Trade booths in the messenger: as Burger King uses the chat-bot in the network

Task

Self-service kiosks of network of Burger King restaurants using which buyers can order food inside all are connected to a single internal system of monitoring. She periodically polls all available terminals and saves information, allowing to obtain, on demand, further information on a status of any booth: about its normal work or the failure code. Actually it is the database with the statuses of all booths.

It was necessary to create on the basis of this system of a bot which would notify responsible persons on faults of booths in the territory subordinated to them.

Why bot

Point-of-sale terminals — a frontend of restaurant the quickest response to their faults is also necessary. And if in mail the report can get lost, then in the messenger, with its point notifications and end-to-end availability on all devices, the report on failure is delivered as fast as possible.

Also history of all requests is stored in the messenger in correspondence with a bot, there are convenient file management and, at last, control instructions using which the functionality of work is defined.

Thus, the bot allows to trace visually and quickly a status of booths, to create a picture of work with their faults at the level of each restaurant or to escalate, 'lift upward' the arising problems, generalizing data for heads of the following links. Eventually all this reduces a downtime of booths, increases efficiency of problem solving and allows to maximize their work.

Implementation

Using a monitoring system the bot periodic requests obtains information and sends it at first to directors of restaurants, after small idle time — to operating officers and when idle time drags on — to area directors.

In a system there is a base of personnel in which it is defined to what user are assigned what restaurants. Thus, automatically we obtain information and on the one to whom what notifications to send. The user, being registered in a bot, it is connected to base, and to it notifications about all accountable restaurants begin to come.

Initially the bot intended for the notification of personnel, first of all directors, about the arisen faults. Then reports and summary statistics on failures for certain periods were added to a bot. It is provided in the form of report files: day, week, monthly or for any necessary period. Within this report users see, for example, what problems arose at this restaurant and for what time they were solved, in a generalized view — about average time problem solving, about restaurants with the greatest number of failures, about the most widespread errors, etc.

All interaction is conducted using a command set — since authorization and finishing with an output of reports.

Interesting features

Project implementation, from premeditation of architecture and before acceptance testing, took about 2 months. In total one architect, one developer, one tester and one project manager so it was rather small for 65apps development worked on the project.

• Acceptance testing just dragged on enough as in real network it was impossible to receive all errors at once — it was necessary to wait for them emergence and then already to monitor delivery.

• Also the interesting feature was investigated that Telegram blocks mailing of a large number of messages, including it for spam — we passed to formation of batch files with reports, but not to mailing of separate messages. In more detail in the following item.

• In terms of reference the message format was not defined and it was necessary to remake it based on the first tests. At a large number of booths the huge flow of messages with errors (in one of days about 100.000 messages came — it was failure on the party of a monitoring system, but after all it gave us many "joy") came — we reformatted the text and integrated messages in files that it became much more convenient for the receiver. Now in real time only notifications about emergence of problems come.

• Based on testing users offered a number of improvements of UX: not make a team manually, and use buttons.

• In day in the Moscow region about 1.500 notifications about errors come.

Business impact

Problem: in Burger King network thousands of booths. The idle booth — the bad booth generating losses. It is necessary to minimize a downtime. For this purpose it is necessary to inform quickly personnel on failures, and the end-to-end analytics how restaurants solve the problems with them is necessary for acceptance of management decisions.

Solution: the bot in the messenger which sends information on failures to personnel and the generalized data on the arisen problems and solutions to managers is created.

Result: after implementation of the chat-bot Burger King had accurate statistics on the number of requests and time of their solution. The most problematic devices were revealed, and over them work on correction was performed. Moreover, it was succeeded to systematize errors and to identify one general for all a problem which was fixed already. Restaurants understand much more quickly now that they had a problem with the booth, and do requests in a support service — a downtime of booths was reduced to a minimum.

In 2016 chat-bots became one of the most haypovy subjects in the technology environment — they had to replace applications and to become the new habitat of brands/companies/services and their consumers. However in 2017 interest in bots podugas and hardly somebody still sees in them new revolution. But bots did not disappear — they just became the new service tools, convenient, functional and working in the spheres.

Bot created 65apps for Burger King — an example of such effective solution. It most quickly to provide information to all interested participants via the familiar messenger, accelerating decision making and minimizing a downtime of point-of-sale terminals. This cool solution in which the new IT platform ideally solves the arisen problem and moves business forward.