easyJet on-board with Tableau to improve BI reporting

Profile picture for user ddpreez By Derek du Preez July 6, 2015
Summary:
Paul Chapman, easyJet's BI lead, explains how the airline is diving deeper into its data with Tableau.

Easyjet in flight
If you've done any sort of domestic travelling within the UK and across the rest of Europe, you no doubt will have come across easyJet's brightly coloured orange jets. The budget airline is responsible for carrying 65 million passengers a year across hundreds of different routes in and around mainland Europe and makes over £500 million a year in profit. It's incredibly popular.

However, not too long ago, only three years ago in fact, the airline was floundering through the use of Excel for BI reporting and visualisation. Paul Chapman, easyJet's BI CoE Manager, admits that just 36 months ago the company was a “very lean” organisation that hadn't invested a huge amount in technology. He says:

A lot of things were Excel based. Actually a lot of data and reporting would be outsourced to third parties, where we would provide them with our data and they would provide us with visualisations and dashboards through a web interface. The problem with that is that it is actually quite slow if you want to change things and it's quite expensive.

There are limitations with Excel, which I'm sure you will be aware. There are a million rows of data that you can use, which meant that if you are looking at 65 million flights, you are having to aggregate that. You are not getting the whole story.

So rather than looking at an individual flight that took off just now from Luton to Ibiza, what we would have to look at is Luton to Ibiza as a route that might have flown 80 times in the month of July – and you would get one number for that. You wouldn't understand all of the detail.

It's a story that we hear time and time again. A large, successful organisation, with a strong brand and a good position in the market still using pretty archaic tools to try and understand how its business is running. Whilst many recognise the value of data, analytics and visualisation, a significant chunk of organisations don't know where to start or how to tackle the deluge.

And that's understandable. Data is often messy, it's locked in complicated spreadsheets and it is guarded fiercely by a few people that consider themselves lone-knights that know best. Data projects require a lot of effort, cultural change and it's often hard to put together an ROI.

However, easyJet has taken on the task and two years ago began working with Tableau to improve insights into its business, to get casual users asking queries of data and to make data a business function that can help boost revenue streams. It is also beginning to use Alteryx to organise, blend and model its data. Chapman explains:

So about two years ago we started with a really small proof-of-concept, we had two desktop licenses and we had a server literally on a desktop machine under our desk. From that we went and engaged with Deloitte and they helped us develop an early BI strategy. We also partnered with The Information Lab, which helped us with our Tableau deployment.

tableau logo
We gave Tableau to a dozen of our analysts with zero expectation and asked them to see what they could do, to see if they could find anything interesting. That went really well and we moved from PoC to pilot phase.

Speaking at Tableau's user conference in London this week, Chapman says that easyJet started taking Tableau seriously as product when he presented the tool to the board. He explains that at the end of the presentation, easyJet's CEO told him that the visualisations looked great, but asked when Chapman could start doing those with easyJet's data. Chapman told the CEO that the visualisation she had just seen were in fact already using easyJet's data. He says:

I was able to conclude that session by saying that it was our own data. What we have been able to do is go from a single number at a total level for the whole of easyJet and drill all the way down to flight level. That was the moment the business realised that we had some traction around what we were doing.

A centre of excellence

easyJet now has 41 desktop users and has a 16 core enterprise-wide server. It has also created a BI centre of excellence that operates as a business function, rather than out of IT. Chapman says that the main benefits of introducing Tableau for rapid visualisation is that it has improved the work of the company's analysts. He says:

One of the other things that our CEO asked us was what she was going to see that would be really different when we started using Tableau. I told her that fundamentally, as a CEO, she was not going to see a huge amount of difference. Because currently the board asks a question that goes down the line, where someone will do the work, then it comes back up the line and you get the answer.

If you ask a follow up question it will go back down the line again and comes back up and you get the answer. Whatever question you ask, there may be a lag in getting it to you, but you will get the answer.

But we used the swan analogy, where it might look pretty and you get a beautiful result, but under the water the legs are going crazy. And that's the original question [being asked], where you might build something that isn't sustainable because you think it's a one off question. Then you ask a follow up and you have to rebuild the whole thing. So what's going to be different is going to mainly be seen by our analysts.

Chapman said that he was trying to fix what he saw as a 'typical analyst' function at easyJet (which he admits

Data center management DCIM businessman iPad © everything possible - Fotolia.com
is still a work in progress) – whereby each department had one analyst that knew how everything worked, but when they went on holiday or left the business created an “absolute nightmare”; where departments weren't sharing information in each other; and where analysts weren't able to present data to the board that they felt confident in, which meant that executives were still basing their decisions on gut instinct.

Chapman's BI centre of excellence now acts as a link between business and IT and is a business function that reports into the CFO. He says:

Before a business user would go to IT and ask for a thing to be built, but that pipeline was never really managed. That now comes through us and we prioritise all the work that comes in, can understand what is being asked, how we can provide it in a better way.

We are enabling self-service, enhancing ways of working, being able to ask a question very quickly and easily, we run Tableau workshops, we have an internal user group. All of this helps us to have consistent processes, increased automation, better data analysis and then ultimately having this pan-easyJet tool suite, so no matter which department you are in, you've got this tool there.

However, Chapman also admits that the journey hasn't been completely stress-free. One of the main issues he faced when creating a BI centre of excellence was convincing the previous IT and BI teams that this needed to be a business function and could be a function that could do a better job, safely.

That paints a really pretty picture and makes it all look great, but obviously it's not always like that. We are not perfect in anyway. There is a lot of engagement that we have to do with IT. We are a business function bringing in a software tool, which was a brand new concept. The business had never done that before.

There are a lot of partnerships that you have to make, especially with IT. And we didn't do a very good job of that at the start to be honest. We were treading on toes.

Use cases

easyjet images
Chapman gave some examples of use cases within easyJet, which highlight the growing success of Tableau. The first of which is allocated seating. For those of you who have been flying with easyJet for a few years will remember that not too long ago the airline didn't allow for passengers to choose their seats. This resulted in a “free for all”, according to Chapman, and a poor customer experience.

We brought allocated seating in a few years ago because we thought it would be a better experience and we thought it would drive some revenue. When we launched allocated seating it was a very simple model – we had three price points and three seating types. The very front row was extra leg room, which was £12 to sit in that seat. The next three or four rows were called up-front and we charged £8 for those. Any other seat would cost you £3. If you chose not to pay an algorithm would just allocate you a seat on the aircraft.

When we started doing the analysis on this, we got three numbers. We could compare that against the total budget for easyJet, but that wasn't really telling the whole story. So what we needed to do was dive into that and compare it on a flight level basis.

Using Tableau, what easyJet can do now is compare each airport and compare each route from that airport on a granular level, where it can easily visualise which flights are selling what seats and where. So, for example, Chapman can now confirm that a five hour flight from Manchester sells lots of high value seats compared to a 20 minute flight, which only sells a few seats near the front.

Chapman says that this will help easyJet to get more out of its seat allocation strategy.

Where this helps is to move away from fixed pricing and move to a more dynamic model, where we can charge for the duration of the flight and on what people are doing. This just wasn't available to us before.

The other big impact use case has been seen with the company's daily reports, where prior to the use of

© Paul Fleet - Fotolia.com
Tableau, they didn't provide any sort of granular detail. Chapman says:

Back in the day that was an Excel report that went out to our main board and each day we got a single number for bookings yesterday, revenue yesterday, seats yesterday, all compared to the day before. It was a single number. What we are able to do is extrapolate that and go down to the individual booking level, produce the last seven days of understanding when that booking had been made compared to last year, compare it geographically and then also when those bookings are actually flying.