Coral’s head of BI places bets on Oracle to modernise data use

Profile picture for user ddpreez By Derek du Preez April 7, 2016
Mike Bartholomaei, Head of Business Intelligence at Coral, explains how the company is moving away from a siloed, outdated architecture.

Coral shop
Data sits at the heart of any betting and gambling company. This is becoming event more pertinent as customers of these companies shift their interactions away from in-store betting towards online interactions.

This is also true of Coral, one of the largest betting groups in the UK. However, the company’s head of BI, Mike Bartholomaei, has the difficult job of updating the company’s ageing data architecture for the modern digital world.

This is no small task and will likely be a multi-year project. However, Bartholomaei has already successfully transformed the Coral’s core financial data operations and is now considering how to upgrade the more valuable interactions with the customer.

A central part of this has involved updating Coral’s Oracle BI architecture - the company was one of the first to migrate to Oracle’s 12c database.

He explained in an interview with diginomica that when he joined the company in April 2015, he inherited somewhat of a messy data environment. Bartholomaei said:

We have been using Oracle BI for quite some time, since about 2007, so we were early adopters of it. Oracle databases have been powering our data warehouses since about the same time. When I joined Coral in April 2015 I found the usual hodgepodge of all sorts of other things in there, so technologies that should have been switched off that weren’t switched off.

Very interesting approaches of using all sorts of other stuff to achieve the same goal. A typical environment where people were not usually mature when it came to building data warehouses and BI environments for an enterprise.

Rethinking the approach

Bartholomaei said that his intent and strategy is to “clean all of this up”. However, because of budget constraints, he had to begin by doing that more all less within the existing architecture and with existing resources. He added:

[We are] a typical stingy retail environment, we do not like to spend any money on anything. That was the situation as I found it.

He began by rethinking how the business viewed its BI function. Traditional Coral employees, when needing analytics or insights, spent their time independently making requests and asking for information, without much thought for what the rest of the business was doing. Nothing was aligned and people didn’t have much of a view of what their counterparts were up to.

Bartholomaei began by fixing this. He said:

Basically I brought in a new approach to actually deliver stuff for projects, I removed the contingency that we had on different stakeholders having different agendas and kind of playing each other, and with us in the middle. When you have a team that delivers to many different user communities and they only have their blinkers on their task list - and don’t know what the other guys are doing and don’t care - you can easily be crushed in the middle.

I removed all of that and put it all in a very transparent way, on a tool that is accessible by anyone (Trello), a bit like a portal, and made all my tasks disappear from spreadsheets into there so everybody can see what we are doing all the time. Basically the whole expectation setting is managed by what we actually want to deliver in a much shorter timeframe.

Following this, a sort of perfect storm allowed Bartholomaei and his team to push for a far more modern architecture.

Firstly, Coral hadn’t bought any new hardware to support their data environment since 2007, which allowed Bartholomaei to convince the board that the company needed some new boxes. The existing hardware was bought to cater for two transactional systems, but this had grown to nearly 50, so it was in dire need of updating.

Secondly, Coral’s finance team were at the same time pushing for some new software, which Bartholomaei said was a “crazy idea to introduce another silo”. So instead of doing this, he managed to prove to the finance team that what they needed could be done with the existing Oracle software.

Thirdly, whilst all this was going on, Oracle released is latest version of Oracle Business Intelligence Enterprise Edition (OBIEE). Bartholomaei said:

September I had the board approve the hardware purchase, in October the finance competitor software versus our existing technology decision was made on a proof of concept, in November we demoed this to a group of executives and got a green light to build it.

And while all of this went on Oracle released the latest version of OBIE, then we basically bunched it all together and said while we have new hardware, let’s go to the latest version of Oracle, take our ETL tool and make this the latest version.

This made us one of the first 12c companies around.

Shifting out of the core

Bartholomaei said that this allowed Coral to build a “focused business intelligent

environment for the first time”, as it let people see data from multiple financial and other sources in one place. He said:

The way that the organisation was set up and was consuming data was more like getting an email with a request for an Excel or PDF attachment regularly - and they abused the tool for that purpose. They left the Ferrari in the garage. And if they did get it out of the garage, they never got out of first gear.

Basically what we did was we twisted it around, put everything on this brand spanking new hardware, put the latest versions of the applications in play and produced something that runs a P&L across our 1800 shops, 9 regions and 80 odd areas in less than 20 seconds.

Which is absolutely amazing. To see that was incredible - hundreds of thousands of finance transactional lines, doing all sorts of fantastic things on the fly calculations in seconds. That’s something that no one here thought was possible.

Bartholomaei explained that this finance project was much more than a BI project, as it required changing all sorts of underlying business processes and altering the ways in which data was categorised. He said it allowed the business to control what they were doing.

Bartholomaei added that this has delivered process and time savings for Coral. He said that typically a month end process for finance takes many, many days. Whereas, now all the data is at the fingertips of the finance people and the whole month-end process has been streamlined. He said:

We will be using that now to drive financial performance throughout the whole organisation. It was basically a rebuilding of our trading pack. When you have a retail organisation they usually send out to the local area, or regional managers, all sorts of big sheets about their performance.

It was like a 30MB spreadsheet that took 10 days to produce and was sent round by carrier pigeon. Now we have that on a dashboard that’s available to the end users.

Bartholomaei explained that this finance project was more or less greenfield because nobody had ever really tapped into those data sources. He said that the company’s data sources were “rapidly growing” and that his job is to convert all of that content into a new platform.

I basically have all this sitting in the old versions of the technology and I need to move all of that content over into the new shiny world, ideally giving it a nice little sprucing up. That’s massive. Compared to what we built now, which was nice and self contained.

You have to think about the kind of data we are looking at - think of bet transactions in shops, any bet that is placed in the shop, there are all sorts of permutations around that. These retail shops also have machines in them where you pay the roulette, poker and slots, there is also the data around that.

There is a heavy amount of customer insight to be gained from this data when you combine it all up. I would say I’ve done 10% of the work, the core, now I need to layer around it my whole betting world and the way that these bets are structured.

It’s a business thing


Bartholomaei explained that he is going to take an iterative approach tot his and that every month there will be new things that come into the platform. He said that this is tricky as he is still having to maintain operations for the existing platform, so is still serving the old stuff, whilst building the ‘new world’ at the same time.

But he said that one of the most important things he has done to get this project going is to ensure that BI is no longer an IT function, but rather a business function that serves business users. He said:

Maybe one thing that’s really important to understand is that we have gone through a structural change in that period as well. I have convinced quite a number of people in high places to move my department from an IT function and made sure people understand that we are a business function.

We are not doing this for our own purpose, IT for IT’s sake, we are business facing thing. We have moved the whole department physically down inside the business, so we are now fully embedded.