Nowadays the TV screen is competing more and more with other devices for the viewers’ attention. Only 19% of those polled of the Deloitte Media Survey 2015 are focusing solely on the rolling programme. This trend is in favour of the usage of second screens, involving PCs, tablets and mobile phones. Surfing the web (40%), social networking (29%) and e-mailing, as well as texting (31%) and online shopping (15%) while watching TV have greatly increased. This can be thoroughly seen as an opportunity for performance-driven TV advertisers, especially for those who are settled in the e-commerce sector (already every third TV ad stems from an e-commerce company, according to a 2015 XAD analysis). The usage of second screens allows to measure the direct effect that the TV ad had actually on the audience.
If the usage pattern on a second screen can be directly allocated to the TV impulse, an indicator can be obtained, how successful a TV ad had been appointed, respectively which effect it has caused. Hence the measurement of the advertising impact indicates, if the TV advertising budget was allocated effectively to the marketing goals. The most important indicators are the traffic on the website caused by the TV ad, as well as the conversion rate.
Measuring the TV effects on the website traffic and the conversion rate requires tracking data exact to the second, an exact capture of the website traffic and conversion, as well as an algorithm, which matches airing data and traffic and qualifies the resulting TV effect. What might sound simple actually brings some challenges in the realization.
Why the exact TV airing data matter
It seems reasonable to use the airing information provided by the TV stations and agencies for matching the traffic and airing data. But it should be noted that these media schedules often differ widely from the actual broadcasting time. This leads to a higher error margin at the correct assignment of TV effects. XAD in contrast captures each ad by itself. And that is the reason why XAD provides exact per-second information on the channel, the advertising environment and the ad break’s position an advertisement has actually been aired on. Those data are daily and automatized loaded to the XAD analysis tool TV2Web Impact.
Traffic data give some indication about the TV audience’s customer journey
The required traffic data can be determined by an open source traffic analysis tool which easily built-in (e.g. by Piwik). Alternatively, those data can be imported from already existing web analysis tools. Visit-IDs, time stamps of the first action, user-IDs, event-IDs, open event time stamps as well as event data, for example of the shopping cart are characteristic values. Another important information is the traffic source that leads the TV triggered visitor to the target page. The most important sources for TV triggered search are paid and organic search as well as entering an URL directly into the browser (in most ads the URL is named or at least displayed). The analysis of the traffic sources allows to draw conclusions on the TV audience’s customer journey.
The centrepiece – the algorithm
The algorithm matches the above-mentioned data sources and calculates the website traffic caused by individual TV broadcasts. The first step is to determine the so called baseline, the expected traffic without strong external triggers like TV ads. There are various strategies to calculate the baseline. For this purpose, XAD applies methods like floating average and linear regression. The period for the value assessment can be freely chosen. 10 minutes before airing an ad has been proven to be workable. The TV effect results from the difference, the so called uplift, between the baseline and the actually observed traffic. The uplift is normally measured in a time interval of 7 – 15 minutes after the broadcast has started.
Identification of the TV audience
At the moment, the greatest challenge is to assign each visitor to a TV broadcast with 100 % certainty. Probability calculations must be sufficient yet. By means of the already described uplift calculation, a probability of being triggered by a TV broadcast can be allocated to each visitor.
The challenge of overlapping
The fact that TV ads are simultaneously aired within 5 minutes on different TV stations is nothing new and that issues another challenge to the algorithm. To evaluate the performance of the broadcast on different TV stations correctly, the algorithm has to assign the individual TV effects to the right broadcasts. In this context, there are two different methods of the analysing tools currently on the market. Some utilise the channel reach and, depending on that reach, match the percentage of the TV triggered website visitors to the respective TV station. Thus, TV broadcasts with a high reach obtain a higher proportion of the TV effects than broadcasts on special-interest stations with less reach. This seems to be a plausible solution at the first sight, but using this method, the specific target group of the advertised product is failed to take into consideration.
XAD has experienced that broadcasting on a small special-interest station can lead to a higher increase of traffic, if an ad reaches the right target group. A lipstick ad, for example, rather reaches its’ target group on the women’s channel SIXX than if being aired on one of the large TV stations during a football broadcast. Ultimately, it depends on a target-group-specific reach. This is why XAD’s algorithm uses empirical values taken from similar placed TV ads (similar combinations of stations / time /programme) and the resulting effects on the website traffic. As a general rule, the more data are held, the more precise is the allocation.
Performance is everything
For measuring a TV ads performance and the consequent optimisation of TV media strategies, the mere determination of the TV effect per broadcast is not sufficient. It is crucial to find the cause of performance differences between ads.
The decisive measured values while analysing a TV ad’s performance are hence the cost per visit (CPV) and the cost per order (CPO) on various TV stations, at different times and in various programme environments and genres. Only after analysing those performance indicators it is clear how efficient a TV campaign has been. To XAD’s experience values, media budgets can be optimised up to 30 % after analysing the data.