“You could consider work with data to be a sort of base for the management development of the company, which is in large part built on standardizing and delegating functions”

Burger King chain leadership have created a rather unique large-scale restaurant chain management system based on objective indicators. Managers have access to objective, demonstrable, and timely information connected to their work, which ensures they remain engaged in improving the company’s efficiency.

Maxim Sharapov, the head of information system devel-opment at Burger King Rus talks about the advantages of a meritocracy, developing a businessman mindset for restaurant owners, and about complete outsourcing of BI processes.

— When did Burger King start transforming?

We had to go through a preparatory stage for the transformation. The Burger King chain began to develop very quickly in 2012, when we had a few dozen restaurants and gained a strategic partner. After the chain got financing, it started expanding. This is when I came aboard, working on information system development strategy.

The basis for all subsequent work was creating a singular data source necessary to deploy digital services after the fact. We decided to implement an ERP system, combining scattered data from different departments into a single database. Moreover, ERP became our master data system. This created the foundation for further digital transformation.

—What’s your history of actively working with the data? When did the push happen, what led to it, and what were the results?

After we finished integrating the ERP system, which was a long, but important process, we could start spending resources on specific business problems. We had all the prerequisites for proper business analytics: we finally got data that we could guarantee to be complete and correct. The company prefers a pragmatic approach to business, the upper management likes using “quick wins”: hitting concrete goals as quickly as possible. While we knew that we couldn’t integrate the ERP system quickly, and we had to bear with it for a whole year, we could easily organize work on BI reporting according to Agile. The process is built on specific application creation tasks that business departments send us.

— What platforms and products did you use?

We used Qlik as our platform of choice, and were quickly connected to a developer team to the project. This platform lets us act quickly and move forward without needing to create a data repository. Using Qlik, we started doing quick projects. Our main task was giving business hints and suggestions for correcting their work in order to maximize profit.

— Why did you decide against having a data repository? Seems like everyone is doing it these days.

We believe that we don’t need one. Sure, a repository is absolutely necessary when you have many sources of diverse data types. But we don’t have that many data sources, and only really have two main ones: the ERP system, and the POS system. They’re rather homogenous, and provide us with high quality data, so we don’t need the intermediate step of converting all data to a single format. To interact with external systems, like client feedback or auditing systems, we use Excel files, aggregated and ready reports.

Qlik itself often presupposes a lack of repository anyway. The platform easts whatever you feed it, and ETL processes can be defined within the system itself.

— Do you have a specific strategy for working with data? If so, what is it?

We don’t have a well-defined strategy per se, but we do have a core principle: the company makes data-driven decisions quickly and regularly. We define our data requirements from this. For example, we make sure that no more than 15 minutes pass between an event occurring in any of our 600 restaurants, and it being registered in reports. This means that all types of data quality and integrity controls must be a part of the system itself.

The most important thing is that a huge number of people are involved in working with data. We have conference calls with each restaurant daily to determine what they are going to do based on the indicators we get from the system.

— Is speed really that important? How do you reach data relevance when working so close to real-time?

The first major result of working with data that we got was the “Routine” project, at the end of which all restaurants started getting detailed daily reports. The next step was focusing specifically on speed. We rethought how we worked with data sources and started connecting to local POS databases, rather than a central data-base that consolidated data at the end of the day. This was necessary to ensure that restaurant employees could see what their indicators for the day were looking like close to real time. If something was trending wrong, they had time to adjust.

— Switching to data-driven management is a seri-ous cultural transformation. People don’t like to change. How do you make sure such a change is even possible?

It’s both easy and hard at the same time. The company likes to stick to a meritocratic philosophy and give everyone their due. Anybody that brings the company a result that we can measure in money should get a slice of that result. Any restaurant employee can bring up an idea that can increase the company’s efficiency. These ideas can be brought up to their direct supervisor, and then to their management in tandem. It’s very important that this process of running ideas up the chain be very fast, and good ideas should be quickly implemented. You don’t need to have endless theoretical discussions, you need to start actually doing things, and the pilot project will show its value in practice.

It’s important to foster a businessman mindset in restaurant general managers. They have the right to try out new ideas, no one is going to punish them, even if it doesn’t quite get the right results. The company gives our employees a lot of ground to self-actualize. This culture works really well with a data-driven approach.

This way, you could consider work with data to be a sort of base for the management development of the company, which is in large part built on standardizing and delegating functions. All management decisions at each management level are based on objective data. Employee motivation is also based on objective metrics, which gives a high level of engagement in improving the restaurants.

