Gliebers Dresses: Decomposing Item Performance

Welcome to this week's Executive Meeting:

Glenn Glieber (Owner): "... needless to say, reaction to my comments at the conference has been highly negative in the blogosphere and on Twitter, but was very favorable among other catalog CEOs."

Meredith Thompson (Chief Merchandising Officer): "Kevin, is that you?"

Kevin: "Yup, it is me."

Meredith Thompson: "We're trying to understand how productive our new strapless tube dress is, Kevin".

Pepper Morgan (Chief Marketing Officer): "In the old days, it was easy to measure the profitability of merchandise, and it was hard to measure the profitability of a customer. But these days, it seems like the situation is reversed."

Roger Morgan (Chief Operations Officer): "Absolutely. This item has never been featured in an e-mail marketing message, and yet, it seems like the item sells really well during the twenty-four hours following an e-mail blast."

Lois Gladstone (Chief Financial Officer): "I want to install a discipline around here, one where we make formal decisions about the merchandise we choose to invest marketing dollars in, vs. letting items be featured online without the need for expensive catalog support."

Roger Morgan: "But that's hard to do, Lois. I mean, we know that when we mail a catalog, we get $1 offline for every $1 generated over the telephone. But that relationship is an average, right? The relationship doesn't hold for every single item."

Pepper Morgan: "And the relationships are confounded by marketing strategy. When we bump up our paid search efforts, we notice that we capture more new customers in our catalog marketing activities, but the new customers are skewed toward items featured in the paid search ads."

Kevin: "Since we're about to select customers for the January catalog, let's try something. Randomly sample 16,000 customers, and assign them to one of four treatments for the first six months of the year. Group One = Receive All Catalogs, Receive All E-Mail Campaigns. Group Two = Receive All Catalogs, Receive No E-Mail Campaigns. Group Three = Receive No Catalogs, Receive All E-Mail Campaigns. Group Four = Receive No Catalogs, Receive No E-Mail Campaigns." Lois Gladstone: "Six months?" Kevin: "Yes. This test will allow us to do a few things. First, we get to see how customer behavior changes over time. Second, we get to see if there is a multiplicative impact between catalog marketing and e-mail marketing. Third, we get to see how paid search is influenced by catalog marketing and e-mail marketing. Fourth, we should be able to measure, at a merchandise division level, if there is a difference in the $1 offline = $1 online relationship."

Lois Gladstone: "Can we afford to do a test of this magnitude?"

Kevin: "I don't think that catalogers can afford not to do this style of testing anymore. Your online brethren are doing this type of testing, they call it 'optimization', in real time, all of the time. The only way we, as an industry, are going to profitably navigate our way to the future is to truly measure how customers behave, with and without advertising."

Roger Morgan: "But we do those things, what are the called, 'matchbacks', right? Doesn't a matchback give us the answer we're looking for, without the need to do testing?"

Kevin: "Matchbacks are good, you're far better off doing them than not. Unfortunately, matchback programs are built off of rules, rules determined by humans, not by customers. For instance, you'll use a 60 day window to match online orders back to catalogs. That's a rule. If a customer clicks through an e-mail campaign two days after receiving a catalog, you allocate 90% of the order to the catalog and 10% to the e-mail campaign. How do you know that? You really cannot get into the mind of the customer to understand if 10% of her thought process was influenced by the e-mail campaign, can you?"

Lois Gladstone: "Isn't that what the online marketing folks do, too? They call it 'attribution', don't they? If they're doing it, we should be doing it too, right?"

Kevin: "Yes, online folks are concerned about allocating orders based on first-touch or last-touch or blended-touches. They are dealing with the exact same problem you are dealing with. And they are making the same mistakes you are making."

Meredith Thompson: "So are things hopeless? Can we not properly measure merchandise profitability?"

Kevin: "Hopeless? No. Harder? Yes, much harder. We can no longer prove what caused a customer to order merchandise. Fifteen years ago, it was easy. If you send an e-mail to a customer offering a Thanksgiving dress and she instead goes online and likes your strapless tube dress, you're kind of sunk. Does that mean you should offer the Thanksgiving dress next year, or a strapless tube dress? Merchandise profitability becomes an inexact science, blending more and more with marketing strategy. At least with the tests you'll execute, you can get a directional idea how profitable various merchandise divisions are, and then you'll apply rules, something I criticized earlier, and you'll get as close as you can get. But by and large, merchandise profitability is harder to measure than customer profitability, and will continue to be harder to measure."

Glenn Glieber: "In some ways, folks, I'm glad this is harder to measure. If it is harder to measure, then it requires more art, more inference, more gut feel. And honestly, we all get a rush from making correct decisions without good information. Nobody likes to have all of the answers on a dashboard, robotically driving where the business is headed. Honestly, I think a golden age of merchandise leaders can emerge from this. Don't use this as a crutch for indecision. Use this as an opportunity to allow your genius to shine through. Now, let's move on to the next topic. Tomorrow is Halloween, and I want to make sure this place doesn't look like a pit on Monday. So all decorations need to be down by 4:00pm."