Opinion piece from Ronny Golan, CEO of ViewersLogic
Ever since the inception of the internet, TV advertisers have been obsessed with trying to measure the ‘performance’ of TV in a way they can online. The assumption being that consumers will see an ad, perform a very immediate action on their smartphone or tablet which can then be measured and attributed back to TV.
Recent technological breakthroughs have enabled brands, for the first time, to understand how users interact with TV ads and identify what actions they are performing after seeing an ad. This data has revealed facts that change the way we need to look at TV attribution – only 1% of the traffic attributed to TV comes within the first 5-10 minutes of watching an ad. For this reason, all of the models assisting with this task have been fundamentally flawed, as they have only been capturing around 1% of the response driven by TV. So how do we measure the remaining 99%?
The only way to measure this is by changing the way we capture these results and start measuring people, not media – or in other words, investing in the Holy Grail of consumer data, Single Source Data. By definition, Single-Source data is the measurement of TV and other media and marketing exposure, purchasing behaviour and location data over time for the same individual. It is passively collected across all of these touchpoints and seamlessly presented as one single source. No need to fuse/stitch data together, hurrah!
The status quo – where have brands gone wrong?
Until recently, the only way to measure TV effectiveness was to use a 5-min attribution model. This 5-minute attribution model calculates a baseline of website traffic – i.e. what traffic would be if there were no TV ads. Peaks above this baseline are attributed back to the last spot that was aired on TV (up to five minutes before the peak happened).
By condensing user behaviour into such a small window, brands are effectively ignoring how people behave in the days and weeks after exposure to TV ads. Now, with single-source data, we can measure the actual long-term response.
Using single-source data, we have analysed the behaviour of users who saw TV ads in multiple sectors and thereafter visited the website/app within a few weeks. We found only 1% of the visits happened within 5 minutes. Secondly and most critically, we see that response in a 5-minute window is driven by the frequency of ads viewed in the weeks prior. This means frequency had primed and nurtured the intent, the last spot was merely a trigger.
Therefore, what was up until recently considered state of the art (5-minute modelling) in fact is not logical (based on data), efficient (bang for buck) or effective (ad performance).
What does the data tell us?
Let’s compare two industries: automotive and online gifts. We used single-source data to understand:
- What percentage of users visited the site within 5 minutes of seeing a TV ad?
- What was the average number of ads seen by these users in the month before visiting, compared to users who visited within two weeks after their last TV exposure?
In the automotive sector, we can see real-world evidence that proves only around 1% of consumer traffic happens within the first 5 minutes of viewing a TV ad. It is even more interesting to see that these users saw on average 60% more ads in the previous month.
This means that brands are making decisions on how and where to spend money while neglecting 99% of their response data.
Automotive (5-minute window)
| Brand | % of traffic within 5 min of seeing a TV ad | Avg. frequency of ads viewed in 4 weeks prior | Avg. frequency of ads viewed by those visiting within 2 weeks of seeing their last ad |
| Cazoo | 1.64% | 10.19 | 6 |
| Webuyanycar | 1% | 11.3 | 7.6 |
The gifting sector is an even more extreme example, with real-world user data showing that less than 1% of traffic is generated within a 5-minute window of the last spot airing.
Online Gifts (5-minute window)
| Brand | % of traffic within 5 min of seeing a TV ad | Avg. frequency of ads viewed in 4 weeks prior | Avg. frequency of ads viewed by those visiting within 2 weeks of seeing their last ad |
| Moonpig | 0.3% | 10.9 | 7.8 |
| Funky Pigeon | 0.57% | 12.33 | 6.75 |
| Thortful | 0.7% | 13.45 | 7.22 |
Online Gifts (1-week window)
As can be seen in the above data, only around 0.5% of the traffic in this sector comes within 5 min of viewing an ad. What’s more, those users saw on average 70% more ads in the month before seeing the last ad than other users who visited a few weeks later.
Clearly, assessing the performance of TV advertising campaigns via a probabilistic 5-minute window is no longer fit for purpose.
What, then, is the solution?
The right way to measure TV: a 1-week attribution model using single-source data
Single Source Data help brands understand the cross-media impact of TV spend by looking at customer behaviour through empirical, deterministic, single-source data. No opinion, guesswork or assumptions – just real data from real people.
Instead of taking a snapshot of data within a 5-minute window, Single Source data shows that we should look at user actions over the course of an entire week.
For brands that want to maximise their TV spend, a 1-week attribution model provides visibility to the missing 99% of their consumer data, which means brands can make smarter, data-driven decisions.
