I'm head of consumer insight at an electrical goods retailer. We recently had a behavioural segmentation framework developed, including a range of both behavioural as well as store-related attributes. As the store network has changed considerably over the past 12 months, is it appropriate to question the validity of this model? Is there actually a right and wrong way to develop such a model?
Richard Lees replies: Segmentation models are designed for different purposes and there is not really a 'one-size-fits-all' approach.
Generally speaking, the logical framework for a segmentation is to start using behavioural attributes that you can influence, such as spend, range, frequency and payment type. This creates segments of customers according to their value to the business, and appropriate treatments or value propositions can be set up for them.
As these segments become apparent to the business, they will provide a benchmark for measuring business performance. You will want their definition to remain as static as possible, except for movement between segments.
Introducing variables such as stores, demographics and products contradict this requirement as they vary over time and you cannot influence them.
People get older, they get married, the business introduces new product ranges and stores open.
For consistency in your model, overlay these variables on to the behavioural segmentation framework and use them as filters, not behavioural segmentation variables.
We are a large UK courier company looking to expand into Europe. We would like to support this with a direct marketing campaign, but need some advice on how we should go about this. What are the problems we might experience and should we be looking for expert companies in each country?
Gary Selby replies: Adopt the methodology that you would use in the UK, but refine it accordingly to the requirements of each country.
Initially, firm up your marketing objectives, bearing in mind that you are taking a proven service and launching it into a marketplace by aggressive acquisition.
Identifying your target market is crucial. New customers should be carefully selected so that the cost of acquisition is outweighed by lifetime value, while minimising the risk of churn and credit problems. Profile customers from your existing client base to help you.
Historically, this kind of targeting is difficult to maintain across national boundaries due to a lack of available country-specific customer data. However, proven analytical and data techniques have been developed that allow modelling conducted in one country to be re-engineered in accordance with the particular market demographics of another country.
This work must be undertaken by a centralised, pan-European specialist capable of viewing it from a multi -national perspective.
Identifying your prospects will help you decide whether to license data from a pan-European data owner such as D&B or Mardev, or alternatively from in-country specialists.
Prospect and customer data can be centralised in a European marketing database, which will facilitate communications for each country.
The specialist you select plays a vital role in delivering the message.
Aspects such as language, cultural differences, call to action and data protection can vary greatly between nations and require special attention.
Remember to monitor the results of your campaign and learn from them.
This will enable you to adjust your approach and achieve maximum success in future endeavours.
As a major leisure company we have a sizeable customer database populated with transactional information. We are thinking of enhancing our database with third-party data. How do we know which data would work best for us?
Stuart Hopwood: There are numerous data sets now on the market, particularly lifestyle files, which may all provide some form of added value to the information you currently hold.
As a business you need to be clear on what your requirements are from a data application perspective. Different business functions will have different data requirements. Building predictive models for product cross-sell or churn prevention may be the focus of the database analysts, while a deeper understanding of customer demographics may be of more importance from a marketing communication viewpoint.
Once all data applications are known and an appropriate budget set aside, the evaluation process can begin, ideally via an independent third party.
This ensures total impartiality, reduces your time spent evaluating a complex market, and results in your objectives being met within budget.
Data attributes, including coverage, match rates, number and breadth of variables, actual data, recency, method of data collection and cost, can be scored along with relative importance based on your needs. Test the statistical uplift for acquisition, cross/up-sell and segmentation to identify the most predictive data set. This could be a retrospective piece of analysis on known responders to a campaign. Finally, consider the data supplier. Does it offer any relevant additional services such as consultancy? Is there a strong data strategy in place?
The final recommendation will be based on an objective and practical evaluation, ensuring the optimum data is selected.
To have our experts solve your data problems, send your questions to: noelle.mcelhatton@haynet.com
DATA DOCTORS
Richard Lees is MD of The Database Group. Richard's multi-national career in database marketing spans 20 years.
Gary Selby is joint MD of B2B specialists Information Arts. Prior to this Gary worked at Identex.
Stuart Hopwood is director and founder member of Datalytics. Before this he worked for Claritas for nine years.