3. Customer segmentation and profiling
Customer segmentation is the practice of dividing a customer base into groups of individuals that are similar in specific ways relevant to marketing, such as age, gender, interests, spending habits, and so on.
Using segmentation allows companies to target groups effectively, and allocate marketing resources to best effect.
The UK Cabinet Office: "There is a significant opportunity for local government to improve its performance through more effective use of customer intelligence. Customer information is an underutilised resource at the moment and its value as an ‘asset’ that can be used to drive improvements to service and performance is not sufficiently recognised by local government. There is increasing pressure for authorities to demonstrate that they can pro-actively identify customer needs and demonstrate how those needs are being met. Measuring performance in terms of impact on customers and their views will become an increasingly important feature of performance management. It is unlikely that the council of the future will be regarded as performing well unless it is using customer information to drive and measure performance. Furthermore, low or reduced grant funding for local authorities together with likely council tax caps over the medium term requires authorities to think about prioritising services and methods of service delivery. It will be hard to do this effectively and even harder to justify without reference to hard data about customers’ use of services and their preferences."
Profile characteristics
Citizens may be segmented by a myriad of different profile characteristics including: age, ethnicity, sexual orientation, ability/disability, gender, level of affluence and state of health.
Difference with profiling
Segmentation is often used in conjunction with customer profiling, but there are areas of difference. For instance, profiles are not suitable for identifying certain population segments: people with disabilities are usually split between multiple neighbourhood profiles. Likewise gender segmentation is rarely associated with neighbourhood. For other factors such as age and ethnicity, composite profiles can only support broad generalities.
Who are your customers?
You should have some Customer Data: information about who your customers are. This will be either:
- Explicit – records of who has or is currently using the service e.g. from a CRM system or other records
- Implicit – knowledge of staff or partners who are dealing with customers in delivering the service.
A useful audit tool for what data you have is http://www.westmidlandsiep.gov.uk/download.php?did=1181. Even if this is a new service, you should be able to make an educated guess on who may use it based on other existing services where the customer group is likely to be similar.
If you have no knowledge or it is sketchy, you can always carry out market research – probably via qualitative techniques rather than quantitative so that the service can be explained before people decide whether it is for them.
A guide to customer profiling produced by the Smart Cities Regional academic network is http://www.smartcities.info/files/Smart_Cities_Research_Brief_Customer_p... There is also a UK guide http://www.cabinetoffice.gov.uk/public_service_reform/innovation/segment...
Explicit
This can equally be done with Population Data: Census style socio-economic information about small areas e.g http://www.statistics.gov.uk/hub/population/index.html If you aggregate customer records to the same area as the census data (in England this would be Super Output Area) you can then produce a similar table of the percentages of customers who have the various characteristics in the census fields.
If your implicit information suggests this customer group also use other services which you have records for your can use the sort of data matching described above to create a profile which says “70% of people who use service A also use service B”.
In addition to statistics and data which are available locally or nationally, there may be further explicit information about customers in academic analysis based on primary research. Although this will probably have been done with customers in areas other than your own, the patterns of behaviour are likely to be equally applicable to people in your location.
Implicit
Implicit information in the minds of staff can be turned into a more formal profile by using workshops and/or questionnaires to ask staff what they think customers are like against key criteria taken either from census style information. Customer Profiles, or in terms of other services used where you do have specific address based customer data.
Even if you have explicit customer information, you should ask staff for their views of customer characteristics as this will help round out the picture of:
- Who they are
- Why they use the service
- How they use the service
- What their other needs are
- What other services they need
- Who could or should use the service but doesn’t
You may also want to use the same techniques to ask questions of the customers themselves as part of the Customer Co-Design process. A useful guide is www.demos.co.uk/files/File/CollabWeb.pdf If you can contact existing clients or have an idea of the sort of people who are potential clients, you can also organise a focus group to find out more about their needs and behaviours.
A guide to understanding demand for services is http://www.smartcities.info/files/Smart_Cities_Research_Brief_Measuring_...
What do you already know?
You need to identify all the information you have about your customers. In addition to the information above you can identify other data. Remember that your customers may also be customers of other parts of the organisation or partner organisations.
For example, you may want to encourage new parents take parenting classes to help them look after their babies. You may only have basic information about them as new parents – from the health service or birth registrations. However they may have older children and those children’s schools may have information about them and their families. The parents may have other contact with the health sector or adult social services – for example for disability. The parents may be in receipt of benefits.
It is important that you do not try and cross reference too much information, but stick to information that is relevant to the business question. This is for two reasons – because you do not want to make the analysis to complex and time and resource consuming, but more importantly to avoid trespassing on people’s privacy. While most people will be happy for you to carry out anonymised profiling work if it means they get a better service, and there are circumstances where comparing personal information is justified to prevent crime, in most circumstances this should be limited to what is needed.
You need to consider privacy and data protection issues carefully when comparing information. Where possible you should make it clear you will be using information for this reason when you collect it and gain consent, which will prevent any problems.
The key thing with all information to be used in profiling is that it should have a geospatial reference and that it should be as “clean” as possible. Records that have lots of incomplete or inaccurate data should be avoided. If you are planning on doing a lot of data analysis (and not just customer profiling) as an organisation you should have good data quality processes and training in place.
There are companies and software that will carry out bulk data matching and identify those records which are incomplete or where possible matches have to be checked (is John Smith the same person as J.S. Smith and is The Laurels, Coronation Street the same as 12 Coronation Street). Doing this work can then be used to clean up the original databases as well as to improve the customer profiling work.
Not all data will be able to be matched precisely, even with the work described above. There is a technique called “Fuzzy mapping”. This takes real customer data and plots it on a graph against views or other characteristics. The areas where there is largest clustering of points together shows the most likely views of that customer group and this can then be attributed to them as an expected characteristic.
What more can you find out?
As mentioned above, there may be information available in other parts of your organisation or in external organisations which is relevant.
Other tiers of government will often have data sets which are geospatially referenced and which can be downloaded or requested from them. The EU directive on re-use of public sector information http://ec.europa.eu/information_society/policy/psi/rules/eu/index_en.htm... a good lever to persuade them to share this – especially if you can then reciprocate with your data or analysis. A list of data sets for each country has been developed.
There are also commercially available data sets and companies who have carried out research and/or data matching. It may be possible to purchase this or commission those companies to do some research or analysis for you.
Esd toolkit in the UK has the Local Government Business Model http://doc.esd.org.uk/#/?tab=data-lgbmwhich links to the Local Government Service List http://doc.esd.org.uk/lgsl/3.07.htmlto show which groups access which services. Some services have already been profiled by local authorities against MOSAIC customer types so you can see what a typical access pattern by those groups are.