• Pat

    The company reported that for these 30 clients, “‘direct visits’ were getting more credit for conversions under a last-touch model, as they were often the last interaction before a conversion. As a result, other channels like ‘organic search,’ ‘paid advertising,’ and ‘referrals’ were typically undervalued.” Slingshot SEO said that consumers on average “took 2.79 interactions before converting” and that the final touch or step before the conversion “was typically a direct visit to the site or a branded keyword search.”

    AMEN! There are critical insights available whenever those other buckets change – whether it’s a drop or rise in organic, changes in PPC ROAS goals, PPC stoppages, new referral partners… myopia abounds.

  • Pat

    compare name traffic PPC to those direct visits, same dealio. segmentation on front end is essential to observing the outcomes, then trying to “see” causality.

  • http://www.booyahadvertising.com drewgrab

    The key part to this study seems to be the simplified and rarely used “flat multi-touch attribution model”. This model assumes equal weight ought to be applied for each step in a conversion path. However, this does not take into consideration any time constraints or latency periods between steps in the conversion path. Also, this model will exaggerate the value of steps that are often repeated (such as organic non-brand search when users are shopping around).

    As an analyst for Booyah Online Advertising, I do believe these mid-funnel steps are clearly undervalued compared to a last-click or first-click attribution model, but the findings in this article seem to be espousing the value of over-weighting mid-funnel steps in conversion paths. I would be interested to see how these findings hold up using a more sophisticated constraint based attribution model compared to the simplified model used in the study.

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

    Any such study that doesn’t differentiate between brand search (paid and organic) and non-brand search (paid and organic) is not just oversimplified, it’s meaningless. There is very little distinction between the marketing value of direct load and the marketing value of brand/navigational search. In both cases the user knew where they wanted to go in advance. As we’ve argued many times, there is far more that goes into an accurate picture of marketing influence than proportional credit allocation delivers as latency, ordering and channel behaviors are important, but missing the brand/ non-brand divide is a fundamental flaw.

    Their marketing team deserves a hat tip for getting this out there, but as Greg noted, keep in mind the interests of those who produced it.

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

    er, uh, on more careful observation, I see they did separate brand and non-brand organic, if not paid…Mmmmm, this shoe is delicious!

  • http://www.acquisio.com Marc Poirier

    I agree with George – of course it’s great to see beyond the last click. But why is an equal part attribution model the right answer? I’m sure I’m missing something.

  • Pat

    okay, i’ll admit it, i read your article again today. delicious the 2nd time as well.

  • Pat

    “But why is an equal part attribution model the right answer?”
    There is no ‘right’ answer, all answers are a chosen compromise, necessitated by aggregating data. Seek out the one the best reflects your channel’s contributions, then (and only then), are you truly winning the pursuit of optimal outcomes.

    Do not try to bend the spoon — that’s impossible. Instead, only try to realize the truth: there is no spoon.

    And yes, I came back again today. It’s only March, but I crown thee article of the year (even though the title doesn’t do justice to the lid you’re lifting).