In a recent blog post, our Head of Innovation, Tim van Rijt, explained the importance of “quality impressions”, which are defined as impressions that show the full creative message in view. This means that the exposure time of the impression must be the same or greater than the duration of the creative. In this blog post, we’ll introduce the concept of “quality users” as our answer to the industry challenge of cookie decay. We’ll explain why it’s important to serve quality impressions to quality users in order to hit both your awareness and performance goals.
Cookie decay, or cookie deletion, is an industry-wide challenge. In order to reap all the benefits from real-time bidding (RTB), you need real-time access to data, which is often stored in a cookie. A user is segmented into different audiences based on their cookie info, but for instance the frequency a user has been served a certain campaign is also stored in that cookie. From our own research, we discovered that almost half of the cookies can’t be found after 24 hours. This is in line with other research regarding cookie decay and correlates with the increasing popularity of the “incognito mode”, which is available in almost every browser. Many marketers accept this as a fact and communicate that half of the users delete their cookies. However, is this really true? A quick check amongst colleagues, friends and family showed that almost nobody deletes their cookie on purpose each and every day. But how can this correspond with the fact that we can’t find half of the cookies after 24 hours?
Cookie decay: an example
Imagine a group of 100 users, where 99 of these users never deletes their cookies (“non-deleters”), while one user is a frequent cookie deleter and deletes his/her cookie after every internet session. Assume that users have three internet sessions per day. After 30 days, the 99 “non-deleters” still have their original cookie, 99 cookies in total (one cookie for each user). The frequent cookie deleter, however, has had 90 unique cookies (30 days * 3 internet sessions), almost as much as the other 99 users in total. If we were to calculate the cookie decay in this example we would say that we’re only able to find 100 out of 189 cookies (99 cookies + 90 cookies). This means a cookie decay of 47%, while only 1% of the users actually delete their cookies.
This example illustrates that a small group of users is responsible for a large share of all the unique cookies. This is both positive and negative news. The positive part is that we can tell our client’s advertising story to 99 users without any cookie decay issues, the negative side is that the frequent cookie deleter is regarded as a new user each and every time they delete their cookies. This means that advertisers will compete with each other to serve this “new user” their impression and, in this example, this happens 90 times in a month. This advertising money can obviously be much better allocated in order to create awareness to other users who don’t delete their cookies, who we now name “quality users”.
This was the start for our data analyst department to use data in order to validate the above assumptions that we should focus on these “quality impressions” for our branding and performance-based campaigns.
Our hypothesis is that there is a difference between
users we can only track for a short amount of time (cookie deleters/non-quality users) and
users we can track for a longer amount of time (quality users)
In other words, can cookie lifetime be used as an indicator of performance? In order to provide an answer to the main problem, we need to define what performance is. To scope the research, we defined the cost per landing as our performance metric. This means that we investigate if there is a difference in the cost per landing between quality users and non-quality users.
Because we’re looking at user-level performance, it’s important that we divide the population into two groups based on a variable, in this case, cookie lifetime. All other factors should stay constant as much as possible. This way, effects on performance can be linked to the one variable that is different between the groups.
To check whether quality users truly are ‘of higher quality’, we set up a test in one of our campaigns. The campaign was divided into two identical parts, with the sole exception that part A targets quality users and part B targets non-quality users.
A few things stood out:
We found about seven times as many non-quality users then quality users. As stated before, this is mainly due to the fact that users that delete their cookies will have multiple cookies during the lifetime of a campaign. Even though it would appear that the contact frequencies are about the same in both groups, we are probably serving far too many impressions on users that frequently delete their cookies. This makes it extremely hard to manage (in-view) contact frequency in such a way that awareness KPI’s can be met or that the performance can be correctly measured. This basically comes down to having no lifetime frequency cap, which can be annoying for non-quality users and which is very ineffective for advertisers.
The conversion rate of quality users was about 5 times as high as the conversion rate of non-quality users. There are multiple possible reasons for this. One of them is that quality users can be tracked longer and we are able to map out a lot more of their conversions. Since we care a lot about mapping out the complete customer journey of a user, this is a big deal, both for advertisers, who want the most relevant users in their targeting, but also for the users themselves, who don’t want to be spammed with products they’ve already purchased.
The clickthrough rate is higher for quality users. The advanced strategies that advertisers use in their campaign targeting have more effect when the information they use for their targeting is more complete. Propositions appear to be more appealing to relevant users. The fact that advertisers have more information about quality users (because we’ve seen them before) means that the message they deliver will be much more relevant.
In short, it seems that quality users are much more relevant than non-quality users for both branding and performance purposes. By looking at users with a longer cookie lifetime, advertisers have a more complete picture of their customer journey and are able to define their relevancy much better.
https://www.bannerconnect.net/wp-content/uploads/2016/08/Alice-in-Dataland-02.jpg400400Alice Bezetthttps://bannerconnect.net/wp-content/uploads/2015/09/logo_bannerconnect_72.pngAlice Bezett2018-04-03 10:00:372018-05-16 17:11:29Regression Models for Trait Prediction Part II
https://www.bannerconnect.net/wp-content/uploads/2018/03/featured-image.jpg515515Tim van Rijthttps://bannerconnect.net/wp-content/uploads/2015/09/logo_bannerconnect_72.pngTim van Rijt2018-03-06 09:59:302018-05-17 09:39:23Quality Impressions Connect Your Creative To Your Media Buying