which to build a restaurant data model.
We constructed an abstract first-party data model for restaurants in such a way that it doesn't matter where the metadata To Eat List requires would come from, so long as the source had similar metadata. We simply match up the data points and feed them into the database, and the To Eat List interface displays it beautifully. This enables maximum flexibility should the data sources change in the future.
To enable perfectly timed push notifications, we used passive polling of users' device locations based only on major location changes, then increasing in frequency and accuracy if the user comes close to the location of a restaurant on your list. This enables the app to know with reasonable confidence when you've just spent time at a restaurant on your To Eat List without being invasive nor draining device battery in the process. When you leave a defined "geofence" area around the restaurant's location, we trigger a push notification asking what you thought of the restaurant, and if you'd recommend it to a friend.
This intelligent notification system drives engagement at the perfect time–just after someone has tried a new restaurant–and encourages them to either remove the listing or add it to their favorites, so that their followers know it's worth checking out!
Learn more at ToEatList.co.
