How We Spent $300k and Learned Everything about Coffee Delivery

The first step was to test the delivery hypothesis quickly and cheaply. We agreed with the coffee shop that their baristas would take orders via a messenger and deliver coffee to the nearest offices. The process turned out to be long and inconvenient, and as a result we didn’t get repeat orders.

Based on this test, I came to the conclusion that our future service needs to meet the following criteria: delivery time up to 15 minutes, hot coffee delivered right to the workplace, ability to link a card to an app.

That year, I had been collaborating on several projects with my friend, a CEO of a branding agency, and I told him about the coffee delivery experiment and the issues that came with it. Together we came up with a project concept and a process that would help eliminate all the drawbacks of the previous test.

The entire system was designed to coordinate brewing and delivery, and ensure high speed.

At that moment, my friend was closing his agency and vacating an office in a large business center, two buildings with a total area of 323,000 sq ft. We rented a part of the office to launch a ghost café in those business centers.

Before opening, we explored all the coffee shops and vending machines with coffee on the territory of the business center. None was offering specialty coffee. In each of the five coffee shops, we spent several hours at peak time, looking at where people tend to order takeaway coffee, and which drinks are most in demand. We found the most popular coffee shop and set our prices 10% higher than they had.

Then it was necessary to ensure the quality of our coffee and organize a new customer experience at the highest level. We bought one of the most expensive coffee machines and specialty beans. The process and technology of making coffee was provided by the former technical director of the Double B coffee chain.

We wanted the coffee not to cool down, not to spill, and a courier to deliver several drinks at the same time. So we designed and ordered thermal bags. We tested them on real orders and changed three models before it worked as we wanted.

In total, the launch took us three months, during which we created and tested the software, bought equipment and inventory, and hired staff.

We started with the distribution of flyers for free coffee in order to introduce people to the new service quickly. The idea took off instantly, and on the second day we already had 160 orders.

After two months of work, we had 120–160 orders a day and about 350 regular customers. But our profit was about $350 a month, and it did not satisfy us. We realized we faced three challenges.

Challenge 1. Low average check

To increase the average check, we added snacks, pastries, sandwiches, and oatmeal, but it didn’t work. Most customers continued to order just one drink at a time. In addition, we did not have the space to deploy even a small kitchen, so the range of food and snacks could not be expanded.

Challenge 2. Morning peaks and afternoon slowdowns

Up to 60% of orders came from 9 to 11:30 AM and about 30% from 3 to 5 PM. The rest of the time there were virtually no orders. This put two tasks in front of us: we had to cope with the morning peak and do something during the periods of complete stillness.

We managed the morning load with our system. It combined orders so that one courier could deliver up to 5 orders at a time without mistaking them. All coffee cups look the same, so in our app for couriers, we indicated which order belongs to which customer.

Unfortunately, we didn’t figure out how to deal with slowdowns. We ran promotions: happy hours, sandwich + coffee special offers, two drinks for the price of one, but it did not change the number of orders. Perhaps another scenario was playing out during that time: people wanted to leave the office to take a little walk with a coffee to go.

Challenge 3. Human factors

If a barista or a courier were late or did not call out of a shift in advance, delivery time slowed down to 30–40 minutes. It was too long for customers who were used to receiving their coffee in 15 minutes max. On such days, churn rates increased. As a solution, we introduced a subscription option. Customers paid for coffee a week in advance and got each drink cheaper. As a result, 20% of customers subscribed.

We continued to test hypotheses, including those that would help to cope with peaks. The service had high retention rates: users made an average of three orders per week. They also highly rated their experience at 4.7 out of 5. Delivery took 12-15 minutes, and thanks to the thermal bags, coffee arrived hot. We opened several more ghost cafés in smaller business centers, but didn’t reach a cash flow that would cover costs.

With no proper profit, we closed our virtual café. But we did not completely abandon the idea of coffee delivery, and were looking for options to use the power of our all-in-one management platform.

