Giving your data a cleaner bill of health

Data is good quality only if it is fit for your purpose. For some pieces of data, that is easy to establish – an email address is either deliverable or not. For others, the standard is less clear – faulty postal addresses may still be deliverable, thanks to the efforts of individual posties.

When you get into more complex aspects of data management, such as using a variable as the basis for targeting or a set of variables to build a segmentation, any gaps or errors start to compound. What begins as a decent database if you just want to send out a mailing can prove unfit for planning your marketing strategy.

Improving that quality involves using a set of tools that get the database ever closer to the (unachievable) goal of 100% accuracy. Data cleansing is a long-established part of that process. Data enhancement is another option that is increasingly being used to build on top of existing customer and prospect records.

Sky is one business that had a particular problem with data that led to it building a bespoke solution. “If there is a block of flats with four or more residences, that causes installation problems and needs a specialist,” says Simon Kaffel, data and analysis director at BSkyB.

Identifying that an address is part of such a block is not easy from standard address databases. For Sky’s call centre, that was causing complaints when potential subscribers were let down because the engineer could not fit a dish. So the media company created a bespoke Multiple Dwelling Units database that flags on its customer and prospect universe – known as Olive – when a caller needs a specialist fitter.

“The quality of our data is paramount, especially when we have customers in hotspots that are difficult to manage, such as Glasgow where there are a lot of tenements,” says Kaffel. “In the Republic of Ireland, the problem is lack of postcodes.” As well as disappointed customers, there is a cost issue with sending an installer to an address that they cannot service or even find.

Olive has been built using data derived from the electoral roll (ER) to gain maximum insight into the potential UK market. As opt-out rates from the use of ER data for marketing have grown – reaching 46% this year – more gaps have started to appear in these sources. To ensure its database remains fit for purpose, Sky recently reviewed all its suppliers.

“We looked at match rates, level of variables and how predictive those variables were,” says Kaffel. “It was a lengthy exercise to understand. We have also reviewed all the variables we hold and removed the less reliable ones.” Working with Acxiom, this has led to the elimination of data such as the number of pets in a household and a focus on more useful indicators, such as “Are you considering purchasing digital TV?”.

Customer service is often the point at which data errors or absences get noticed, particularly in a call centre. The need for richer information about customers is also being recognised in marketing and business planning functions.

Fiona Sweeney, industry strategist at Acxiom, says scenario planning is becoming a critical tool for clients. “We have built a fuel price indicator to look at what happens to discretionary income in every household if petrol prices go up,” she says. “For the very affluent, it has no effect, but there are those in the middle who will be disproportionately affected.”

A similar tool has been built to model the effect of next year’s VAT rise. The purpose of this sort of data enhancement is to sense-check marketing plans and the assumptions they make about the affordability of products and services.

For many organisations, getting to an accurate level of customer identification is just as important. Steve Gardner, database manager at TUI UK, says/ “The main issues we have are around matching transactional data back to an individual without a unique customer reference at the point of sale.

“So matching a customer and all their purchases is particularly difficult when they change address, or use a slightly different name, new email address or different phone number.” Enhancing with contact data from third-party sources was a key way forward, says Gardner.

We used solutions with our data bureaux to improve the matching process, introducing ’fuzzy’ logic in some cases, and encouraged accurate data capture at point of sale through targets. We now have a cleaner database that has more accurate customer information and we have morechannels to contact customers.”

The ability to contact a customer is central to marketing, but the data which allows that to happen is not always easy to get. Permission-to-market rates are the levels of contact data and appropriate opt-ins or opt-outs for each channel that are present in a database.

If these permissions are too low, marketing campaigns will not have sufficient volume to meet business targets. Data enhancement to fill the gaps is possible, but it is complicated by the fact that each channel has a different set of rules.

“Email appending is a grey area,” says Dee Toomey, managing director of Scientia Data. “We follow DMA guidelines and all our data is opted in for third-party marketing.” For compliance, a person must have consented to unsolicited email. Even then, best practice is for commercial data owners to email them twice, seven to ten days apart with a further cooling-off period, giving them the chance to opt out.

This double opt-out approach explains why data enhancement of email addresses is difficult. “Clients often expect to get an 80% or 90% match rate,” says Toomey. “But because we have to match tightly, you get more like 10-20%.” Using a third-party source can help to fill a gap, but capturing this contact information through improved internal processes is a better option.

Appending telephone numbers does not require a permissioning campaign, but there must be an opt-in for third-party marketing plus screening against the Telephone Preference Service. “TPS registration is so huge, it can take 60% off your matches, so the telephone channel is not that viable,” says Toomey.

We now have a cleaner database that has more accurate customer information

SMS as a marketing channel is expected to grow hugely in the next five years. Appending is simpler, requiring just a third-party opt-in, and there is no preference service as yet. The channel can be very powerful, especially for charities, where “text to donate” is proving a highly effective mechanic.

Organisations that rely on donors for their income, such as charities, have a duty to maintain their data to the highest possible standard. This is not always easy – unlike commercial organisations, supporters do not always tell charities when their data is wrong.

Breast Cancer Care has used a data bursary from The REaD Group to help develop its new direct marketing programme. In particular, it wanted to improve its matching against deceased suppression files and look to reactivate dormant donors. The REaD Group was able to identify 5,400 new addresses from its relocation file.

Cleaner data has also improved segmentation of its supporter file. By profiling using the Cameo tool, the charity created seven support segments. Profiling showed that a supporter type it called “Susan” was prevalent in all seven segments, but with important differences for targeting and propensity modelling.

Data enhancement and cleansing are means to an end – usually higher revenues, more customers and supporters, and less wastage. Closing that gap between targets and marketing performance is worth investing in.