We build segmentations and models that indicate likely characteristics about individuals, households and postcodes. We also hold a small range of actual attributes about some individuals, such as date of birth, house or motor insurance renewal month, or information from the County Court Judgment (CCJ) Register.
These segmentations, actual attributes and models are held on our marketing database against UK adult names and addresses, which have been obtained from, our third party data partners, public data sources and Experian credit bureau data (see The personal data we obtain). These provide insight to our clients to enable them to:
- Select potential new customers where the product or service being offered is more likely to be of interest to them, meaning that you receive marketing that is likely to be more relevant.
- Enrich their client's existing customer data to provide additional insight to inform their product and marketing strategies, which enables them to communicate with you more relevantly and effectively.
- Identify areas where public sector support should be made available to the population, for example to decide on the best location for additional health services or a new fire station, or to allocate resources appropriately to improve adult literacy levels.
Dividing things up - "Segmentation"
Nearly every organisation in the UK uses segmentation techniques in some form.
Segmentation involves dividing large sets of people, households or areas into smaller groups, or 'segments', that are likely to have similar characteristics. When it comes to marketing, segmentation aims to match groups of people with common needs or likely interests in similar products and services.
We build several segmentation products and the most widely used is a product called Mosaic.
Mosaic works by taking data about individuals and households and aggregate data (for example for postcodes, towns, or regions) from a variety of sources, such as modelled data from our marketing database, publicly available data such as information from Companies House and national statistics databases such as the Office of National Statistics Census 2011. Using statistical techniques, we use this data to create groupings of people who are likely to share similar demographics, lifestyles or behaviours.
Those groupings or segments are then assigned to every household in the UK to indicate likely characteristics of these households or a particular geographical area.
View the Mosaic type for a UK postcode
A brand or organisation might use these groupings to gain a greater understanding of their existing customers, so they can communicate with them better, or find potential new customers like those they already have relationships with.
Public service organisations use segmentation to help direct resources to those people in most need. For example, we worked with the National Literacy Trust (NLT) to use segmentation to identify those areas of the UK with the highest levels of literacy problems. By working with us, the NLT were able to accurately identify where they could most appropriately target their support and improve literacy levels. Another example is our work with StepChange Debt Charity where we used segmentation to identify the areas in the UK - right down to UK constituency and ward level - where they could help the most people struggling with debt problems. For further examples of how our marketing products and services are used by clients in the public sector see the Who uses our services page.
Below you can find out how we build a segmentation like Mosaic.
Segmentation in action – an example
Imagine that from a market research survey we have data on some older consumers and some younger ones, some of whom have families and others who don’t. This data is "anonymised" meaning that no individual can be identified.
The survey tells us two things about these people – their age and the presence of children in the household. Based on this information, analytics puts these consumers into groups (which we often refer to as "segments") based on how likely they are to share similar demographics, lifestyles and behaviours – in this case, age and children.
We can label these segments in a way that's meaningful and representative of the group as a whole (appreciating that all members of the group will not have identical characteristics). So, we might end up with segments made up of younger singles in "Student Days", families with young children in "Middle Aged Families" and older couples in "Retired Couples".
If we know the likely age band and family structure of households across the UK, we can attach these segment codes and labels to them, which organisations then use to understand the likely characteristics of their customers, prospects or a geographical area such as a store catchment or a local authority area.
This is a simple example. Segments can include individuals (customers and prospective customers), households, businesses, physical outlets, places or products. And, segments can be based on a higher number of the similarities they share.
When we split people into different groups based on data about them, we are carrying out a profiling activity. See the section on What is “data profiling”? further below for more details.
Our segmentation products
We build a number of segmentations focused on different themes, but all using the same general principles as we use to create Mosaic.
See below for a full list of all the segmentations that we build.
Mosaic UK. A consumer classification which provides an understanding of the likely demographics, lifestyles, purchasing behavior, technology adoption, communication channel preferences and location of all individuals and households in the UK.
Mosaic Digital. Provides insight into the digital lives of UK consumers by classifying them according to their likely attitude to new technology, device ownership and online usage.
Mosaic Shopper Segments. A consumer classification which provides insight into the likely shopping habits, preferences and behaviours of UK consumers of retailers and brands, identifying how to engage with consumers across different channels.
Mosaic Global. A consistent consumer classification for over 25 countries around the world based on a simple proposition that the world's cities share common patterns of residential segregation.
Financial Strategy Segments. A consumer classification used primarily by financial services organisations, it classifies the UK population based on a range of data inputs into a series of segments with similar socio-economic and demographic characteristics but also based on their likely financial product holdings and behaviours. The models of likely financial product holdings and behaviour used in building the segmentation are based on market research data from YouGov where data has been anonymised and no individual can be identified. No behavioural information from our credit bureau data is used. Financial Strategy Segments enables clients to better understand their existing or prospective customers, specifically from a financial lifestyle perspective, with the aim of achieving more positive marketing outcomes for the consumer and the organisation.