My business partner controlled all the ghost café communications, branding, and offline operations. I handled all the digital services. And business and product strategies we built up together. We also had a team of 6 developers—3 for backend, 1 for frontend and 2 for mobile.

After two months of working 24/7, we developed all the software we needed: website, apps, an admin panel. It took one more month to debug. We launched the delivery in our office, collected bugs, improved the logic and placement of product cards, and shopping carts.

Customer app and website

Everything is simple here. Users select a business center and see a list of drinks and snacks available for the location. When placing an order, the system automatically remembers the entered data: phone number, card, and office location, so that all further orders are completed in one click or tap.

The virtual café was open only during working hours. On other days, users saw a notification that the order could not be made.

Most customers ordered the same coffee every time, so we offered to repeat the previous order right away in the cart.

After ordering, users can track the order status and delivery. They get a push notification when a courier is nearby. After receiving the order, they can rate it and post feedback. All reviews immediately go to the admin panel, the team Telegram channel and the Barista App. If there is any negative feedback, a barista can respond immediately with an offer to replace the drink or an offer to provide a gift at the next order.

If the café runs out of some ingredients, relevant menu items are disabled automatically. The cart recalculates order items in case something runs out while the clients are placing an order.

We were improving the service non-stop, so we suspended the Client app development and focused on the website to make changes faster and cheaper.

Barista app

At the same time as the order is taken, the system checks if there is a free bag for delivery. If it is in place, a barista immediately starts preparing and assembling the order.

To start, a barista taps “Prepare” on the order card. From that moment on, the system starts a timer for 90 seconds and distributes items from the order to the bag. As soon as the order is ready, a barista taps “Done” and the system waits for another 3 minutes. If there are no more orders, the bag status updates to “Full” in the Courier app.

If there is a new order in 3 minutes after cooking, it goes to the same bag, provided that it has empty slots for it. Such a system allows to run the most optimal process of cooking and delivery—couriers carry maximum orders at a time.

The app also helps to monitor feedback: in case of a low rating, a barista can create a reminder to add a bonus to the next order.

The app allows to view delivery status, ask the delivery team to hurry back to the café, start and finish work shifts, calculate payments (we paid hourly), disable menu items and close the café if something goes wrong with the coffee machine or some ingredients run out.

Courier app

The Courier app displays all the bags. As soon as a bag is full, its status changes to “Start delivery”, and the app shows the shortest route to customers. When approaching the office, a courier taps the “I’m here” button. It’s important to notify clients, as a lot of offices have access control systems.

Admin panel

The entire process is centralized in a single system that controls stock levels, orders, ratings, and feedback, has a built-in loyalty program, and allows you to manage staff. Due to the automation of each action, we ensured a high work speed and 15-minute delivery time.

Before the day starts, our system checks the leftovers in all cafés and shows what needs to be reordered. As soon as one ingredient runs out, all menu items that need it automatically turn unavailable. Until stocks are refilled, the menu items cannot be removed from the stop list.

The Admin Panel helps to manage staff. Each barista and courier has a personal profile with information on open and closed shifts. This allows managers to control work schedules and correctly calculate wages. The employee profile displays the number of hours worked and the amount of money earned.

We stopped working in the ghost café mode, but continued to test hypotheses to integrate the system into food services. We tried offering food and drinks delivery from places in business centers, but the model did not work, since Baristika was loved for high-quality coffee, and other cafés and restaurants could not provide it.

We turned to coffee shops in business centers, offered our brand, equipment and business process for launching cafés with high-quality beverages, takeout and delivery. It could be a win-win: we do not invest in staff, and the second party’s team doesn’t stand idle, since their peak hours come at lunch and ours in the morning, so that the order flows do not overlap.

This option would have solved the issues with staff control and increased the average check. Unfortunately, we did not implement it, since the promotion among the owners of cafés and diners took a lot of time and effort, this project was not our major business, and the $300,000 allocated for it, ended. So we suspended this venture.

We faced major difficulties offline, but the all-in-one management platform is still with us. It ran all the field tests and performed well. It is easily integrated into any delivery business, and we are already implementing it in a big ecommerce project.