By now most multichannel merchants know about matchbacks and the importance of understanding where online and retail buyers are coming from rather than guesstimating the percentage of sales driven by the catalog or an e-mail campaign.
To help marketers more accurately assess the impact of their catalog, e-mail, and other campaigns, several advanced software programs have evolved during the past few years. Basically these programs allocate unsourced catalog, Web, and retail transactions by matching name and addresses to mail files.
Because matchbacks are so hot, “any provider who’s taking customer data, whether it be for a relational database or a customer file, has to have a way to deal with matchbacks if the customer wants it,” says Coy Clement, founder of East Greenwich, RI-based catalog consultancy ClementDirect.
Matchback tools can come as a stand-alone product, such as ChannelView from Abacus and Channel Match from Experian, or from a service provider who includes some sort of matching capability as part of its overall services. Matchback services are available from database and marketing services providers as diverse as New York-based HarteHanks, Central Islip, NY-based MBS, Rensselaer, NY-based I-Centrix, and Little Rock, AR-based Acxiom. ChannelView 1.0, from Broomfield, CO-based cooperative database services provider Abacus, launched in January 2002. Fellow database services provider, Experian countered with Channel Match in May 2002. Abacus upped the ante this past December with ChannelView 2.0, which the company says allows marketers to track sales in up to six channels, including paid search, affiliate marketing, and e-mail marketing.
Both systems use a merchant’s promotional files and response data. ChannelView is Web-based; Channel Match is not. Costs range from $10,000 to more than $30,000 a year, depending on how many catalogs, e-mails, or other promotions are mailed annually.
Perhaps the biggest difference between the Abacus and Experian products is the methodology each uses to match records. Experian uses “merge logic,” while Abacus uses “match logic.” Channel Match uses more than 1,500 business rules to ferret out discrepancies or match data between files, says Doug Gilbert, Experian’s senior director for product marketing. The process is similar to that of a merge/purge of mailing lists. For example, if a catalog mailing to John Roe at 123 Main St. preceded an in-store purchase from John Roe of 132 Main St., Channel Match would flag the retail purchase as having likely been driven from the catalog mailing.
Conversely, in the match code process used by Abacus’s ChannelView, a unique match code for each transaction that represents an individual in a household. The match code is a function of components, including name, gender, and address, says Casey Carey, Abacus’s vice president of marketing, data solutions. Through algorithms in generating the match code, ChannelView can recognize misspellings or alternate versions, such as Aaron vs. Erin.
Linda Spellman, director of direct marketing for Corte Madera, CA-based cataloger/retailer Restoration Hardware, says the tools improve the accuracy of reporting catalog dollars per book for transactions without a source code, including Web and retail sales.
Before the home furnishings merchant began using ChannelView in 2002, it made mailing decisions based on sales by source code. For orders that did not have a source code, Restoration would allocate the unknown sales by percentage across its other channels, says Spellman. So if the catalog represented 35% of its known sales, 35% of the unknown sales would be allocated to catalog. “We were making decisions based on inaccurate information, and we were cutting lists and house file segments that were successful and not cutting lists that were bad for us,” Spellman says.
For example, catalog names drive a tremendous amount of store traffic for Restoration. “Without a matchback tool, we wouldn’t have the visibility to the true impact a mailing has on our business regardless of channel,” she says.
For a marketer with stores such as Restoration Hardware, the retail matchback process begins by obtaining zip codes at the store’s point-of-service (POS) system. Then your service bureau or provider matches name and credit card to your house file. The non-matches go to the company doing your matchbacks or your service bureau.
Once it began using the matchback tool, Restoration Hardware found that more than 40% of online purchases were attributable to the print catalog; what’s more, online purchases from catalog recipients are 30% higher than other online orders. And catalog recipients who shopped in the stores spent 25% more than the retail customers who didn’t receive the catalog.
Outdoor apparel manufacturer/marketer Patagonia has been using ChannelView only since the beginning of the year. But already, says circulation manager Ken Storey, the Ventura, CA-based merchant has been able to decipher the sources of its Web sales. “We’d had a real good idea based on when the catalog dropped that it was pushing buyers to the Web, but we weren’t quite sure,” Storey says. Now Patagonia has enough solid data that it can rent lists and use database models previously thought to be below breakeven.
