News

Get a sneak peak inside some of Clarivoy's newest developments!

BY Vice President of Marketing Katie Robinson COLUMBUS May 03
Rick Otten

Rick Otten, Principal Data Architect at Clarivoy

Q: I understand you’ve just integrated a new attribution model into Clarivoy’s TV Analytics dashboard. What is the new model called and how is different from the others in the dashboard?

A: We’ve added a “Linear Time Decay” series of models to the TV Analytics dashboard. After a spot runs, we assign credit for the inbound web traffic following the spot for a certain amount of time, which we call a ‘response window’. The other models consider all of the traffic within the window equally – whether the response is seconds after the TV spot ran, or seconds before the window ends. The Linear Time Decay model assigns diminishing importance to the web responses as they get further and further in time from when the TV spot aired. This model follows a straight line from full credit up to one minute after the start of the window, to zero credit at the end of the window.

Q: Why is it important?

A: We believe this model is closer to approximating the real world impact of experiencing a TV spot. In most cases, as time goes on the impression and memory of a typical TV spot fades.

Q: Can you provide a real world example of how the Time Decay Model illustrates a customer’s purchase path?

A: The excitement of experiencing a compelling TV spot is real and may motivate a potential buyer to take immediate action to begin researching their next big purchase. In a world of short attention spans and myriad distractions, the excitement and intent to follow up on that idea can be quickly subverted. Factoring this into our analysis of the impact that spot had on their purchase decision helps us more accurately assess the value of the TV campaigns that have been recently run.

Q: In addition to the type of attribution model, what other factors should be considered in assigning credit to TV spots?

A: The attribution window you use and the type of fractionalization are two important factors to consider in addition to the type of attribution model.

With regards to the attribution window, you should ask yourself the following questions:

  1. What is my spot cadence? (i.e. how often do you re-air the spots when they are running? Every 20 minutes, once an hour?) For the best accuracy you should choose a window that is shorter than your typical cadence.
  2. Do my spots tend to run at the beginning of a show, or towards the end? Sometimes customers will defer doing any serious web surfing until their show ends.
  3. How memorable is my spot? Do you think viewers will forget about it very quickly, or will it stick with them for a while?

Fractionalization is how we share credit for a web response between spots that ran concurrently. Even Fractionalization gives equal credit to all the spots that were running around the same time. Graph fractionalization uses a proprietary algorithm to take into account the impact some networks and dayparts (and other spot characteristics) may have had over other concurrent spot placements based on Clarivoy’s real-world observations of those characteristics collected over several years of analyzing hundreds of thousands of automotive TV spots.

Q: Can you give us a sneak peak of what new and exciting features might be on the horizon for Clarivoy’s Multi-Touch Sales Attribution solution?

We are looking at non-linear models for TV Attribution, which might further approximate the real response curve. We are looking at offering much finer granularity in window selection. We are also looking at more sophisticated models for Multi-touch sales than the symmetrical parabolic model we currently offer as our top-of-the-line model. We have some new visualizations in the works and more data dimensions we’ll be adding into the mix as well.

We hope to continue to lead the industry by providing the most accurate and most insightful views possible, enabling our clients to dive as deep as they need into the details of their marketing successes so they can be even better and more successful in the years to come.


ALL NEWS