February 19, 2018
4 minute read
We’re used to being defined by demographic information. From census data to market research, grouping people according to their gender, age bracket, nationality, income and/or occupation is arguably the most common form of segmentation. In marketing, demographic segmentation is a tradition rooted in print media; when assessing the suitability of a publication for promotional purposes, the demographic profile of its readership used to be all there was to go on. The concept behind segmentation is to tailor a marketing message to suit a particular ‘type’ of consumer, but is assuming that everyone of a similar age or gender thinks the same way really the best way to create a campaign?
Today’s online activity provides a wealth of additional data about consumers, with the potential to paint a far more accurate picture of what makes them tick and, more importantly, how they interact with brands and businesses. Here, our Head of Insight and Analysis, Tom Hutchings, explains why it pays for marketers to embrace a new approach to segmentation, and how we put it into action for our clients.
TH: “Because most transactions happen online these days. With data insights, we can see that online behaviour varies hugely, even within the same demographic, so we need to look outside the standard classifications to get a clearer perspective on how people behave. Consumers today expect you to understand their preferences from day one, so we can’t make assumptions if we’re going to grab and hold their attention.”
“Here’s an example – we have a client who is a motorcycle insurance broker and they had a lot of male motorcycle riders as customers, generally within the 35-50 age range. The issue was that relying solely on demographic information didn’t really segment their base that much, they needed another way to dig down into the data.”
“We used extra information about them – how they use their motorcycles, the value they add to their bikes, whether they were repeat customers and so on – to create five segments, which helped us improve both acquisition and churn rate.”
TH: “Absolutely. Many companies don’t have any segmentation in place and are using a broad-brush approach to communications, but we believe every consumer wants some degree of personalisation from their favourite brands. Segmentation can provide this, but the key is getting the balance right.”
“Approaches to both segmentation and personalisation have to acknowledge consumers and their preferences, and can even predict what they don’t know about themselves, but the trick is not to scare them off. The proof is in the results; how segmentation influences engagement, spend patterns and long-term retention. We always try to test any segmentation against a random control set of data to assess its efficacy.”
TH: “Then it’s up to us to be a bit more creative! We can help them gather data via surveys to their clients, loyalty schemes and on or offline transactions. There’s also third-party data and a wealth of valuable open source data out there, such as postcode information – data.gov.uk is an amazing resource.”
TH: “Firstly, we usually hold a workshop with our clients to understand what they want from segmentation and how they plan to use it, which helps steer our methodology. For example, if a client has a very clear idea, we might adopt a business rule or hierarchical approach rather than letting the data itself inform the segments. We’ve worked with many different kinds of clients, from football clubs to the Home Office, retailers to insurers, and their different aims require different approaches.”
“More often than not though, we go down the statistical route. We use factor analysis or principal component analysis to pick up the patterns and trends that shape the segments, and then move on to clustering – defining the centre of each segment and assigning customers to it to see how well they fit. Each segment must be distinct and have its own defining traits and behaviours, so there can be a lot of movement during the process.”
“Once we’ve generated the segments, we profile them to check that they’re what we expected them to be. It’s usually a case of several rounds of segment building, then segment profiling, then doing it again. It’s quite a cycle.”
TH: “During the workshops, it’s useful to hear what clients think their segments will look like, even if it’s just based on anecdotal experience or word of mouth.”
“Sometimes we end up dispelling rumours – a retail client had an idea of who their key fanbase was, and actually, it was quite different to what they were expecting. 80 per cent of their sales were coming from a segment that was much older than they thought it would be, but discovering that allowed them to alter their marketing campaigns accordingly.”
TH: “Once the segments are defined, we need to understand how best to talk to the people within them. The risk is that poor segmentation can offer worse results than no segmentation at all so measurement and testing is key.”
“We do further research to really understand our segments’ motivations and interests beyond what the data says, before creating draft campaigns. Then we use a test and optimise process to refine the campaigns, say, testing a campaign across multiple segments to see whether or not it’s well-received by the one it’s geared towards, and whether or not it resonates well with the other segments. A lot of effort goes into the test strategy, making sure there’s measurability when we communicate with the segment in the right way, that it’s not just a fluke.”
“Segmentation-led marketing is an ongoing process that’s never really finished. Over time, we work with clients on segment-led planning sessions, including the acquisition journey, the in-life journey and the retention journey from each of their segment’s perspectives.”
TH: “It’ll be exciting to see the impact of real-time data on segmentation in terms of joining up the dots of consumer profiling. Usually data on a new customer is placed in an information depository so that we have time to gather all the available historical data about them and their behaviour before assigning them to a segment.”
“Real-time could enable us to place them immediately based on real-time browsing or transactional behaviour. Machine learning could help us work through more data and more variables too. We create a finite number of different workflows, or journeys, for our clients’ segments, but machine learning could probably automate the generation of thousands.”
Segmentation is just one of the ways we can help you unlock the marketing power of your company’s data, contact us today to find out more.