Could you start off just by telling us a bit about your background and how you came to be working in the area that you’re working in with analytics?
I studied statistics with economics undergraduate level, then I went on to do a Masters in statistics at the LSE part-time whilst I was working at GFK. I worked on a joint project between the LSE and GFK looking at using statistical models to forecast demand for consumer goods. I’ve stayed in the industry since then. I’ve worked across a number of large organisations, and I’m now at Populus.
What’s your current role?
I’m Director of Analytics at Populus. I’ve been at the company for 18 months and I have a team of three including myself – me the Director, one junior analyst and also a data scientist. Our team services the whole of Populus in respect of not just analytics needs, but also helping with survey research as well as internal and external training (we do training for the Market Research Society). Primarily we’re running analytics on the data from our surveys but increasingly we are using the client’s own data in addition to our own survey data. Sometimes a client will send us extracts from their CRM or other internal metrics which they want us to merge and combine with our survey data ready to generate further insights.
What would you say the main challenges that organisations like yours are facing at the moment?
The main challenges really are that companies want results faster. In addition to getting the survey data itself, they also want us to give them the insight that goes with it – they want us to tell them the “so what.” Indeed, sometimes they don’t even want the data itself. Gone are the days where you can supply them a whole load of tables and hope that they would get something useful out of that. Market research companies, when we’re given data, need to be able to quickly understand what the headline is. What’s the main insight? What’s the main finding? Once we’ve identified that then we can dig down further and get into the “why’s” and maybe use some of the other data as evidence of that. That’s one of the main challenges.
The second main challenge is how best to manage the amount of data that clients are increasingly capturing themselves and trying to get that to dovetail or merge with our data. Companies are capturing a lot of data and many of them don’t know what to do with it. They have some sort of systems for capturing it and for holding it but often they don’t really understand how they can use it to understand their customers more. They’re looking for companies such as Populus and others to find the hidden gems within that.
Are clients are expecting you to be making next action type recommendations?
Definitely, yes. It varies from client to client but the people we deal with are always doing the research for a reason. It could be to help with marketing, or perhaps to help with strategic prioritization or investment decisions. It depends on the client, but we see our role as to give our clients evidenced recommendations.
It’s critical that, almost before we start the research project, we understand what the client wants to get out of it. That’s not just in terms of what they want the research to contain, but also how they’re going to use it. Who are the internal stakeholders? What are their main interests? What are the processes that they’re going to undertake as a result of it? For us, to be able to actually help with that is an essential part, it’s not just a “nice to have”. It’s not just producing a nice table or deliver an analytic package. It’s actually making something useful out of that.
How do you manage that process?
When we’re working on a project, either me or the people in my team will form part of the client’s project team. We’re not just an added extra. We’re very much an integral part of the team, so we’re there at the project set-up meeting. Indeed, we’re often there right from the pitch stage, so we explain how we would work with the client, so they get to see us. We’re not just people who work in the background, producing clever stuff but never being seen. We try and get involved as much as possible.
Is there a typical kind of client you tend to work with or a sort of typical project at the areas that come up again and again that people are interested in?
A lot of our clients are interested in targeting, in trying to understand how to reach particular groups of customers and how to understand their requirements. We also work on a lot of project to help clients understand how their brand is perceived or how their media strategy is working out. Some of our clients are in public sector, so then they’re not necessarily competitive. They’re just trying to better understand what it is that they’re doing. Sometimes that’s trying to influence people’s behaviour in some way or it might be just trying to understand how well they’re meeting the customer’s needs. It varies really.
Do you notice any trends in terms of the kinds of things people are asking for? Does a new technique or thing come along and suddenly everybody wants that kind of analysis?
Yes. At the moment lots of people are talking about data science. Another thing that’s come out of a lot of our client feedback is behavioural economics. The theory has been around for a while, but there are special techniques we can apply to data now that incorporate an understanding of behavioural economics.
Apart from that really, the other trends are, obviously, the increasing volume of data that clients hold themselves and a growing requirement for solutions that work with their own data and internal analytics teams.
Have you noticed any other changes in the broader analytic space during your career from when you were doing your degree for example, through to now?
Yes, there’s definitely more emphasis on the “so what?” It’s not enough just to produce an output, it’s actually getting the insight that tells the client “What does this mean for us? What can we do with this?”.
I’ve also noticed that clients are cutting out the people who are just interested in the research and going straight to those who are actually doing something about it, whether that’s the advertising teams or some of the product teams, so as a research agency we’re increasingly getting a bit closer to them.
Also, the other trend is for analytics at a much more granular level. We still do lots of segmentations for clients but increasingly they want to know what happens when they target these people, so that they can actually plan strategies at a very detailed level as opposed to just having a broad view about how to target a particular group of people who have a particular demographic or sort of attitudes.
What are the kinds of tools and techniques that you’re using most frequently these days?
SPSS is still my day to day main tool, but my colleagues use all sorts of different resources especially the data scientists in the team. They’re using a lot of machine learning tools, anything that can help them be able to access stuff faster, to handle bigger datasets and to do more things with them. Greater volumes of data and more sophisticated analytics means that you need faster machines with bigger capabilities, and then you need more support from IT in terms of being able to handle all the data and all the processes that are running.
What is it that you like or find valuable about SPSS specifically?
I do think it is very well designed for what we do in terms of analytics, and it is still very flexible in terms of how you can create data sets. A lot of the algorithms which it uses are very good and they’re still industry standard in many ways. It’s regularly updated so there’s often new things coming into it, I quite like that as well.
When you’re recruiting are you looking for people who are already skilled, experienced SPSS users or people who’ve got a little bit of experience of a lot of things?
I’m not too worried if they don’t have SPSS experience because I can bring them up to speed on that. I’m more interested in seeing whether they have the right attitude. They need to have some level of analytical skill but it’s their ability to understand research data and our client’s data or client context and how they might combine the two which is more important really.
I’m after certainly some sort of software skills definitely, depending on whether it’s junior or more senior level. Obviously the more senior they are, the more I’d expect them to be highly expert in a couple of tools or a couple of software packages. Which software it is is becoming less important because there are so many ways of people picking that up themselves, self-teaching.
We all have to be client facing and so it’s critical that people can talk to internal clients or be able to explain something to an external client. Things like being able to write a professional email are really important. There’s not really much room for people who are brilliant programmers but poor communicators. Those experts and analysts who aren’t able to communicate their outputs clearly, I think those are the sorts of people that will struggle now.
It seems like there are opportunities in analytics for people from all kinds of different backgrounds.
Definitely. It’s quite rare to have a stats degree, amongst the people I’ve worked with. There are some people with a stats background, but they’re by far the minority. Most of the people I’ve worked with are from different backgrounds, but they ended up doing a bit of analytics somewhere along the way and they’ve just enjoyed it and going from that.