Necessity of new approach
The first thing for bank managers to do was to take care of the “warm” audience – people, who already seek information about loans. For that ad campaigns in paid search ads were reviewed and strengthened. That allowed obtaining almost all traffic for the main search queries. Campaigns using the cost per click model targeting website users (remarketing) were also optimized.
After covering the “warm” audience further demand increase was possible using media advertising (video and banners). For maximal efficiency of these tools it was decided to use new, for the Kazakhstan, market tools developed by Google — Display & Video 360.
Configuring new approach
As a starting point, the target audience and its needs was defined in depth. Marketing messages were segmented for people who are looking for funds for:
- a wedding;
- and a broad message to reach people outside of aforementioned groups.
Based on created audience segmentation videos and banners were divided into:
- General group:
- In-market audiences → Business services.
- In-market audiences → Consumer electronics.
- In-market audiences → Financial services.
- Custom Intent by keywords → Branded.
- Custom Intent by keywords → Consumer credit.
- Custom Intent by keywords → Competitors.
- People looking for repairs:
- In-market audiences → Home and Garden.
- In-market audiences → Home and garden / Home appliances.
- People looking for funds for a wedding:
- In-market audiences → Gifts and Occasions / wedding Planning.
- Custom Intent by keywords → Preparing for the wedding.
- People looking for vacation funds:
- In-market audiences → Travel.
The results of such an approach became visible literally from the first week of its engagement.
Improving the campaign
Getting desired positive results could be the ending of this case. But managers understood that evaluation of video and banner campaigns only by clicks is not quite correct. While loan applications number increased, the received clicks conversions from media formats were not pleasing. Managers asked for newage. expertise.
From newage. we got a recommendation to add post-view data that was received through Google Campaign Manager. This solution allows using data from cross-device conversions in addition to cookie-matching (linking to an entry in the user’s cookies).
Taking into account post-view data generated a different conclusion.
Initially, bank managers needed to understand the data structure and how the user generally interacts with the brand. To do this, we have studied how users behave after a click during the first, second, and so on days (after ad exposure):
Adding to the previous chart data on users who come to the website after ad exposure (post-view), we see a completely different picture:
Conclusions at this stage:
- it is incorrect to consider only clicks when evaluating media ads, since in our case this was at most 1/10 of the useful actions users did on our website after contacting the ad. Rest 89% of performed on our website actions after the ad exposure were post-view.
- the chart does not contain cross-device conversion data. The Auditor doesn’t allow us to present data in such a way.
- evaluation in the standard post-click model will lead to a typical survivorship bias with conclusions for the entire campaign that take into account only those who clicked.
- the media long tail is quite extensive, and we can see that its main effect is present in the first 5-8 days after contact with the ad.
Moreover, in this advertising campaign, we also evaluated cross-device data. It was important for us to evaluate users who could see ads on one device, and make the final conversion on another. For example, from smartphone to desktop computer, and any other similar bundle. This is important, given that we used YouTube, which had a lot of ad servings in the app.
Final data distribution looks like this:
Such an analysis added to the funnel a total of 92% of the data. At the same time, we see that at each stage of the funnel, the share of each stage of the funnel may differ, which should be taken into account in the future.
The post-view actions dynamics is very different from the dynamics of post-click. It is more smoothed, and, indeed, the user may be not immediately interested in the product/service and visit the website after a while. This is very relevant for banking services advertising.
The most interesting thing — a lot of data is good, but how can it be used and how can it help the marketer?
The conclusion we could make using traditional market analyzing methods for a media campaign, namely by click:
Here are the conclusions we were able to draw after Holistic Analysis with Display & Video 360 was connected:
We think it is clearly visible how much we could have been wrong and how our priorities are changing
Using this data, we quickly identified which segments were not working and which were working. For example: being in the middle of the advertising campaign, we realized that the message “Take a loan for the wedding” did not find such amount of response as, for example, the message “Take a loan for a vacation”
So we disabled the first one and strengthened the last one.
It is also important to evaluate YouTube. If this platform is considered using a standard approach through clicks, it may seem that its use is ineffective. The reason for this is the high percentage of video views on YouTube in the mobile app. As a result, it is not possible to link the user’s video viewing with his further visit to the website via the browser on their smartphone or desktop.
But if we add to the picture the data from Google Campaign Manager, which can track user interactions across different devices (cross-device), we can see that our placements on YouTube are not inferior to a number of banner placements, and even surpass some of them in terms of conversions percentage.
We were able to make significantly more conclusions and insights using this data. We could determine the effective frequency, the conversion chain, which channel was used for this media long tail to return, and much more.