• http://www.smsrd.com Bob

    Great article.

    How do you handle conversions that are likely time delayed in the pipeline? The basketball example is great but traffic often has a life cycle that can be 30-90 days or more. We can see this impact in the start up or shut down of a campaign.

  • http://www.rimmkaufman.com George Michie

    Thanks for the kind words. You’re absolutely right, the longer the ‘sales’ cycle the more challenging this becomes. Extending the window of consideration makes sense, but that also highlights the challenge: we never really know with certainty what the ‘first’ touch is, we’re really stuck with a notion of the first touch within some window, or the first touch since we turned on this tracking. Imperfection is the norm.

    The other challenge related to this is: whether those early early touches were actually reflective of a different shopping mission. I visited your site after searching for ‘shoes’ 4 months ago, bought shoes somewhere else. I visited the site again looking for a scarf and made the purchase. Should the shoe ad/ads get credit for this scarf sale? How much?

    The folks who claim to have the perfect solution are either delusional or deceitful. There is danger in false precision.

    We do recommend a longer window when it’s a long sales cycle/ high consideration purchase. Doesn’t make sense when you’re selling flowers or running shoes, but for real estate… Often, these businesses don’t have to worry too much about the department store effect either.

    Hope that helps.

  • RobK

    Hi George,

    Interesting read. What is your view on measuring and reporting on banner conversions ‘post view’.



  • http://www.rimmkaufman.com George Michie

    Hi Rob,

    As I mentioned in the italicized note: view through conversions are real, and carefully measured one finds that most of the incremental traffic lift created by display ads comes from folks who don’t click on the ad. Unquestionably the degree of view through effect will depend on the business, as well as on the Display creative. “Check out the great deals at Foobar.com” will probably have a bigger view-through effect than other messaging might.

    Attribution is useful here, particularly when calibrated by hold out tests.

  • http://www.brianfosse.com brianfosse

    George – Based on your experience at RKG where do you see value in affiliate marketing? This example taken in isolation would indicate that they bring little value. However, are their certain business types that you’ve found to be a good fit for affiliates? What best practices have you established to measure their value?

  • http://www.rimmkaufman.com George Michie

    Brian, great question. The value of affiliate traffic depends entirely on how they generate their traffic. There are affiliates who do their work through social networking, through organic optimization and other legitimate techniques that require hard work and know-how. There are others that focus all of their efforts on ranking well for their client’s trademarks and trademark + coupon searches. Whether the balance of affiliates, affiliate traffic, and affiliate commissions are produced by the type that drives incremental value or by the type that produces significantly less incremental lift will depend on how well regulated the program is.

    We think it’s helpful to classify coupon affiliates differently from other types of affiliates so the model can identify those distinct behavior patterns correctly, and not tar all affiliates with the same brush.

    We have found that for most of our clients moving from last touch attribution to a sophisticated attribution approach drops the perceived sales generated by affiliates by 50 – 70%.

    That’s actually less of a drop than we expected to see. It seems a pretty good chunk of customers start their shopping at coupon sites and search for deals on products and services.

    Hope that’s useful