Opinion: watch your assets

What you don’t know will hurt you – and your business. The ongoing management of data must become part of business as usual.

By Steve Tuck, Datanomic Ltd

Last month, we heard customer data described as an organisation’s second most valuable asset. Lessons have, I hope, been learned about maintaining the security of customer data at all times – the Nationwide Building Society laptop debacle and the resulting fine of nearly £1m earlier this year was a warning to us all. However, poorly designed software, sloppy business practices and the good old typographical, of 'fat-finger' error present an ongoing threat to the quality of the data we’re protecting in our CRM systems.

In a survey commissioned by Datanomic and carried out by eMedia Research between July and August this year, 35 percent of those responding said that the quality of customer data is critical to their business and 31 percent reported that errors in the data had had a negative impact on business. But only 21 percent of respondents said that their organisation can measure the cost of poor quality data to the business.

Most worrying of all, 30 percent said that they had no confidence whatsoever in the contents of their CRM database. The Datanomic survey highlights the fact that the vast majority of organisations have nothing in place to measure and monitor the quality of their customer data.

The choice is simple; you can either take a proactive, analytical approach to understanding your data, or you can turn a blind eye, sit back, and wait for problems to bite you on the bum.

Anecdotal examples of problems are nothing more than symptoms of a deep-rooted problem. In the same way that a doctor will carry out tests to establish the nature and extent of an injury or illness, an organisation that is suffering from the symptoms of poor quality data needs to put its data under the microscope.

Symptoms of critical data issues include:
• Customer complaints about incorrect personal data
• Difficulty in selecting and targeting marketing campaigns accurately
• Growing volumes of returned mail
• Customer dissatisfaction about duplicate mailings and inappropriate offers
• Staff have to query multiple systems because information about customers is fragmented across them
• Business users losing faith in the CRM system

Medical before the triathlon

I am not for one minute suggesting that any company with less than perfect customer data should hang its collective neck in shame. Even if it were achievable, the cost of delivering perfect data would far outweigh the cost of living with a degree of inaccuracies.

On the other hand, operating your business without any formal way of measuring the fitness of your customer data is like venturing off on a triathlon without having any sort of medical first. At best it’s irresponsible; in the worst case scenario it could be fatal.

Start by assessing the current situation. You need to understand what data you have, where it comes from and what it is used for. Then you can profile your data to see what it actually contains rather than what the documentation says it should. At its simplest, profiling counts the number of occurrences of each unique value in each data attribute, telling you what the range and distribution of values is.

If you’re a dab hand at SQL you might choose to do some of this manually, but a data profiling tool can slice through large volumes of data and deliver results far more rapidly. A tool will also help with more complex areas of analysis, such as identifying relationships between attributes and finding duplicate customer records using 'fuzzy' techniques.

The point of profiling is to add to your understanding of your data by highlighting things you don’t already know about it and it often results in more questions; why for instance are there four different values in the gender field, or why do so many of your customers appear to have been born on the first of January? As you start to discover new facts about your data it’s likely that you’ll want to begin 'fixing' things, sometimes so that you can ask more questions of it.

Unless you already have all the answers, you’ll also need to collaborate with others to get to the bottom of things (if you’re using a data profiling tool it should support issue management).

A word of warning: it’s not uncommon for people to get so embroiled in the detail that they miss the big picture. Take care to assess the impact and cost of each data deficiency so that you can make sensible decisions about which (if any) need fixing and what the relative priorities are.

Getting your data fit – and keeping it that way

The process of understanding your data will give you a diagnosis. If that shows costly faults in your data it will be necessary for treatment to begin. Radical surgery is often required to improve data and remediate the problems that already exist. Thankfully, modern software integrates the capabilities to understand and improve data so that you can move quickly from one to the other. Software of this type will also come with templates that you can fine-tune to your particular requirements – you should become proficient very quickly and not need to undergo years of training.

A mistake that is often made is to carry out any data improvements as a one-off project, but unless we protect our data, new errors will assuredly creep in. Rather than having to repeat the exercise every year or two, it’s far better to keep your data fit by protecting it. You should be able to exploit many of the rules that you’ve used to improve data with minimal effort. Protection should prevent common errors from ever reaching the CRM system, ideally stopping them at the point of entry.

Finally, you need to keep control of your data on an ongoing basis; regularly monitoring the quality of your customer information so that you can see how things are improving and be alerted to any setback before they grow into a serious issue. The ongoing management of data must become part of business as usual. With data quality acknowledged as a critical business issue, the senior management should demand access to an online dashboard that lets them quickly and easily see the health of your organisations data today and also how it has changed over time.

Steve Tuck is chief strategy officer for Datanomic Ltd

Recent articles by Steve Tuck