Expert insight – Emma Brooker, Customer Insight Manager, L&Q

What’s your background? How did you come to be working in analytics?

I suppose accidentally really. I studied politics at university, then I happened upon a short-term contract for a market research company doing data entry and progressed from there. I found the work interesting and enjoyed the projects I worked on. I moved to L&Q six years ago and have progressed up to where I am now, running the customer insight team.

Did you study statistics at university?

I had done bits and pieces of analysis as part of my degree but it’s mostly been a case of learning on the job with the odd bit of training on top of that. I certainly didn’t come out of university with a statistics degree or anything like that.

What does your current role involve?

At the moment, I’m managing a team who do customer insight and policy. I’ve got six analysts. Three of them are policy people. Three of them are on the customer insight side. We are a central point for people who want to know a bit more about the business beyond the straightforward data and reporting that they have access to.

Can you give some examples of the typical projects that you work on?

We conduct analysis of resident satisfaction data where we might look at satisfaction results in more detail to try and understand where are we seeing differences, perhaps by local area, by customer demographic groups, by the contractors who carried out our repairs, that kind of thing. We also look at trends in arrears levels to try and identify what kinds of people are falling into arrears. Another example might be examining trends in maintenance to identify whether we can see any patterns in the types of work that are needed, for example in different areas, or in different ages of stock, those kind of things.

What would you say are the main challenges that face providers like yourself these days?

The big thing for us is the change in people that we’re housing. Increasingly, the people coming in are vulnerable – people who have been homeless, people coming out of the care system, that kind of thing. That brings with it challenges around understanding who’s moving in, where are they coming from, what are their circumstances so that we can then make sure we’re tailoring early intervention programs where they’re most needed.

There’s also a challenge around the business wanting to transform and to do more in terms of perhaps moving services online but trying to balance that with the needs of our customers.

A specific analytics challenge is data quality. The more analytics we want to do, the more we run up against barriers of missing or inaccurate data. Some of it comes from people, customers not wanting to tell us about themselves, some of it comes from stock data that wasn’t properly captured at the time that we took over the stock. Increasingly we’re looking at issues around fire safety, asbestos and so on, so accurate stock data is really important and we’re suddenly finding that we don’t have that in some places.

How does the use of analytics help address these challenges?

When people move in they have to fill in several forms asking for information about where they use to live, their household composition, why they’ve left their previous home and so on. We can take all of that information in and use it to inform our analytics and point them towards the services we think will best meet their needs.

For example, if somebody looks like they’re at a high risk of falling into arrears and not being able to sustain their tenancy, we can get them straight in touch with our financial inclusion team so that they can work with them. We’re not just using data to identify these people but also to develop the intervention that we hope will help them.

Similarly, we look at stock quality alongside the feedback we get from our repairs contractors and use that to try and better plan our maintenance works based on where maintenance is likely to be most needed, rather than just a rolling program of maintenance, postcode by postcode.

We’re doing lots of work on understanding people’s payment behaviour and their contact behaviour more generally to learn more about why people are contacting us, and what can we then do to improve either the information on our website or the information given to people when they move in.

Do you see other opportunities for how you might be able to use data and analytics to improve the services you offer?

At the moment, our data is very fragmented and we’ve got different data sources that tell us slightly different things. Some of it’s on spreadsheets, some of it’s in databases. Trying to get a rounded view of a particular service area or a customer is really difficult. We’ve got a big project going on at the moment to put all of that data in one place, a data lake which combines different data sources.

One of the big challenges we always have is to do with maintenance – is it the unit that drives maintenance demand or is it the person living in the unit? If you have somebody who always feels cold and you put them in a new house, they’re going to ring you and tell you they’re cold regardless of the quality of that house. Equally, if you’ve got a really drafty, cold house and you put somebody who’s perhaps relatively well off, they’re still going to be spending a lot of money trying to heat that house. Which is it that drives the demand that we see? Because of how our data is structured and stored, that’s a really difficult question to answer because you have to go through three or four steps to join everything together and get into one place.

Having this data lake will start to make those processes easier. We should be able to start to get into some of those things around better understanding of demand. Where’s it happening? Who’s asking for these things? Then, working with the business to say, “This is what other data is showing us. What does your local knowledge say?”

Predictive maintenance is another thing – if we identify that the boiler’s gone in five units in a block, let’s go out to the others and just do a pre-emptive check to make sure they’re okay. That’s the kind of thing that’s beyond our capabilities at the moment because the data’s all over the place. Going forward, that’s the sort of thing that we want to be able to get to. It sounds simple but the reports might have been taken by different agents, perhaps one of the calls went to a local neighborhood officer. We want to be able to linking all of that stuff up together so we can go back to people and say, “We know there’s an issue here.”