For example, a model might reveal that a certain list performed at $1.25 per book, below the profitability threshold. But now that online and retail sales can be matched back to specific catalog mailings and even specific files, Patagonia can see that some lists previously thought to be unprofitable actually performed quite strongly, Storey says. SKU-level matching
For catalogers that mail rather frequently — say every three or four weeks — it’s not enough to know that a catalog mailing drove a Web sale: It needs to know which catalog mailing drove the sale. This is where SKU-level matching, offered by St. Paul, MN-based database solution provider CMS Direct since September 2003, comes in handy, according to its chief information officer Andy Cossette.
For example, if a customer receives a holiday catalog in December and a catalog in January and makes a purchase after the January mail date, the sale would ordinarily be attributed to the January catalog. But suppose the customer purchases an item that wasn’t even in the January catalog but rather from the December catalog? The SKU match can allocate the sale to the most recent mailing that contained the product the customer purchased, not just the most recent mailing. “You can expense your advertising promotion cost against the right sales,” Cossette says, “reducing errors in assuming what prompted the consumer to place the order.”
In the above example, if a cataloger assumes that an order placed in January was driven by the January book, when it in fact was driven by the December catalog, the company may make a strategic decision to cut its December mailing, “and they would be wrong,” explains Cossette. “Now with SKU-level matchback, the cataloger can see that the December catalog did better and may find that they should actually mail more in December.”
Experian’s Gilbert says that more catalogers are beginning to inquire about SKU-level matching, and the company does offer it. But he warns that because there are hundreds of thousand of data points for catalogers to track, the process can be challenging to manage.
In fact, none of the matchback tools are infallible. One problem, says Restoration Hardware’s Spellman, is that they all assume that the catalogs are the drivers to the Website. While the system’s algorithms allow mailers to segment specific channels, its logic uses catalogs as its default mechanism. And if you mail more than one title or market by more than just a catalog, the matchback results can be misleading. E-mail matching
Matching back e-mails is a bit trickier than matching back catalog mailings, says Casey Carey, vice president of marketing, data solutions, for co-op database services provider Abacus. If you don’t have a street address for an e-mail client, he explains, you can’t match retail and telephone sales from the customer. That’s why it’s important to capture the name and physical address of as many people on your e-mail lists as possible.
Simultaneous e-mail and catalog campaigns provide another challenge. Many merchants send e-mails just before or just after the in-home date of the catalog, Carey says, then follow up a week or 10 days later with a reminder: “Typically they leave the e-mail campaigns open for five to seven days and the catalog open for four to six weeks depending on the next mail date.”
Also keep in mind that the e-mail universe is a subset of the universe being sent catalogs, so most have 25%-50% e-mail coverage on their house files and 10%-20% of the prospect pool.
In the end, Carey says, most of his clients prefer to attribute only Website orders resulting from an e-mail click-through to the e-mail campaign. All other orders will be matched to the catalog that is open during that period. If they do not match to the catalog and they have an e-mail keycode, then they get allocated to the e-mail campaign, Carey says. “This approach was driven by the needs of our almost 100 clients who for the most part don’t use the e-mail matching capabilities and just look at catalog response.”
Clients no longer view e-mail and catalog as standalone marketing events with separate ROIs. Rather, Carey explains, marketers now look at the total campaign of multiple contacts and expect that the incremental sales created by the e-mail in conjunction with the catalog will more than cover the incremental cost to the overall campaign. — MDF Matching back on a budget
Even the simplest method of matching back — taking all of your Web customers and running their names against the names of your most recent catalog mailings — usually requires assistance from a service provider. According to Linda Spellman, director of direct marketing for home goods cataloger/retailer Restoration Hardware, this method is relatively inexpensive, at roughly $500-$1,500 per project.
In fact most catalogers use the company that performs their premailing merge/purge to conduct this sort of matchback. The downside is that it generally requires a lot of manual intervention and trial and error. And to keep down costs, analysis is done at the end of the campaign.
Network
