Quarterly Product Review — Q1
At Local Kitchens, we build technology that delivers a seamless and delightful culinary experience to our guests — from browsing and ordering on one of our guest-facing apps (web, mobile, and in-store kiosk) to facilitating efficient and accurate kitchen operations with our KitchenOS. We are a tech company and a food company, but first and foremost, we are a guest-obsessed company. Today, we are excited to share five of our product launches from Q1.
1. Mobile
Local Kitchens seeks to boost selection and convenience while offering a unique mix-and-match experience, regardless of our guests’ current food mission. Based on guest feedback, Mobile is the most convenient platform for digital ordering on the go. We listened to the feedback and shipped our iOS mobile app in 8 weeks. We are now live in the App Store and will be launching in the Play Store very soon.
Mobile guests enjoy a more seamless check-out experience and unified order status-tracking across all our direct sales channels. When a guest orders from the kiosk, we identify their mobile account and trigger a push notification that links to a status-tracking view on their mobile app. Since launch, mobile is driving 45% of new guest acquisition.
2. ETA Prediction
After every order, we send a feedback survey to our guests. We read every piece of feedback we get. A common driver of negative experiences is guest wait times that exceed the original ETA. Similarly, it is a negative experience for guests to arrive on time only to see that their order has been sitting out and getting cold.
In order to solve this, we began modeling the cooking times of every line item against the current state of the kitchen when it enters cooking. The way we do this is by first calculating the count of every menu item in the kitchen at the time the new order enters the kitchen. We then add in the menu items from the order we are predicting the ETA for. We labeled each of our historical examples with ETA categories (5–10 min, etc.) and then trained a Random Forest Classifier on this historical data, using 100 estimators and bootstrap sampling. We are now able to predict ETAs with > 90% accuracy within a 10 minute range. Currently, we are simulating the model in production with new and unseen data to determine how the model generalizes to new brands and new locations.
3. Chit Printing
The most inconvenient and frustrating experience with a family delivery meal is when Dad’s food is missing from the bag. We take these instances very seriously and aim to minimize missing items from orders. We think of our kitchens as an A-plant orchestrated by a staff member we call “the expo.” The expo ensure items in a multi-brand order are ready simultaneously and that the right items are going into the right bags. When the kitchen would get busy and the expo would get backed up with many similar looking items, throughput would screech to a halt because they would manually open boxes in order to ensure they were placing the right items together. During these stressful scenarios, sorting would end up incorrect and issue rates would increase due to missing items.
In order to mitigate these cases, we have developed a set of sticky labels that cooks place on dishes. This helps our baggers ensure the right items are going into the right bags. Downstream of that, our order manager app ensures the right bags are handed off to the right people. The added benefit is that on group orders, our guests will now know which item belongs to whom.
This product launch introduced interesting hardware challenges to the team. For normal receipt printing, we utilize a server direct print API while polling for receipt objects. However, the printer we sourced that was capable of printing and splitting sticky paper receipts did not support this API. In order to workaround this limitation, we had to install firmware onto the device that would support server direct print.
4. Dine In
We launched our Palo Alto location on California Avenue in December. We were overwhelmed by the positive reception and feedback from the community. We noticed that relative to other locations more guests were ordering at our in-store kiosks and eating at the restaurant. We heard our guests loud and clear — they wanted the option to dine in. Now our guests can select the dine-in option on the kiosk and we will notify them when their tray is ready.
5. Referrals
We have found that the most efficient growth comes from creating magical experiences for guests. When we succeed at that, guests tell their friends who tell their other friends, leading to an organic growth loop. In order to augment this loop, we launched a referrals product where our guests can give $10 to a friend and in turn receive $10 in store credits to be redeemed on their digital orders.
We are building to delight at Local Kitchens. Come join us!
