Introducing: Project Think Tank
On June 27th, Adjust launched Project Think Tank. Project Think Tank brings together some of the brightest minds across a variety of different verticals to tackle problems and discuss potential features in mobile marketing.
We’re coming up with ways to help Adjust’s roadmap reflect our end users’ needs and goals in a setting that can best be described as a happy hour brainstorm. For Project Think Tank we wanted a relaxed environment - think food and cocktails - where ideas can come to life and thoughts can be not only shared, but challenged.
For our Think Tank launch, we hosted eight different leaders in mobile at the San Francsico Battery Club. Our agenda revolved around the best attribution methods for short-term user acquisition versus long-term user retention.
For our part, we prompted Think Tankers ahead of time to consider six different marketing ‘buzzwords’. We wanted to tackle these concepts and dive deep into what they might look like in practice as measurement tools - how can they help our clients reach their own goals, and what do industry leaders see as the positive outcomes or challenges of implementing them?
Here were some of the considerations for each buzzword:
- Multi-touch (or MTA) - Is it needed for install as well as re-engagement? What types of multi-touch methods do you use currently - U shape, positional, something else entirely? Which has been the most effective? How does multi-touch work from a network callback perspective?
- Re-engagement - When should you begin running re-engagement and which inventory is most effective? How does re-engagement affect LTV calculations?
- Temporary/Limited Attribution - For how long should a user's actions be attributed to a certain campaign?
- Cross-Device - When a user has multiple devices, how should attribution work? If a user installs your app on two different devices, should both engagement sources be rewarded? Which devices do you primarily see as becoming big for mobile advertising?
- View-Through versus Impression and Click - Is there a standard definition for both? Or are they interchangeable? How effective are the two, and how do they measure up against click engagements?
- First Source vs. Dynamic Source - In which cases do you need to know the user's original source (versus their dynamic source)? Which KPIs lend themselves towards the different sources? Which method do you prefer?
Senior Marketing Manager, DoorDash
There were too many outcomes to fit in the length of one post; however, here are some of the highlights we took away from our session:
- Aggregated data forces you to compute averages. Averages may look nice as high-level KPIs, but they shouldn’t be used for optimization, as averages in mobile advertising are a close cousin to guesses. Knowing that 10 users installed via creative ad 'kittycat' and that this creative spend was $50, does not mean that each user cost you $5. The actual user cost in this scenario could be $2, $8, $1, $9, $3, $7, $2, $8, $5, $5. So while yes, at a high level you may just want to know your effective (average) cost per install to allow you to budget acquisition accordingly, this leaves you wide open for mistakes when it comes to LTV and re-engagement strategies. Without knowing data at the device level, short term acquisition may refuel and refill as expected, but anything long term- cohorts, LTV, reengagement, retargeting will be adversely affected.
- For short term acquisition, it's key to test out new partners. In doing so, having a fraud prevention suite for protection and limited custom attribution settings is critical.
- Transparency shouldn't go both ways. While publishers need and expect certain data from networks, exchanges, etc. they shouldn't, in turn, be responsible for giving away all of their data and user activity for retention targeting.
- Last click is still the install attribution method publishers want. However, this isn't the case for re-engagements. When it comes to retention, clients feel that it's critical to first understand the user journey, knowing average frequency of engagement driving certain activities etc., and then work to be able to charge based on fractional in-app conversions.
- View-through is a scary topic, as it has many different definitions. Is it merely an impression, is it a two-second view, ten-second view, completed view? There's currently no way to monitor what type of engagement is sent to attribution providers. To see what’s working for multi-touch, it will be critical to know the exact type of engagement that’s occurring, in order to ensure the data isn't skewed.
Adjust plans on hosting many more Think Tanks in the future at a global scale. Next up: Berlin and LA. The topics will vary and the Think Tankers will change based on specialization. Be on the lookout. We might need you for our next Think Tank!