Creating models and making predictions
We use something called "modelled data" to make predictions about the likely characteristics of individuals, households and geographic locations in the UK.
The stages of modelling
There are two stages to modelling:
Building the model: we look at the personal data we have in our Marketing Services business together with data from other sources such as geographical datasets, publicly available records, consumer survey information, aggregated consumer panels and market research surveys and apply analytics to create a set of "rules" or models which allow us to make a reasonable estimate of the things we don't know.
For more information about where our data comes from, see The Data We Obtain page.
Applying the model: We then apply these estimates to a larger set of data to make predictions about how the people in that group might behave. Let's say a holiday company in Newcastle wanted to promote a weekend holiday to Paris. We'd apply our model to the Newcastle area to estimate which individuals or households in that area are most likely to be interested in a short break. Or, if a family-focused Italian restaurant chain was considering where to open a new branch, we'd help them find an area where there were lots of families who like to eat out and who'd be interested in the opening.
Of course, no model can be 100% certain. Our modelled data simply provides insights to our clients about which products or services might be most relevant to the people they're communicating with, or which services are most relevant to the area in which they reside.
When we create a model about an individual, we are carrying out a profiling activity. See the section on What is “data profiling”? further below for more details.
See what predictions we make about you
You can see what your Mosaic profile looks like here. These are general predictions based on the likely characteristics of people in your area, so some things will be right, and some will not!
To see a full list of the models we build, click here.
You can also ask us for a copy of all the personal data we hold about you. This full information request is known as a Data Subject Access Request (DSAR). To request a copy of all the personal data that we hold about you, please visit Experian's Data Access Request page or contact our Customer Services Team.
What is "data profiling"?
When an organisation splits people into different groups based on data about them, it is carrying out a profiling activity. We create "profiles" using combinations of our segmentations and models which best describe the likely characteristics of groups of individuals, households or geographic areas.
These “profiles” allow organisations to make sure their marketing is being sent to people who are most likely to be interested in it. For example, a clothing store might divide its mailing list into customers with children, and those without, so they can decide who is most likely to want to receive their children’s clothing catalogue. Public service organisations also use profiling to make sure their services are supporting the areas that are most in need. For example, a council can use profiling to see which of their local residents are most likely to find targeted mobile library services useful, to make sure they’re going to the right places.
Organisations have been creating profiles since the birth of marketing. It is a well-used, suitable marketing technique providing it complies with data protection rules.
See below some examples of "profiling" in marketing:
Examples of how profiling can be used in marketing
Profiling takes many forms, but every business or organisation will, at some level, use profiling to inform their marketing activities:
- Albert buys a cake from the local bakery once a week. When the bakery has an offer on cakes, it’s likely that Albert will be more interested in the offer than Geoff, who has never bought a cake.
- Laura is opening a new restaurant in Wigan. Profiling will help her to advertise her new store to people who live close enough to visit, and make sure she isn't investing her resources in advertising to people who live 50 miles away and will never visit.
- Brenda lives in a fifteenth floor flat. She doesn't have a garden, so she’s not interested in marketing about lawnmowers. We can help make sure she doesn’t receive irrelevant messages, which is good for her, and for the lawnmower company too!
- An online fashion retailer can use profiling to better understand the people visiting their website. For example, they might find out that 20% of visitors are likely to be female, aged 25 to 34, and interested in high-end fashion. The retailer can make sure they’re offering ads which display the right clothes and accessories to appeal to that audience.
- A local authority can use profiling to identify those areas with the highest levels of literacy problems, so they can use their resource as effectively as possible and improve literacy levels.
Using profiling responsibly
Whilst "profiling" is sound business practice, in certain contexts advertising and marketing which has been informed by profiling could lead to poor outcomes for an individual if carried out irresponsibly. Examples might be: where an individual's wishes to not be marketed to have not been respected; or where the marketing is conducted to minority groups or vulnerable adults, for example someone in financial difficulty who is continually served ads for on-line gambling; or where the profiling is being used for fraudulent purposes.
When we create a segmentation like Mosaic or a model like household composition, we don’t make any decisions based on that information. It’s our clients who make those decisions. However, we expect our clients to use our segmentation data and models for responsible marketing purposes. We have controls over the kinds of organisations that we do business with. For example, we don't supply marketing services to organisations for the purposes of promoting pay day loans or tobacco products. In fact, some of our products and services, such as our marketing suppression data, help our clients to ensure their use of profiling is responsible.
Provided that the profiling is carried out responsibly, it's good for the consumer, good for organisations and good for society.