February 26, 2009

Matchback Comment From "Anonymous"

An individual named "Anonymous" left the following comment this week:
  • "I have the best matchback algorithm in the industry and you are completely wrong. 90% of our orders are sourced to catalog mailings, we get about 5% from emails. Why don't you show us a graph of your brilliant algorithm that shows how emails drive more sales. You don't have one because you don't know what you're talking about. This is the most retarded thing I've ever seen. Anyone can tell just from the order flow that catalogs drive virtually all the sales (all the orders come in right after a catalog drops). Who benefits from your airhead analysis? Emailers?"
Courtesy of my web analytics tools, I can see the companies represented by folks who leave comments like these. What a shame that the individual deprived us of the name of the company s/he represents.

Most catalogers know to compare matchback results to mail/holdout results.

If matchbacks suggest that 80%+ of volume are matched back to a catalog --- and mail/holdout tests suggest that you can mail fewer catalogs without a significant drop in sales, then matchback analytics are wrong.

If matchbacks suggest that 80%+ of volume are matched back to a catalog --- and mail/holdout tests suggest that there is a linear drop in sales as each catalog is taken away from the customer, then you are in the same company as the individual leaving the comment above.

That assumes, of course, that the individual leaving the comment was savvy enough to execute mail/holdout tests.

Sometimes our industry steps to the plate, swings, and misses.

6 comments:

  1. Anonymous5:44 AM

    This is a great issue to debate and I believe the answer depends upon the company and to a certain degree, customer demographics. It's unfortunate that a respectful debate is sometimes difficult to solicit. At least there's passion around analytics:)

    ReplyDelete
  2. I think this may be a symantically arguement to some degree. Matching back a sale can be finding the marketing channel that a sale came in through. Or it can be matching to the channel the created the sale.

    This also makes me think about Organic sales. Shouldn't one find the organic sales of each marketing channel and NOT match that sales back to anything?

    Also, as primarly a non-email database marketer I would like to raise my hand and say that I benefit from your airheaded analyses.

    ReplyDelete
  3. In my opinion, if you are matching your conversion to a single marketing event, you are really missing the entire picture. In today's market, it is really about the accumulation of the marketing events that can drive a purchase decision and not just one catalog or one email. I'm a big fan of looking at the path to purchase in the response allocation process rather than attempting to use a "Last touch wins" methodology. Sorry to say this, but if your "algorithm" doesn't take all relevant marketing events into account, it is not near the "best in the business."

    ReplyDelete
  4. Hi John, demographics certainly play a role!

    Matt --- yes, organic sales should be viewed separate, though many matchback vendors allocate those back to catalogs.

    Jeff, my algorithm is actually called Multichannel Forensics, and is very different than the matchback algorithms that are mentioned here. I forecast future sales on the basis of the past channels customers order from --- I leave allocation of orders to folks who specialize in that stuff.

    ReplyDelete
  5. Hey Kevin -
    My comments were more directed towards the anonymous comment that spawned the post rather than your algorithms. I'm familiar with your book and your work and I'm a huge fan. I think that your Multi-channel Forensics forecasting is some of the best thinking around this topic that I've seen to date.

    Jeff

    ReplyDelete
  6. Ah, thank you for clearing that up, Jeff!

    ReplyDelete

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

This Statement is Accurate, and it is Inaccurate

Here's the link for all of you who love print and think #printisback, which most obviously ... is ... not ... back. https://www.paperage...