Would You Accredit It

With data volumes at their lowest level in a couple of decades tempting bureau to skimp on staff training or overmatch, is it time for a rigorous bureau accreditation scheme, asks James Lawson? 

With no formal standards or certification in the customer data process industry, clients are struggling to confidently select bureaux that can reach defined standards in their processing work. Larger companies can invest in testing prospective suppliers but smaller customers tend to buy on price only. With data volumes at their lowest level for 20 years and razor thin margins, the temptation to skimp on staff training or to overmatch (and so overcharge) is there. With the DMA now working closely with bodies like the BSI, has the time finally come to introduce rigorous bureau accreditation? 

Where to Turn?

Options for customer data processing are wider than ever. As well as in-house tools, there are numerous online as well as offline providers. Then you have suppliers that specialise in processing, those where it is only part of a wider range of services, bureaux attached to printers, bureaux attached to agencies and a raft of 'spare bedroom' operators.

"It's a hugely fragmented market that has become bewildering to new entrants, especially smaller companies," states Marco De Caprio, Commercial Director at Greenstone Data Solutions. "It's hard to distinguish one supplier form another." 

Because hit rates vary between different suppliers anyway (see below), finding out whether an individual job has been done well, badly or merely adequately can be a tough assignment. The supplier selection process is not simple even for those that understand the various processes involved in data cleansing and list building. Hence the attraction of comparing the bottom line cost of quotes, particularly when the buyer is a member of the purchasing function rather than in marketing or IT.  

"The whole market has become incredibly commoditised," says Mark Roy, Chair of the DMA Data Council. "The only discussion is around price." 

But focusing on cost alone is dangerously misleading as a badly run job can easily result in a smaller invoice than a good one. There can be a number of reasons behind the bald hit results on a bureau report. While one supplier might slightly over suppress a prospect list, another might under suppress either because of the different nature of the job (valuable customer data rather than relatively expendable prospect data) or simply in order to win business. 

"We have specific experience of this where we have audited data and lost the work to data cleanse for a client due to another cheaper overall quote, to find that later that they has a multitude of goneaways returned from their campaign," says Nathan Shilton, Head of Data Management at Tangible Data. "So their cheaper data cleanse has cost them more in printing and pack costs." 

Other suppliers report some clients unable to tell the difference between matching between the same files at household level (higher match rates) and at individual level (lower rates). Running test files and interrogating bureau sales staff does require good experience and a sound knowledge of what the processing is aiming to achieve to be effective.  

Without that knowledge, potential clients have to rely on price and other simple indicators to show help them make their choice. Unfortunately the fact that a supplier that belongs to an industry body is no firm guarantee of quality.  

"The DMA is good, but their best practice guidelines are open to interpretation," notes De Caprio. "We would certainly welcome some form of accreditation." 

As well as minor variations in results between suppliers and client ignorance, there is also incompetence and maybe even outright dishonesty to deal with at some bureaux. Research run by the Suppression Providers Alliance (SPA) in 2008 and 2009 certainly looks to support the need formal standards.  

Its mystery shopping exercise found "a great deal of misunderstanding, lack of the knowledge of the basics, inadequate internal processes and suppression practices that leave a lot to be desired." Some bureaux match rates were more than four times the average.  

Match Maker

This came on top of research by The REaD Group back in late 2006 when a number of bureaux were found to have breached their licence by not declaring the full amount of income derived from running client prospect lists against the company's reference files. The need to monitor the accuracy of royalty payments is what prompted the formation of the SPA in the first place.  

Writing in early 2010, Chairman Colin Lloyd noted that the SPA "has continued with its audit programme (both overt and covert) that has identified that clients and resellers have very different perceptions about where suppression sits within the marketing mix. The latest audits show that misunderstanding and confusion still prevail among those bureaux audited."  

All suppliers interviewed for this article supported the concept of industry accreditation with some regretting that the SPA never moved beyond just talking about such a programme shortly after its formation. Most cite the benefit of being able to stand out in a crowded market by virtue of the extra certainty accreditation would bring to the selection process.   

"A scheme that allows marketers to buy with confidence is important," says Roy. "But how do you do it?" 

The first challenge is a technical one. How do you compare results? Simply comparing match rates against a test file is not enough. The business goals for the data processing are required, along with some inspection of other elements of the processing workflow.  

"You would need to go beyond software used to test processes and also- and most importantly - test and audit operator skill and the briefing process," says Shilton. "The problem would be to guarantee consistency of testing." 

