TV Campaign Tracking — Measuring Direct Response

By now, a considerable number of e-commerce companies spend money on TV advertising campaigns. TV is an important mass medium and plays a significant role in terms of reach. But how can the effects of a TV advertising campaign be quantified? At least for e-commerce businesses, the direct response effect of TV spots can be measured. In this post, we’ll give you an overview of how it can be done.

Goals of TV Campaign Tracking

The main goal of TV campaign tracking is to quantify the impact of TV advertising on traffic, orders and revenue. Resulting data can be used to improve the effectiveness of TV campaigns. Measurements of the direct response effect of TV spots are based on the observation that mainly brand traffic increases significantly after a commercial is aired. This is a result of users browsing while, or after watching the commercial. We consider direct type-ins of the top-level domain as brand traffic, as well as searches for the respective SEA or SEO brand.

Approach to TV Campaign Tracking

The simplest way to measure direct response is to compare brand traffic after airing a commercial with the baseline, i.e. the “normal” traffic evolution. The delta within a defined timeframe is considered the “TV induced uplift”. Respective brand traffic (clicks and visitors) is flagged as “TV induced traffic” in the data warehouse. Conversions of these visitors within a couple of days after the TV induced visit are considered “TV induced conversions”. These conversions are associated with visits via cookies.

The reason for this methodology is that TV induced conversions cannot be clearly attributed. Thus, brand traffic is credited with the respective conversions.

This way of measurement provides us with precise, comparable performance data for every airing, such as visits, sign-ups, orders, revenue, cost per visit, cost per order, and a cost revenue ratio. The respective data is aggregated in order to assess the effectiveness of TV spots depending on channel, day of the week, time of the day, TV show, and other factors.

Requirements to measure the Direct Response Effect

So far, so good, but what is actually required to implement TV campaign tracking and measure the direct response effect of your commercials?

• You need granular, at best raw data, e.g. from web analytics tools.
• Your brand traffic must be flagged in the campaign tree.
• Sales data from the backend needs to be attributed to traffic.
• Precise data about airings is required (usually provided by third parties). A TV media plan is not accurate enough, since actual airing times mostly differ from planned times by at least several minutes. An alternative would be data integration via real-time spot recognition.

The respected timeframe after the airing varies from venture to venture and must be tested. Due to spot overlaps, long timeframes may cause vagueness or incoherence regarding resulting data. Overlaps of airings (see graph below) can be handled by weighting TV induced uplift according to the reach of the corresponding spots.

Pros and Cons of TV Campaign Tracking

Besides generally quantifying a TV campaign’s impact on your business, one of the main advantages of TV campaign tracking is the comparability of spots regarding all relevant dimensions. Immediate or near time recommendations for optimizing your media plan are provided since placements and motives of TV spots can be evaluated quantitatively. Furthermore, KPIs across businesses, industries, and markets are available.

There are, however, some disadvantages and critical aspects to be considered. The total effect of a TV campaign consists of direct and indirect or late response. While the direct response can be measured as described above, the indirect response cannot be covered by this methodology. Therefore, TV efficiency cannot be compared to other marketing channels.

In addition, there are several technical issues worth mentioning. First, the effort of implementation must be taken into account. Depending on your setup and the scope of your TV campaign, it might take quite a while until you can see first results. Especially the availability of accurate airing data might be an issue depending on the country where commercials are aired. And even if you implemented TV campaign tracking properly, there is still a risk of data errors.

Cross-device tracking can also be an issue, since visitors who look up your brand more than once are not counted as one but as new users when entering the website from different devices.

The main challenge is that brand traffic cannot be separated properly because SEO keywords are in most cases no longer provided. One option to get around the issue is to decide on the destination URL, depending on whether the traffic source of a visit is SEO brand or non-brand. If the landing page is the homepage, it would be SEO brand, otherwise non-brand. Though, not only brand keywords direct visitors to the homepage. As a consequence, all SEO traffic can be considered as TV induced traffic. Still, we think that flagging only the brand traffic, if possible, is closer to reality than taking all traffic into account for TV induced uplift.

In conclusion, direct response measurement is very useful for comparing performance of TV commercials in regards to various aspects, but not for an overall evaluation of TV as an advertising channel. It will not be able to provide you with a holistic view.

Outlook

With an eye on the future, there are a couple of measures which could be taken in order to obtain a complete picture of TV advertising as a marketing channel.

It would be worthwhile to flag every brand click with a percentage that indicates the probability that the click was induced by a TV campaign. This would increase the precision of the conversion attribution considerably compared to our current approach, where we flag concrete brand clicks randomly and according to the percentage of uplift above baseline. By now, we are testing this approach and we will post an update on this methodology.

Moreover, our current approach could be supplemented by taking into account the late response effect of a TV campaign. In this case, a certain percentage of brand traffic would be added as TV induced, assuming that TV has an effect on brand traffic on the website beyond the considered timeframe. Of course this approach depends on the quality of the assumptions regarding the percentage of TV induced brand traffic to be added as “late response”.

This post was written in collaboration with Evgeniya Anikina.