The other thing that’s increasingly coming through, after the Grenfell fire, is the importance of listening to the customer voice. We did a big exercise looking at queries where people had mentioned the word fire. Essentially, we had to manually read them all, understand what each query was about and try and categorise them ourselves. What we’d like to be able to do is use text analytics much more intelligently to be able to do a lot of that heavy lifting for us and to start to pull out the seeds without having to have one person reading all those and relying on them spot a pattern. Are we suddenly getting a spike in calls about something? Are we getting complaints that are starting to look at a different subject?

What would you say has been your favourite analytics project or a project that you really enjoyed or found particularly valuable and why?

We looked at people falling into arrears to determine whether we could predict who was going to fall into arrears based on certain criteria. We identified half a dozen characteristics that were highly predictive of someone falling into arrears in the early part of their tenancy. The model was almost 100% accurate in terms of predicting the red, amber, or green risk people and people got different interventions accordingly. If they were predicted as being falling into a red category, they got three or four phone calls in the first week to check on them. If they were in the amber category, they got a phone call after two weeks. If they were in the green category, they didn’t get any intervention at all.

However, it turned out that actually the people who we were trying to have these conversations with actually ended up in a worse situation than the people in the control group, who we didn’t do anything different than we would’ve done otherwise.

The analytics side worked perfectly. What didn’t work with what we decided is to do with that data. It comes back to this: the business needs to understand that while data is incredibly powerful, it can’t tell you what to do. You need to have people in your business open to saying, “Okay, what are we going to do as a result of what this data has told us?” rather than expecting you to say, “This statistic means we should do this thing.” That’s not quite how it works. Some people forget that you can have armies of people tapping away at spreadsheets and using SPSS to come up with all sorts of information, but if you haven’t got a way of translating that into some meaningful action it’s worthless.

What we’re finding at L&Q is that there’s an increasing demand for frontline staff to have a much better understanding and knowledge of data. They’re being expected to understand that, to manipulate data, use reporting tools when they were brought in to be a housing manager, to deal with antisocial behaviour cases. Their skill set is not in Excel or that kind of thing. We’ve got people who are frontline housing staff, or call centre agents, or maintenance operatives who are increasingly being fed this data but have never really had any need to understand or use it in the past, yet we’re suddenly expecting them to become quite data savvy.

As a business, if you want people to start to be more data-driven and using insight and taking action off the back of it, you need to invest in your staff and give them the training and the tools to be able to do that, whether that’s systems that are more user-friendly or whether it’s the basics of simple Excel.

Have you noticed any other changes in the analytics space over the course of your career?

The big obvious one for me is just stuff has become much more accessible to people more easily. Where ten years or so ago, getting data into any nice visual format for people was a real struggle. You had to really put a lot of work into making a report look nice and easy for people to read. Now, with things like BusinessObjects, Tableau, Power BI, all of those things, you plug them in and suddenly you can get graphs, pictures, maps. Visualising your data is so much easier.

There’s also a lot of people thinking that big data is the solution to everybody’s problems and not realizing that it’s not necessarily the appropriate thing for us. If you’re Coca-Cola or Unilever, selling millions of units of things every day, you need to be able to process that and come up with trends and see why you’re selling more in Lewisham than you are in Southwark. For us, it’s not quite the same process. People get a bit hung up on thinking that we’re behind the curve because we’re not doing these things. Actually, we don’t need to do that and we don’t need to go off and spend thousands of pounds on a new server to run this data stream when we don’t have that kind of data.

Do you have any words of advice, maybe, for a recent graduate who might be thinking about analytics or indeed for somebody who isn’t considering analytics and perhaps why they should?

The first thing I would say is that it’s probably not as scary as it sounds. It’s very easy to be dismayed when you see the jargon that surrounds analytics and statistics. If you’ve got the soft skills, if you’re inquisitive and you want to understand the detail of things, you can apply that to analytics very easily.

You don’t have to be a maths genius to be able to do it. Something like SPSS is so easy to use. As long as you understand the basic concepts, it doesn’t take a degree or doctorate to be able to get the best out of it. The real skill is the ability to not only do the analysis and use those tools but to communicate it successfully. To be able, in relatively plain English, to explain to somebody the steps you’ve gone through to a position, a result, a number, a statistic, and then to help them understand what that actually means for them on a day-to-day basis.

 

Emma Brooker is Customer Insight Manager at L&Q Housing