E-commerce: Calculating the profitability of TV advertising
Via time series decomposition
Any business that does online marketing has probably become used to the fact that everything – even the success of marketing activities – is measurable. The times are gone when huge sums were spent on ads and no one really knew whether they got noticed or who noticed them. One does not want to give up the accustomed comfort of online marketing, especially when offline advertising channels such as TV command six- and seven-figure budgets.
Many providers now automatically collect information on when and where a TV spot was broadcast and connect this information with the program’s media data. This gives the impression of knowing who might have received the advertising messages. But especially in performance marketing, it is about more than this – it is about producing a measurable (re)action. And this is what separates the wheat from the chaff. The procedures that measure the effect of TV advertising often do not include more than a simple baseline correction. Here the traffic on the website is divided into two components. Methods such as averaging or time interval comparisons are used to determine how many visitors were actually expected in a specific time interval (so-called baseline). The difference between this and the actual number of visitors is then attributed to a TV spot broadcast at just this time. If one looks at the data more closely, one quickly discovers that this method adequately describes a number of simple cases but often fails to take into account the complexities inherent in practice.
Our flexible, mathematics-based approach aims to achieve much more. It also performs reliably when …
- the traffic on a website fluctuates significantly,
- the impact on the number of visitors to the website is small, or
- the advertising weight is heavier causing the windows of impact of the TV spots to overlap.
The procedure integrates seamlessly into online attribution, thus enabling companies to analyze and, most importantly, optimize the impact of TV spots in their accustomed environment.
Such an analysis requires data on the actual broadcasting times of the spots (ideally including the reach of the program positions). Tracking data is also needed in order to measure online traffic. In addition, this data can be enriched with budget data so as to make inferences about the cost-benefit relationship.
We first identify in which online channels (SEO, direct, etc.) the TV spots generate traffic. The analysis begins by modeling the so-called unbiased baseline, that is, pure online traffic without the traffic generated by TV advertising. The model used here forecasts the traffic for each relevant online marketing channel while taking into account all relevant factors. We can then derive the TV impact from this model. As long as the model has a high level of validity, it is possible to achieve a substantially higher accuracy. This means the procedure can also detect a comparatively weak TV impact (e.g., when measuring broadcasts during off-peak times or on special interest channels). The next step – which is also model-driven – involves filtering out the impact of the TV spots with overlapping windows of impact. Here we typically achieve a model validity of 70 to 80 percent (coefficient of determination: R²), so you can rely on the results. The analysis produces, in the end, data at a level of granularity that enables you to make connections between traffic (for each online marketing channel) and each spot broadcast.
In contrast to commonly used baseline correction methods, our procedure takes into account the complex reality of TV impact and also produces reliable results when there is heavy advertising weight. The high level of quality and our clients’ positive experiences justify the extra effort and expense.
After all, it’s ultimately about measuring the success of TV advertising as precisely as possible. You’re able to see which spots work best on which channels and in which time slots. This often reveals where substantial improvements can be made. Particularly in cases where specialized or niche products and/or services are being advertised, the analysis often identifies time slots where the audience has a great deal in common with your target group. And such information is worth its weight in gold: It provides the basis for optimizing your media plan while also enabling you to allocate your TV budget more effectively and efficiently.