We’re promised a lot when it comes to data and the value it can deliver. In our daily lives, we’re all used to seeing and working with large-scale implementations that are indistinguishable from magic.
Take Android’s “time to work” feature. If you have an Android phone, it’ll warn you if the traffic is bad, so you can leave earlier for work. It’s completely automatic – there’s no need to tell it where you live, or where you work. It’ll figure it out, and let you know.
It works by merging your phone’s GPS movement patterns (where does your phone show you are from Monday-Friday? When do you leave, from where, and via what route?) with predictive traffic data, developed in turn by combining GPS data for other local Android phones (where are they going, and how fast are they moving compared to the speed limit?), and comparing this against averages for the area and the time of day.
This complex blend of historical trends, predictive analytics, highly personalised individual data and aggregate information from millions of other devices is distilled into a friendly, well-timed notification that you should consider leaving a little earlier for work today. It’s elegant, and powerful, and works so smoothly that most users don’t even stop to think about the complex processing that sits behind it.
So why can manipulating the infinitely simpler datasets we use in social housing seem so awkward by comparison? If Google can tell me whether I’m going to be late without even knowing where my office is, why can’t my maintenance management system warn me that this quarter’s costs are going to be above budget?
While most software platforms come with a series of predefined reports which cover the most popular metrics, these are rarely able to deliver the insight needed to drive real transformation in business practice and the experience of your customers and tenants. How do we get from the clunky reporting dashboards we’re used to, to this sort of magical data?
We need to ask the right questions
Data is only as powerful as the questions we ask. Computers are very good at giving us the answer to our questions as we phrase them, regardless of what we really mean to ask.
In data analysis, there are four different types of questions:
The more sophisticated the questions you’re able to ask of your data, the more powerful the answers become, and the more value they provide.
For social housing providers, increasing data value often means moving up the hierarchy from descriptive and diagnostic questions (what’s happened and why?) to predictive and prescriptive questions (what’s going to happen and what should we do about it?).
Upgrading the questions we ask
To see how upgrading to more sophisticated questions can help us get more value from data, let’s look at the underlying question behind Google’s “time to work” alerts. What type of question would we need ask to get the information it delivers?
1. Descriptive: “What’s the traffic like in my area right now?”
Asking for a simple status update isn’t enough – the traffic might be heavy, but if it’s always heavy at 8.15am on a Thursday, that doesn’t help us understand whether we need to change our departure time.
2. Diagnostic: “What’s the traffic like in my area compared to the average Thursday morning?”
This gets us a little closer, but still doesn’t really answer the question. If it’s worse, how much worse? What’s the likely outcome for my arrival time?
3. Predictive: “If I set off at the same time as usual, what time will I arrive at work this morning?”
This is much more helpful, but still doesn’t necessarily tell me what to do. If I’m going to be 15 minutes late, but I usually travel at a time when the daily rush in my area is starting to ease off, setting off earlier might mean I hit a different wave of traffic, delaying me further.
4. Prescriptive: “What time do I need to set off today, if I want to arrive at my destination at my usual time?”
This is the version of the question that covers all the bases – what’s happening now, how does that compare to what usually happens, and crucially, how do I need to change my behaviour to achieve my desired outcome?
The extra attention Google have spent on ensuring they fully understand what users want to know is what takes their traffic alerts from a daily irrelevance to a vital tool. They’re not just providing the data they have (“what’s the traffic like”?), but using this to form a prediction of future performance (“what time will I arrive”?), and to recommend the best course of action (“what time should I leave?”).
How do we apply this to social housing?
In social housing, taking the same approach can help us move from asking simple questions that return less useful information to more sophisticated options that deliver deeper insight.
For example, we might want to “upgrade” our analysis of property voids from “how many properties have become void in the past quarter?” to “how many voids can we expect in the next quarter?” and then to “how should we schedule next quarter’s maintenance to minimise the number of voids over the next year?”.
Of course, a side-effect of asking better questions is that often it exposes gaps in our knowledge that we need to fill. Sophisticated questions are good questions because they help us uncover complexities we hadn’t thought about before.
For Google, answering the question “what time should I leave for work this morning” draws data from the Google Maps infrastructure, the GPS system of the user asking the question, and aggregated “big data” from millions of other Android phones. Asking this complex prescriptive question meant processing and using their existing data in completely new ways.
In our sector, prescriptive questions will, hopefully, not require quite such complex data, but it’s very likely that in asking these types of questions we’ll discover new requirements for data we don’t have, or that we don’t store in the right format, or for additional processing of existing data in ways our current systems aren’t capable of.
This means that we might not be able to jump straight to a prescriptive “what should we do” question – we might need to settle for an earlier version, while we improve our data systems to allow us to ask the more complex questions we really want answered. However, if we start with ambitious questions like this, it also helps us set a roadmap for what our data systems need to be able to do, and ensure that we’re developing our capabilities in a way that will provide the insights that will make the biggest difference to performance.
Orchard Data Value Services
At Orchard, we want to open up a discussion with housing providers to understand the questions they’d really like to be able to answer. By helping you to ask better questions, our data experts can help you to develop a data roadmap that will deliver the insights you need to transform your business.
We’d love to hear about the questions that are keeping you awake at night – whether they’re on customer engagement, regulatory compliance, fraud detection, void management, or anything else. While we might not be able to answer them straight away, we believe that asking the right questions is a vital first step in improving the value data delivers both to our organisation and yours.
We’d like to talk about how we can help you reach the “magical” level of data value – what questions would you ask?
To find out more about Orchard Data Value, click here.