This is the second instalment of our series on Marketing Attribution. If you didn’t see our first one then read about Google Analytics as an attribution tool. Today, we are going to explore Spot Matching – a method used to measure the impact of TV advertising. We’ll look at the theory behind it, highlight its strengths and weaknesses, and compare it to econometric attribution. While offering granular attributions and a fast turnaround time, it has limitations in terms of measuring response only in a short window of time.
Spot-Matching vs. Econometrics
The theory
Spot-matching measures the uplift, typically in web response (e.g. site visits, registrations, sign-ups and so on), in the few minutes following the airing of a spot. The systems used to carry it out are usually fairly automated, taking feeds of the airtime data from industry systems and web activity, and then crunching them (often at night) to give you a set of attributions. And that’s what you get back from it, a list of spots with the number of responses, let’s say visits, that the algorithm has determined are due to each TV spot.
Of course, there’s a lot more to it than that, but that is the essence of how it works. The strengths are obvious: highly granular attributions, a high frequency of analysis and a fast turn-around time. The weaknesses are also pretty obvious. We’re only measuring response in a short window of time (minutes) after each spot airs. It’s TV-centric, although we’ve seen instances of radio spot-matching. It’s best suited to web response, although we’ve seen call centre data plumbed into spot-matching systems.
Contrasting with Econometric Attribution
You won’t be surprised to know that the number of responses (e.g. sign-ups) attributed to TV by spot matching is generally far lower than the number attributed by econometrics.
simply because the spot matching response window is so small (5-10 minutes) compared to the econometric response window (weeks to months). The result is that the measured Cost per Acquisition (CPA) for spot matching is many times higher than the econometrically-measured CPA.
The chart below shows 2 examples based on real models (but suitably anonymised). In Case 1, spot matching returned a CPA of £56.24, more than 7 times higher than the true, econometric CPA of £7.28. In Case 2, the disparity is smaller (4.9 times) but still orders of magnitude in size.
You can see how an investment case predicated on spot matching would look very different from one based on econometric measurement and that captures more of the media effect.
If you’d like to know more about Spot-Matching, then read our blog ‘A Guide to Spot Matching (By People Who Don’t Do It)’.
The Bottom Line
Spot-matching is great for fine-tuning your TV mix. But for most brands, it gives a misleadingly low attribution and, correspondingly, a misleading high CPA for TV. By omitting, in effect, the majority of TV response, it penalises TV unfairly and makes it look more expensive than it really is. Thus it’s not well-suited to proving the value of TV to the CFO, or for making decisions about your media mix. Don’t use it for that.
Look out for our next instalment on marketing attribution alternatives, we’ll be looking at claimed attribution i.e. How Did You Hear About Us? In the meantime, download our free econometric guide or drop us a line.
Philip Gaudoin