— Full outsourcing of all work related to data is an interesting approach: Burger King has no BI specialists, and functional managers add data tasks to the implementation queue themselves. How did you manage this?

Yes, all data work is outsourced to a company called Qlever Solutions, and we really don’t have any BI specialist on staff. We do, however, have a strong business analytics department, whose members aren’t technical specialists, but have expertise and a deep understanding of the company’s economic indicators. We also have an information system development manager, who manages changes and sets priorities. This way, all changes get controlled at a relatively high level.

This approach is largely a consequence of the fact that company sets hard limits on the number of salaried employees in any given department. This has led to the creation of a flexible structure that lets us precisely scale resources according to the amount of work, and buy the necessary services using the time&material model.

I have to admit that we didn’t hit the perfect approach right off the bat. We had a rather negative experience working with a very famous company that works with the Qlik platform, which turned into a perfect example of the kind of relationship we didn’t want. Closing off the solution’s source code, and blatant extortion forced us to break with them, and to pay special attention to these aspects in the future.

We now have a relationship full of trust with our current partner. We provide a significant workload for them, and expect honesty and an interest in long-term cooperation. In return, we get loyalty and attractive prices.

— Your partners have access to all of your data. How do you protect against leaks?

I can only repeat the word “trust”. Of course, we have confidentiality agreements signed, but those can never give a 100% guarantee from leaks. This is the company’s choice: we take on a significant risk in order to have more efficient processes.

— You use the Agile approach to ensure that changes are quickly implemented. When did you realize that the company needs it?

I can’t really say that we rigidly adhere to any one specific method. We have a need for fast development and quickly executed pilot projects. When your consultant and developer have great expertise, and especially if they are the same person, then it takes very little time to go from discussing the product to releasing it. On top of that, all of the company’s business departments are included in the solution implementation process, which limits the “game of telephone” effect.

Another important aspect is that we work with the three-layer data model, where already existing indicators can move from model to model. This is important architec-tural know-how from our partner Qlever Solutions, which significantly speeds up the development process.

— At the moment, predictive analytics is the big thing – using mathematical models for predicting revenue, demand, and other business indicators. How relevant is this to your case?

I would say that it’s relevant, but we’re not fanatical about it. Quite often, the marketing aspect of IT does more harm than good, and using fashionable technologies isn't always justified.

We have an example of this in using machine learning to predict revenue. For a long time, we’ve used a rather basic algorithm that gave acceptable results. But it was important to the company that we raise the precision of these calculations since they affected purchasing and staffing. The more precise our predictions got, the better the company could adapt to external situations.

We invited experts in machine learning to improve the algorithm’s quality. After long time of working on it, they managed to improve the quality of predictions by a whole one percent. This demonstrates that the important part is understanding your business and the key factors that affect it. In this case, really simple models often have a level of quality comparable to modern solutions.

— Is your data-related work mostly focused on internal needs, or on providing external services?

We started working on an external block: a mobile app, and a CRM system and services based on it. We started working with delivery and added a pick-up function. The end goal of these is providing personal offers for each guest.

Yes, we need to develop personalized work with guests. We will probably need to analyze the customer journey, application user friendliness, fill out our CRM, and connect external data sources to it, including social networks. This is a serious challenge for the company, since the volume of data is growing exponentially. We will have to solve infrastructure issues and take another look at how we work with data.

We think that these “external” digital services will bring additional revenue to company in the future.

— How has the work you’ve done in the past couple of years affected you professionally?

You could say I’m on the rise. Our shareholders have finished buying a large number of restaurants of a well-known brand. I was invited to come on as the head of IT in this new company. I will also have to implement what we’ve done here there.

We will need to do an enormous amount of prep work, that was already done in Burger King. We can’t do without it, but, because we’ve already done it once, I think we’ll be able to do it much faster this time around. Right now, we’re launching the ERP system, with only half a year allotted to the project. We’re starting work with data sources, taking stock of our current reporting, and, with the experience behind us, getting ready to develop it.

A part of the old management team has stayed on, and a part has come in from outside. I’m curious to see what will come out of this clash of cultures and approaches.

— What types of new digital services can we expect in the restaurant industry in the future?

Client identification is a promising technology, and there are already a few strong algorithms that can help achieve this. There will be a number of projects on the market connected to video analysis.
On top of that, the recent achievements in deep learn-ing let us move onto a new level of working with product and robotizing many processes. Innovation is good, but everyone is keeping track of their money, and as long as human labour will remain cheap, we will continue to use it. Nevertheless, I think we will see fully robotic restaurants in the next five years.

Interviewed by Nikolay Smirnov, CIO.ru editor-in-chief