In our group we’re proud to have gathered more than 80 of the most talented people within the field of programmatic advertising. One of these has, alongside his job as programmatic specialist, decided to examine the media buying strategy retargeting from a theoretical perspective. This is the first installment of the result of his work, where Anders Munkesø Kjærbøll weighs in on whether retargeting works – and if so, how and why.
With the large volumes of data generated by the digital consumer, advertisers are forced to follow suit. The data volumes make it possible to communicate intelligently with the consumer in real time – day by day, hour by hour and minute by minute – and not least measure the success of this communication. This poses a great challenge for the advertisers and the speed which they have to adapt to new contexts – and their way of conducting brand management.
Pssst! Are you in a hurry? We’ve highlighted the most important points if you’re looking for a quickread.
Retargeting has long been used in the industry by placing cookies on websites, also with regards to paid searches and paid social media. The next step is orchestrating the opportunities provided by retargeting – to merge business data with market data to create a connection between customer records, CRM and owned and bought media, whether communicating to existing customers or chasing new leads. The aim of this is to create a continuous customer journey across all touch points to ensure the business always appears relevant in its communication.
We’ll get to all that later, but first we will take a look at how we can (and should) measure the success of retargeting.
Retargeting is one of the most commonly used digital strategies. It was one of the first digital disciplines to use first-party data (from advertisers) to target the right users with the right message at the right time. The concept behind retargeting is simple; a person has shown interest in a website or an app but left it without performing a desired action such as making a purchase. To start a dialogue with this person through banner advertisement makes sense for several reasons (see these 15 examples), and retargeting has also proved to be extremely efficient from a ROI perspective.
But why is retargeting so efficient and arguably the best thing that has ever happened within programmatic buying?
An obvious question is, whether the targeted users have, in fact, already made up their minds to not commit to a purchase – or alternatively would return anyway to purchase the product at a later time? Or, in a worst case scenario, would they become annoyed by being followed by the same banner ads and change their mind? It would be relevant to examine what would have happened if the retargeting banner had not been shown.
To determine the added value it makes sense to compare the users that end up purchasing your product by looking at the ad interactions of users exposed to banners as opposed to users who bought the product without ad exposure. (See also Integral Ad Science).
Attribution is often made solely on the basis of the last banner interaction before a purchase, which rather than being a descriptive attribution is probably more of a convenient simplification of the available attribution options. Essentially, this indicator is used as a proxy to track a development and monitor the success of your retargeting campaign. It is important to note, though, that it is impossible to determine directly how a retargeting banner has caused a sale, based on the last interaction.
Whoever gets the click, gets the money. Google introduced the so-called last-click attribution, and for many advertisers in Denmark this attribution standard has crossed over from search to banner advertisement.
It is a very concrete way of documenting one type of effect, as a customer must have seen and reacted to a banner if it has generated a click and a resulting sale, but there is an inherent risk in adopting this means of attribution, as it involves comparing a click medium (search) with a visual medium (banner advertisement), where click rates are often around 3.0% for search compared to 0.06-0.15% for banners.
To this can be added the different behaviors of users who click on banners versus those who do not; some surveys of internet users have found that 8% of users on a news site generate 85% of all banner clicks (ComScore, 2009). In other words it’s far from certain that you’re communicating to the target audience if you only address users that click on banners.
Furthermore, optimizing a retargeting campaign that is measure by post-click poses a challenge. First of all the data is limited, as post-click conversion rates are much lower than post-view, and a campaign will not only need to target users from a conversion probability, but also from a click probability. This will naturally result in an optimization that leads to a high frequency of few users with high conversion and click probabilities, which is not ideal in terms of reaching the entire retargeting audience.
But it is one thing to assess the value of a retargeting banner, and quite another discussion whether retargeting creates value. Let’s take a closer look at this question.
Basically humans live mainly on autopilot. We do that to save ourselves from spending energy on processes that can run in the background, but this autopilot also blocks out a large swathe of the advertisement we are exposed to on a daily basis.
If you’ve recently rented a car, you’ll quite automatically start noticing similar cars in trafic. The same is the case for adverts, where we subconsciously will notice the brands we’ve recently come into contact with. This is called Reticular Activation System and plays an important role in explaining why retargeting banners get more attention than random adverts.
In fact, a large part of the effect of advertisement happens in the subconscious. Similarly, a high frequency of retargeting banners will create a positive effect in the subconscious. Several studies have been made of this phenomenon known as the mere exposure effect. The studies provide solid evidence for a correlation between frequency and liking/recognition. Take for example this simple eye.tracking experiment conducted by Bornstein & D’Agostino:
It is particularly interesting to note how subliminal exposures – 5 ms exposures that are only subconsciously perceived – outperform consciously seen exposures of half a second in terms of liking, whereas the opposite is the case for recognition. This throws an interesting perspective on the discussion of viewability, where the currently dominating standard is that a banner has to have been 50% in-screen for at least one continuous second in order to represent value. With these study results in mind, perhaps viewability should increasingly be seen as a way of measuring a banner’s opportunity to be see.
All things considered, retargeting plays an excellent role in digital media buying. Based on a consumer journey perspective it is vital to be top-of-mind when a consumer is close to deciding on a purchase, but a sale is not simply generated by showing the consumer the same pair of shoes 300 times, just because he or she clicked on them once. Retargeting mustn’t become a simple exercise in stalking; it must happen as part of a relevant and consistent communication with the consumer, or we could risk impacting the brand perception in a negative manner. This aspect of retargeting will be the subject of a future blog post.