Data processing in general and record matching in particular are not exact sciences. When the same input file is run against the same reference files at two separate suppliers, results will often vary due to the settings the operator has chosen, differences in the way the software defines a match and so on.  

Any standard would have to take account of this, perhaps setting a tolerated variance above and below an average match rate for a test file. There would also have to be some sort of standard brief for the test job, for example, to specify that it would be run at individual level, and the test itself would necessarily have to be run anonymously. 

"If you send five files to five bureaux, you will get five different answers," Argues De Caprio. "But if you have four jobs that are similar and the fifth has a match rate that is 20% higher, then that will ring alarm bells." 

Given the focus by the SPA and suppression data owners on more rigorous reporting of suppression matches backed by audits and mystery shopping, these monthly reports themselves could form part of  an accreditation programme.  Royal Mail already asks for the record quantity of the base being screened and their report form automatically calculates the percentage matched.  

This in itself is a form of quality control that could be inherited by the other suppression file providers and would act as an instant flag if percentages seem particularly high. It also fits in well with the idea of monitoring and periodic inspection rather than an initial check with no follow up. 

"It should be like an ISO," says Linda Churches, Account Manager at Occam. "Companies should be audited regularly or they would lose the accreditation."  

This leads onto the other challenges to accreditation: who could manage it and how it could be funded. The obvious answer is to have the bureaux fund it themselves (perhaps with contributions from data owners) under the aegis of the DMA, with the actual testing carried out by a trusted independent like DQM. But given the difficult previous similar initiatives have faced over the years in gaining adoption, such a scheme would risk irrelevance if only a few of the larger suppliers signed up to it. 

Remember the Business List Audit? After years of struggle to establish it as an industry standard, only 12 out of around 250 list brokers and data owners had the logo on their letterhead. It finally folded in 2008 amidst recriminations that the DMA's efforts to promote it had been insufficient. So there would have to be a commitment to a credible marketing campaign to sell an accreditation scheme to the market and maybe even a degree of coercion to make sure the majority would sign up.  

"There could be a body like the DMA behind it whereby suppliers are charged increased membership fees if they decide they need the accreditation," says Emma Thwaites, Client Services Director at Alchemetrics. "However as we've seen with the launch of the DataSeal [data security] standard, the industry doesn't seem to have the confidence that a new standard can be rolled out effectively.  If people aren't being checked and audited correctly, the whole process becomes meaningless." 

Cost Benefit Analysis

With the DataSeal audit apparently corning in at around £5,000 a similar cost for accreditation would hit smaller suppliers in particular, who would lose a larger proportion of their income in funding it.  As there are now multiple ISO and other standards that bureaux are under pressure to conform to, finding the cash for accreditation might be a step too far.  

"ISO standards aren't cheap and anything that requires an assessment and review costs money," says Steve Day, Director at UK Changes. "It would be a pity if accreditation stopped innovative new entrants to the market" 

In the absence of a standard, it's still down to clients to make sure that the processing is up to scratch. If formal testing takes too much time and effort, then comparing match rates (and cost) from different bureaux by running multiple free audits should give a good feel for the hit rate your file will generate in the way of suppressions and so on. Any strangely high or low match rates should show up in this process. 

Thwaites suggests a similar alternative: "They can compare match rates against work with former suppliers to identify and query anomalies within results, which the data bureau would then have to justify." 

Other straightforward questions to ask include finding out when the supplier's reference files were last updated, whether they run all their advertised services in-house and what other accreditation they may have, for example, the ISO 27001 data security standard.

The DMA badge should give some comfort if the code of practice is being followed properly and correctly, while taking up references from larger clients that do require suppliers to run test flies is another way to increase confidence in processing. 

According to Shilton, a client should look for a bureau that does more than just take the data. "What are their briefing processes, what options do they give, are they chatting to you about different deduplication levels or asking if there any special output requirements?" he advises. "Do you need 'tight' matches or does it suit to have looser ones? Can the bureau show you the difference?"  

Although accreditation has the potential to raise standards and performance in the data processing industry along with providing comfort to its clients, there doesn't appear to be much motivation in this struggling industry to kick off such an initiative and solve its inherent problems. For now, clients will simply have to realise that cheaper processing doesn't mean better and otherwise rely on that old adage: caveat emptor.

"Purchasing departments are typically not looking to review quality or evaluate processes," concludes Day. "Unfortunately that doesn't protect one of their most valuable assets - their customer data." 

Article by James Lawson from Database Marketing, Dec 2010