I work for a brand in a highly competitive telecoms environment where churn rates are traditionally high. If we can predict those people likely to change operator we can cut our churn rates. But how can we analyse our database to identify the most likely candidates?
Marc O'Regan replies: The good news is that you can predict customers who will churn. I'll leave the bad news to the end.
What data would be useful? Certainly look at tariff used, call types and volumes. We have had success in analysing the trends in changes in customer spend over time and using these characteristics as predictors.
If you have them on your database customer service call records may be a powerful way of identifying unhappy, at-risk customers. Do those customers who complain about billing, network performance and handsets have a high churn rate?
Transactional data is generally a lot more predictive than external demographic data, however this data may help describe high-risk customers so may add a useful dimension to the study.
The bad news is that the model alone won't retain any customers. You need to develop and test an offer, and compare churn rates among a test group and a control group. Cost benefit analysis of the test campaign is critical to determine which segments to roll out to. You need to compare churn rates between test and control groups overall, and then by segment.
This will evaluate the campaign overall, and identify the best targeting approach for roll out.
I work for a not-for-profit association and am concerned we're not generating enough response from our member e-mail data. How can we achieve better results?
Niroo Rad replies: I suspect that one of your problems is caused by the poor quality of the member e-mail addresses in your database and their preferences. It's no good simply having generic addresses - especially with the new rules surrounding the classification of spam. You must use current and individually named e-mail addresses.
If your e-mail contains a call-to-action, make sure it is clear what you want your member to do and that the task is simple to execute.
For example, if you are trying to get members to update their contact details online, it's no good having a link on your home page from which they have to click through to get to the personalised page - you'll lose most of them at the first page.
The trick is to insert a hyperlink that not only takes the member straight to the relevant page but also immediately updates your central members database so automatic follow-up is created and executed. Make the process simple and painless for your members.
It is highly unlikely your entire membership database is interested in receiving the same e-mail message. If you are sending out an e-mail relating to the promotion of a specific event or a new service, the message is only relevant for those members who match a certain criteria.
Here profiling will prove invaluable as it allows you to select and extract suitable members as part of the targeting process. Profiling of your data is not as daunting as it sounds. There is technology available to facilitate this and enable your association to become more productive, while at the same time generating higher responses to marketing campaigns.
In summary, a personalised e-mail, targeted at the right member with relevant messaging will make a significant difference to the results of future e-marketing campaigns.
How can I make money from my database without mailing them?
Simon Jones replies: The obvious answer here is list rental, where the names of your customers are offered to other companies offering similar products or services.
There are numerous companies who will manage list rental on your behalf, for example Prospect Swetenhams, and provided the necessary steps are put in place to comply with legal and best practice recommendations - opt in/out flags - the process is simple and painless.
If there are no rental opt in/out flags on your database, then you won't be able to rent or swap them without breaking the Data Protection Act.
There is, however, a way to sidestep this, whereby you communicate with your customers by offering them the services of the third-party. Naturally, the third-party pays all the costs.
You may also be able to swap lists with similar companies, thus gaining access to their customers. There is no need to be worried about 'giving away' your customers to a competitor - the chances are they already have them on their database. You could consider pooling your customers with a co-operative database, which will give you access to potential customers at a low cost.
To have our experts solve your data problems, send your questions to stuart.derrick@haynet.com. The questions we publish will be anonymised.
DATA DOCTORS
Marc O'Regan is head of the data consultancy division at Proximity London, which offers CRM analytics services.
Niroo Rad is the CEO and managing director of ASI Europe, a provider of fundraising and website management software for the not-for-profit sector.
Simon Jones is data director at Liquid Communications, where he helps clients solve their data and CRM problems.