Case Study - DeNA
Offering games for mobile devices since 2006, DeNA has since accumulated invaluable specialist know-how in development, operations and user acquisition. Today, the gaming giant boasts an impressive collection of games across a range of genres. They also work in collaboration with other companies and partner developers, extending their focus beyond internally developed/operated titles.
We recently interviewed Yuichi Fukada, who has been taking care of DeNA’s ad-data environment for years. We wanted to learn a bit about the reasons DeNA chose Adjust, how their data management works and what their best practices for working with Adjust’s tools are.
Why DeNA chose Adjust
Before working with Adjust, DeNA's games division tracked their apps through the Mobage SDK — their own gaming platform — along with the SDK of another company. When the company eventually decided to promote their apps overseas, they made the move to the Adjust SDK to track performance going forward. With many of their global agents and DeNA's US subsidiary already using Adjust, it was a seamless transition, Fukada recalls.
Different SDKs have different ad campaign processes. If you have multiple apps in operation, unifying the SDK will make the entire process smoother. DeNA initially had some difficulties unifying the SDK for multiple apps, and tracking via cookies was already prohibited at this stage - so they focused on using the tracking data from Adjust.
Fukada is now often approached for support by other departments and divisions including entertainment, sports and automotive. "Using the same tools makes it easier to support other divisions when preparing the data environment for digital marketing and infrastructure for campaigns - it also offers a big advantage to data managers," he said.
Sharing the latest data with the whole team
At DeNA, they forward raw data from Adjust to their database for analysis. The latest data gets automatically aggregated by the dedicated data analysis unit and is then shared internally via their BI tool.
Their data analysis team also uses data obtained from sources other than Adjust. For example, they obtain cost data from agencies to check the total amount of budget being used - including media costs and margins.
In addition, at DeNA, domestic and overseas teams operate in-house and work with agencies, individually determining the operation processes they deem most efficient. When several teams operate one app, they do not ask everyone to produce data sets with the same granularity and accuracy, but to use the data acquired from Adjust, and to then share the processed data with other departments.
Analysis flow for creatives
What’s really interesting about ad analysis at DeNA is the flow in place to assist marketers in making decisions regarding creatives. According to Fukada, they generate reports that include the images and videos used in ads along with the aggregated data so that marketers and creators can see which creatives are performing best.
By internally storing the data layers such as creative data, networks, and campaigns made to track ads, and aggregating them with raw data acquired from Adjust after serving the ad, they can identify which creative drove the most traffic.
There are various people involved in this flow including creators and staff responsible for data submission - not just marketers and analysts. It’s important that the job can be done effectively by a large range of team members, ensuring that data accuracy is always maintained.
Adjust's app tracking allows clients to generate URLs for tracking, and to easily set parameters from within a spreadsheet. The simplicity of Adjust’s features is a key driver in keeping DeNA’s data analysis processes lean and effective.
Changing the data view depending on KPI
Fukada emphasizes that understanding which KPI is used as the basis for campaign measurement is fundamental in data analysis. Since there are various definitions for concepts such as the number of devices, number of accounts, and number of users, the main KPI will differ depending on the timing and purpose of analysis. In Japan, there is also the concept of ‘risemara,’ which can be translated as ‘reset marathon.’ This describes games with a mechanic where the player is rewarded with a random character after a certain level of progress. If the user doesn’t get a sufficiently powerful character, they will often uninstall and start the process again until they roll a good character.
When tracking an event like the reset marathon, an SDK that can maintain accuracy of attribution — like the one offered by Adjust — should be used so that installs that take place during a reset marathon will not be considered organic. DeNA are able to determine installs and reattributions based on their own definitions by aggregating session data from Adjust. By acquiring data within a longer reattribution window than the actual evaluation period, and sending postback using the period set in Adjust to media, the attribution window for evaluating ad performance internally can be adjusted for each data aggregation.
Predict impact with uninstall and reinstall tracking
Uninstall and reinstall tracking with Adjust offers an opportunity for users who have uninstalled an app to reinstall it and have their lifecycle and LTV boosted. Fukada says that the accuracy of LTV prediction has been significantly improved since DeNA adopted this function. For example, users who reinstalled the app showed positive CPI while CPI for those who did not reinstall was higher than the target. DeNA continually evaluates campaigns and makes adjustments to their budgets based on this tracking data.
In addition, Fukada says that they have been able to gain interesting insights from uninstall and reinstall tracking, including the revelation that LTV might be predictable depending on how users returned to the game. His team discovered that "there are cases where reattributions after reinstalls showed higher LTV than usual reattributions."
An example of this is the users who left the game after playing it for a while, but then started from the beginning again once they returned, as opposed to picking up where they left off. The attribution windows of titles that have been on the market for a few years demonstrate that users are often likely to reinstall after a period of inactivity.
DeNA's gaming ad campaigns are most commonly run for one to two weeks while in-game events are happening. As the first few days are especially important, they typically plan their budgets as early as possible. This is why they aim to predict the impact of an ad based on the assumed LTV provided by each source.
The accuracy of LTV prediction has been significantly improved since DeNA adopted the tracking function.
DeNA Co., Ltd.
In the app industry, the protection of user privacy is crucial, and becoming more important than ever before. In response to this, DeNA is currently exploring ways to plan and develop promotions while enhancing privacy. Fukada explains that "multi-channel analysis which combines data with granularity and meaning - such as data from the platform/SDK - will be a challenge going forward. It requires a tool which can integrate this in an effective and convenient manner.”
In the Adjust dashboard, only certain users with admin permissions can access raw data. This means that any other users are not permitted to download device information from the dashboard. For companies considering the implementation of an SDK, it is highly recommended to evaluate app tracking tools for security and user privacy to avoid data leakage.
Support from Adjust and future outlook
Fukada explained that he "is very satisfied with the detailed feedback" he has received from the Adjust support team. "For verification during the implementation stage, it would be much easier if there was a simple verification process that even users without knowledge of the Adjust SDK would be able to handle. It is very helpful, however, that Adjust offers great support for these technical issues."
DeNA is currently using the trial version of "Automate", Adjust's marketing automation solution. Fukada detailed that their current outlook is focused on increasing efficiency and improving results, "Considering that we’re currently handling ad operations with agencies as well as in-house, we’re working on making our overall process more efficient - and this will be the driving factor for any methods that we choose.”