June 30, 2013

Dear Catalog CEOs: Terrible Attribution or Matchback Reporting

Dear Catalog CEOs:

For the past ten years, you've been told to match back online orders to the catalog that "caused" them.  We conveniently ignored the "cause" portion of the statement - we just took every single online order and gave the catalog credit.  This is fun, of course, because it guarantees that we keep mailing catalogs, no matter whether the catalog had any impact or not.  In fact, if you mail every single customer in your database a catalog, then you will attribute every single online order back to the catalog.  You see this, often.  Of course it's nonsense.  But it keeps the vendor community employed, and it keeps those who love producing catalogs employed, so the behavior will not change.  I've railed against the behavior for the better part of a decade, few listened.  I cannot change those who do not want to change.

I can, however, point out nonsense in the reporting generated by the vendor community.

Too often, we see attribution mistakes so nasty that they should be pointed out, immediately.  And they aren't.

For instance, say your business possesses an average order value of $150.  You look at a distribution of average order values, and notice that only 10% of your orders are under $35.

Then your vendor produces a matchback analysis for you, by channel.  And you see something like this:

  • Catalog = 1,000 orders, average order value = $211.
  • E-Mail = 200 orders, average order value = $160.
  • Search = 200 orders, average order value = $160.
  • All Other Online Orders = 500 orders, average order value = $20.
  • Totals = 1,900 orders, average order value = $150.
What is wrong with this picture?

Well, the attribution routine completely butchers actual attribution, doesn't it?  We know this to be true, because only 10% of company orders are < $35, and yet, the attribution program says that 500/1,900 = 26% of company orders are attributed online, with an average of $20.

The data clearly indicates that this attribution vendor has no concept of reality.  None.  It simply doesn't matter that the vendor uses a "proprietary and sophisticated algorithm" ... the algorithm is simply and horribly wrong, and will cause you to make terrible business decisions.

I see this outcome, often.  Your reporting is just plain terrible.  Either you don't care, or you don't even look at the reports.

Do you even look at the reports?

Do you ever challenge your marketing team to actually think about the reports generated by their favorite matchback vendor?

It's one thing to outsource marketing attribution to industry experts, giving them the authority to know your business better than your own internal staff.

It's quite another thing to completely accept terrible, terrible reporting as truth.

And that's what too many of you you're doing.

No comments:

Post a Comment

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

No Context

Read this article and you'll be struck with a notable finding ( click here ). There is no context here. "Pureplay decreased by 51%&...