User Acquisition KPI: the good, the bad and the ugly!
Working as a data specialist with an app development background in mobile games, I have seen the good the bad and the ugly of tracking KPIs on user acquisition- including once being asked to calculate a KPI based on how many Olympic swimming pools could be filled by potions popped (and it was a lot). So here is my personal guide to KPIs which I have used in the past to assess the quality of marketing for the app I’ve worked on, and common pitfalls I’ve seen when using these.
User acquisition
AKA digital marketing in the world of web
The vast majority of companies will track the first touch attribution or on occasions re-targeted/cross-promoted attribution. This makes the calculating cost per user/install and returns much easier to estimate compared to the work of multi-touch marketing.
In part, this is because in most cases the attribution being all about the install and the quality of the install compared to the CPI. This said, companies are getting more in-tune with multi-point attribution, and now often looks to re-engage or rotate app users across different apps around their eco-system to maximise the income on the initial spend to obtain the install.
The UA KPIs I used:
This is not a definitive list and often there are many breakdowns and views associated to User Acquisition metrics.
eCPM Estimated cost per Mille
In advertising spend, this is simply the cost per 1000 impressions. This is often impacted by the competitive bidding for your audience, so expect to pay more for those grade A payers. It is a great estimate of the base cost of your audience, and often if the eCPM increases your CPI follows suit. This could be down to a number of reasons but indicates you may want to change your targeting OR a competitor could be bidding for your audience against you.
One trick on certain networks is to ensure you are not bidding against yourself for the same users if you have multiple apps. You should also not view a high eCPM as inherently bad if you are also buying good quality users, who spend.
Managing eCPM is an art in itself and can be supported with machine learning however still often requires human input to really understand the root causes of changes especially when using multiple networks.
CPI/CPA Cost per install/user
Simply put — this is the cost of buying a user divided by the number of installs. It allows you to ensure your Lifetime Value will be enough to cover the cost of acquisition and therefore profitable as often analysts will calculate upper and lower bounds of what you can pay for your users in each audience.
Historically, iOS users always were more expensive to buy than Android users, however, in the past 4 or 5 years, this reached some parity due to the better quality of devices and leaps made in the Google Ads algorithm.
However, as with all things — this should not be measured in isolation, with some lower CPIs actually delivering less profit than higher, variances in targeting age and location, plus interests, all of which could impact your overall CPI.
CTR Click Through Rate
This is a great indicator of whether your targeting or ad is performing. A good click-through rate could reduce your CPI if your install rate is also high. I know lots of rates!!
If your CTR is low you can often establish whether your targeting is off, a trick to assess if it is creative or targeting, is to use a creative which generally works with your audiences and test this on new targeting, if you still see low CTR then look at your targeting. Likewise your creative could be generating low clicks so test alongside a benchmark creative you know works.
A creative can also be generating high CTR which could give warning signs if the install rate is low. If this is happening, make sure your creative and text are not misleading, and also ensure it is connected to the right app (trust me this does happen).
Most networks will optimise good performing adverts so the impression volumes will help also give a good idea of performance (hint if your relevancy score on Facebook is low — delete !!)
Install from click
The third of the golden triangle of working out your CPI — install rate. This gives a strong indication of your app landing performance, and seriously listen to this number. If your install rate is low you can end up wasting thousands of dollars and the effort of your UA managers working hard to optimise the campaigns. In previous roles, we have saved huge sums by optimising our icons (seriously I mean HUGE sums by the icon optimisation alone), and the images and text on your landing page.
In addition, Android’s play console and iTunes developer console provide tools to help you run your App Store Optimisation, which can ultimately save an app.
Install from impressions
One last sneaky UA KPI I recommend tracking is installs from impressions rate. This gives a holistic view of your user install journey without the noise of the issues impacting the UA KPIs, if this changes you should be looking at your CTR and install rates immediately. The reason for tracking this is due to increased CTR may decrease your install rate and vice versa.
There are many other KPIs that will support your user acquisition including Customer Acquisition Cost (CAC), Return on Investment (ROI), Retention and Life Time Value (LTV) which I will be talking about in detail in my next article.
SO where can you get this data?
As mentioned previously the play store and the AppStore connect all provide great landing page insights.
In addition to these, your ad providers will provide insights on the core KPIs such as reach, impressions and clicks.
Ad mediators such as Adjust and Appflier offer ad mediation services which can then be connected via callbacks on unique identifiers to enrich your BigQuery Firebase data.
In addition, there are some great Python packages that help extraction of detailed data which can then be ingested into your data warehouse, which allows you to be the continuous reporting and automate tracking of your multiple user acquisition sources.
