Delivering on data: build it and they will come
There is a lot of hype around ‘smart analytics’ and ‘big data’. But starting small, learning by doing and then waiting for the ‘wow’ factor is how one global business travel agency is going about it. Pamela Whitby reports
“Aah, wow, in all the ten years I’ve been in this job, I never thought this possible.” On the dashboard and in real-time a member of staff at Carlson Wagonlit Travel (CWT) could see all the bookings made by the company around the world.
William El Kaim, Marketing Technology Director, Global Product Innovation team at CWT calls this the ‘wow factor’ and is one of the responses he expects to hear if an innovation project has been successful. “CWT set up an expert innovation team two years ago with the aim of developing a deep understanding of the ‘data’ challenges facing the firm,” he says.
Like all companies, CWT today has vast amounts of data, which it needs to process in order to better understand and improve its business. One of the biggest challenges firms face, El Kaim explains, is having the capability to access the data in real time. When this is possible firms can move from “hindsight to foresight”. However, while this is the main issue there are others too including: compliance with data privacy regulations and storage and analysis of vast amounts of data. The challenges also vary from company to company. CWT is an intermediary which sits between the companies (clients) and their business travellers and the suppliers and here is a lot going on between the two. “Before we would have to wait one or two days to get access to certain data but now we react immediately in order to deliver the perfect trip,” he says. “This is why we needed real-time bidding.”
Accessing the data
Once firms have the capability to access the data, there are two options, says El Kaim:
1. First you have a question and you want information from the data to answer that. In the case of CWT, for example, how many travellers impacton the total spend of a company?
2. Secondly once you have accessed the data, then you may want to “dive into it” says El Kaim to really make some serious business decisions. This is heading into data mining and analytics and is a much more complex process which requires greater organisation and a particular skills set.
According to El Kaim, one risk is that firms get too wrapped up in the tools and process associated with data and smart analytics rather than focusing on the business need. It was for this reason that the innovation team, which operates independently to the IT team - comprises both a highly skilled data scientist and others, like himself, who are tech savvy but also have a sound understanding of all aspects of the business. El Kaim admits, however, that the analytics part of the business “really exploded” when the data scientist came on board.
“Our first project involved grabbing a small amount of one or two year of transactions data relating to all the air tickets we’d sold in that period,” says El Kaim. “The first product created from this initial study was CWT Meeting Optimizer. You just need to enter the number of participants, their location and the meeting dates; the tool then generates a prioritised list of places suggesting the best location for the meeting along with an average cost estimate. “We also filed a patent for the algorithm used to compute the results,” he says.
With that information this small but growing team started to look at what business needs could be met in real-time, whether the tools could answer any unresolved questions and how the existing system could be improved. Things like data security and privacy of people’s data had to be considered along the way.
El Kaim believes strongly in starting small. Identify the business need, and then “build [a pilot] and the rest will follow”. It is about getting people interested, training the core people involved in accessing the data, disseminating the data, identifying the outliers (a statistical marker that varies greatly from others in the sample) and then trying to find the best ‘visualisation’ solution. Then they move on to build from the pilot (short-term proof of concept) a product, which is fully supported and managed by CWT. “We always build something first because we need to be sure we have the data and we can actually do it,” he explains.
In every project there is a business leader who will take ownership of a project once the innovation team has completed and validated its pilot and then work with the newly created ‘innovation council’ (a group of people from all global business and IT lines) to decide how to develop it further. Each part of the business may need something different but there is always an enabler and a clear business need. Some teams may need forecasts, while others may need, for example, to establish a link between online and offline bookings. “Our job is to enable a new system, to train people on it so they can leverage it,” says El Kaim.
CWT organises short one-hour training sessions and, importantly, systematically gets people to work with their own data. “We never make them learn something on a system that they won’t use,” he says.
Some businesses spend millions investing in data and analytics systems for client reporting, for example, but at the expense of investing in internal resources. El Kaim says it is imperative that all stakeholders’ needs – both internally and externally – are catered for. “You should provide the best analytic systems internally as well he says, adding that operating a silo only leads to duplication of data and loss of momentum.
Another common mistake is to think that by having data and a tool you will get the answers to your questions easily. Sometimes, you also need to build the algorithm and make it work.
While developing projects in-house is important for CWT, the company also outsources. “For our dashboard we are currently working in private cloud mode with a Business Analytics tool and a service provider selected and managed by our Global IT department; we wanted to be able to concentrate 100% on the ‘agile’ business needs,” says El Kaim.
• Train your people; make them think about what they need before they choose tools.
• If you do not know the data you are using, look at the results twice before jumping to conclusions.
• Recognise that one size does not fit all.
• Don’t compromise on data (quality, security, and privacy) but be more open on the tools/algorithm to crunch them. Stay agile and keep trying.
• Work in a ‘consulting mode’ with a data scientist when questions are too complex.
• Leverage your security and IT department, and use outsourcing to avoid investing in ‘nascent” technologies.
• Find out what you do not know!
For more insights from CWT join us from January 17-18 at our Smart Analytics Conference New York where William will be speaking.