Q: I work for a mail order company and am finding that customer acquisition is growing across our retail and web channels but catalogue sales are falling off. As a result, I am considering reducing the targeting of consumers with catalogue mailings and concentrating on online and retail promotions. Would this further improve our acquisition growth?
Andy McDermott replies: This is something that many mail order companies seem to be experiencing and they are coming to similar conclusions in terms of what channels are 'seemingly' most effective. However, this is a very simplistic way of viewing the situation.
What such an approach fails to recognise is that in the multi-channel world in which we live, it is essential that you understand the impact that one channel promotion has on consumer purchases made across the other channels you offer.
One approach that some companies are beginning to adopt, following on from a catalogue mailing, is to analyse their total sales across all their routes to market for the duration of the whole campaign. This is done by matching back all sales in all channels to the original mailing file. These results can be quite significant and the companies that follow this approach are finding that a high proportion of incremental revenue and profit from sales in the other channels can be directly attributable to the catalogue mailing.
Q: We're a financial services company mailing 150,000 cold names every month. We're paying our bureau £4,000-£5,000 for merge purge. Surely it's as simple as running a 'Find Dupes' query in Access. Should we take it in-house and process it on our desktops?
Martin Rides replies: Deduplication processing is far more sophisticated than you'll ever be able to achieve through 'finding dupes' in Access.
'Find Dupes' works on an exact match principle and isn't sufficiently sophisticated to cope with typical errors in data. You can write queries to overcome these idiosyncracies but that's the slippery slope to writing your own dedupe routine.
A typical bureau-based dedupe routine looks at data from many different angles, incorporating algorithmic matching, phoenetics and fuzzy logic to help cope with the myriad of different ways a name and address can be presented. Further, your bureau should be able to add features, such as hierarchical or random selections, to match at individual, surname or household level and to undertake suppression processing at a different match level to the main dedupe.
Most dedupe programs run to hundreds of thousands of lines of code. The process is still an art as opposed to a science, but it's worth every penny you pay to your bureau to avoid the damage to your brand caused by mailing duplicate or non-existent people. Pay the fee and be glad you didn't have to write the program yourself.
Q: Most DM we do directs some of our prospects to our website, but we know that more people visit than buy. Is there a way of ensuring they either transact or give some personal details that we can then follow up?
Matthew Mills replies: This is a common problem with any marketing that carries a generic URL, and is awkward when you want accuracy. There is an answer however. If you personalise your DM to each individual using digital print or variable e-mail you can include an individual URL, www.intimis.biz /matthew.mills, for example. In this way every visit can be measured and the information used to re-engage with the prospect through telemarketing or DM - subject to the law.
Embracing one-to-one more widely will give companies greater scope to manage the web experience based on the individual prospect and what you already know about them. Even at a basic level you'll know enough to reduce your website content to what is relevant to them.
If you make your brand relevant, they'll listen and if you offer them what they want, they'll buy. This may sound obvious but is seldom done.
And it really does work.
There are some issues with this process - difficulty being one, but there are solutions that allow this level of DM and online integration.
Q: Following severe problems with data management, we've been left with a database that has chunks missing. What is the best way to bring the file back up to its original volume and original quality?
Colm O'Hara replies: Try and see this as a good opportunity to clean up your database. The best place to start is by auditing the data you do have to uncover how incomplete it really is, and the location of the gaps. But before you start buying in more data, take a rational look at what you'll actually use. If you use the database to regularly target customers when their car insurance is up for renewal, this is the information you require, not their children's dates of birth!
Once you know where the holes are, you can start sourcing additional data. Look internally. For example, if the data hasn't been de-duped in a while, see if you can cross-match contacts, which can potentially complete some of the missing data. Once internal avenues have been exhausted, find a data provider to supply what you need.
Once the database is up to scratch, make sure you keep it that way. Ensure all sources that feed into the database are collecting the information you require to stop any further holes developing.
To have our experts solve your data problems, send your questions to firstname.lastname@example.org.
The questions we publish will be anonymised.
Andy McDermott is MD at Abacus UK which he joined in 2001. He has more than 14 years' experience in the data industry.
Martin Rides is director of Data Services at DPS, where his focus is data integrity through integrating software.
Matthew Mills is director at Intimis, an individualisation service provider.
Colm O'Hara is database manager at EuroDirect where he joined in 2002. He has 13 years of database marketing experience.