April 04, 2011

Tuesday Mailbag: E-Mail, Attribution, Gap

Remember, these are real questions from fake individuals.

Mary in Modesto asks ... I'm really fascinated by attribution models right now.  These scientists are somehow able to parse an order into pieces, allocating each piece to the advertising channels responsible for causing the order.  It's so neat!  Why don't you talk more about attribution models?

Kevin ... Let's pretend that you visit a restaurant.  At the end of the meal, you're still hungry.  You have three choices, apple pie, hot fudge sundae, and rhubarb cobbler.  So you're going to get something, and you are leaning toward apple pie.  You ask the waiter if he has any specials today, and he says that the rhubarb cobbler is 15% off, and you get free ice cream and he'll heat your apple pie for you.  You decide to order the apple pie.  The waiter walks back to the cash register, and on a piece of paper, he places a "one" (1) next to apple pie ... he is attributing the order of apple pie to his sales technique.  Is he right?  Well, you would want to parse your apple pie order into pieces ... the menu, obviously, should get some credit, because without the menu, you wouldn't have gone down the dessert path in the first place, right?  The waiter deserves credit, because he helped make the sale happen, in theory, right?  And the promotion deserves credit, because 15% off rhubarb vs. no deal on the hot fudge sundae vs. free ice cream and heated apple pie sways customer demand, right?  How you allocate the percentages makes all of the difference ... and then there's one little detail left outstanding ... you were going to order apple pie regardless!!!!  In other words, the waiter and the discount/promo strategy deserve no credit, 0%, because the decision was made prior to the initiation of marketing.  This is what attribution experts struggle with ... the most important part of the attribution process is the determination of "organic demand", demand that would happen anyway, without advertising.  And that's why I don't talk about attribution very often ... the whole endeavor is a big guess.  Personally, I like to leverage mail/holdout tests in email and catalog marketing to influence my attribution activities, when required to do so.  Hint:  If you are an e-commerce or retail brand, the majority of your orders happen organically, without the aid of marketing ... if your attribution vendor claims your orders are largely due to marketing activities, quietly seek bids for a new attribution vendor.

Bill in Ocala asks ... What do you make of this big "relevancy" trend in email marketing?

Kevin ... Let me ask you a question, Bill ... when was the last time that you planned an email marketing campaign that was supposed to be "irrelevant"?  Recently, we talked about "the frightening", a situation where an old-school channel is being buffeted by new channels offering "new rules".  Email marketing is going through "the frightening", courtesy of social media.  Anytime a channel goes through "the frightening", interesting and unusual responses are expected.  Suddenly, email marketing weakness due to social media isn't the fault of the channel, no, weakness is due to you, the silly email marketing director at a big, dumb brand.  You are the reason that email is being cannibalized by social.  If you only produced relevant content coupled with triggers (vs. irrelevant batch and blast programs), your email program would be just fine.  When a channel goes through "the frightening", blame becomes an issue.  The email marketing community wants to blame you for not promoting relevant content.  It's your fault, Bill.

Ross in New York asks ... This name your own price stuff at Gap is really innovative, isn't it?

Kevin ... Sure, it is innovative.  I also wonder why innovation requires price cuts?  Apple innovates by producing products you didn't know that you wanted, at prices that are, on average, greater than the market average.  Listen, retailers are going through "the frightening".  Gap experienced stock prices around $20 to $25 in 2001, stock prices are around $20 to $25 in 2011.  Comp store sales have largely decreased over the past decade.  That's "frightening"!  So you have to do something.  And trust me, somebody thinks that product is the answer.  But product takes a long time (ask J. Crew), so in the mean time, you have to do something.  The question, of course, is what do you do with what you learn?  What do you do when your customers tell you that they only want to pay $35 for a $49 pair of khakis?  If you have a viable strategy for using this information in a way that doesn't damage your full-price business and your use of the information can potentially increase profitability, then have at it!


  1. What people are currently calling "attribution" in the online space is a huge suck on resources relative to the potential benefits; do it correctly or not at all. Correctly probably means 24 months, huge budget, tons of testing, and an immense amount of corporate discipline. It's *not* about who saw what ad when.

    Online analysts would generate far higher increases in profitability per unit of work by getting into the back end of the business and making a hard connection between what they are measuring and the bottom line of the company.

  2. Great post Kevin. You’re provocative with your ideas as always. Solving the attribution riddle takes a clever analyst indeed. Although from what I gather you feel that the only way we'll discern real attribution is by peeking into the brain. Now that’s “the frightening”. How soon before near-field communication devices start reading us?

    Yet, I believe that you can solve for attribution with an algorithm. Perhaps not precisely, but there’s some sneaky technologies out there today that enable you to combine business acumen with mathematical formula. At least enough to make an informed decision about where to spend your bucks.

  3. Jim --- yup, we agree!

    John --- I agree that you can do attribution well ... some people have advantages, catalogers have the biggest advantage with mail/holdout tests in catalog and email.

    I had experiences with a reputable attribution vendor when I worked at Nordstrom ... the attribution vendor wanted to credit 35% of store sales back to direct marketing activities. Holdout tests showed something like 2% of store sales should be attributed to direct marketing activities. And the attribution vendor beat me up over it, told me I didn't know how to evaluate the success of a multichannel business.

    To do attribution in a way that doesn't damage the business, you need a person who thoroughly understands the concept of the organic percentage ... the percentage of orders that happen without the aid of advertising. So few people understand this concept, the idea that a customer is going to order something without being prompted to do so.

    In my projects, I routinely find e-commerce brands with 40-80% organic orders, catalogers with 30-70% organic orders, and retailers with 95% organic orders ... I review their attribution rules, and I see how the attribution rules cost the company 10% or 15% of total company profitability ... seriously, it's bad.

    So you have to know the organic percentage, and you have to have testing, where possible, to educate the rules. When you do that, I'm ok with attribution.

  4. The funny thing is some e-mail marketers do market to irrelevant audiences. That is why we have so much trouble having to deal with spam, or mail that has NO business in our inbox.


Note: Only a member of this blog may post a comment.

Mohawk Chevrolet

Here's a little bit of levity for a Monday morning (click here) . Going forward, you might think about how marketing efforts fit on this...