Steering the Google Algorithm for App Campaigns on a limited budget

Cost per Free Trial reduced by


Cost per registration reduced by



Moving into April 2021, Germany remained a lower priority market for Clue, but they did still want to maintain a presence there. As such Germany was allocated a smaller monthly budget on Google App Campaigns. The goal remained a strong Cost per Subscription and actually we were keen to reduce this metric further from what we had seen the previous month.

We were at present optimising to Registration due to ongoing reduced budget allocation for Germany but now that we had a goal to reduce our Cost per Subscription further, we needed to revisit our strategy. Optimising to Registration over the past month was returning us a strong CPR, 58% better than the top line (across all channels) for Germany at the time, but conversion rate from Reg to Subscription was very low at 0.28% (compared to the top line figure of 0.47%), leaving us with a high Cost per Subscription. We were struggling to optimise creatives, headlines and descriptions to steer the algorithm and so needed another option.

Optimising to Subscription was off the cards because Google recommends a minimum daily budget of 10 times your estimated CPA. The figure that the previous month’s Cost per Subscription had fallen at meant the monthly required budget for a campaign with this objective would fall at 45 times the figure we had been allocated. We decided to trial combining several events along the conversion journey for our campaign objective, to give multiple success indicators to the Google algorithm, all whilst hitting the minimum daily event fires within budget.


There were many considerations leading up to this test, we know Google is laser-focused on its optimisation goal, we were not sure how the campaign would react to multiple event objectives, of varying frequency. 

Converting Clue customers from registration to subscription is naturally difficult due to a really strong free version of the app which you can access after registration. 

Ultimately our core challenge here was that optimising to Subscription – our main campaign goal – would require a monthly budget of 45 times the allocated monthly budget – reducing the campaign flight to 2 weeks would still pose a problem and the nature of Google App Campaigns lend themselves to an always-on strategy.


  • Our revised strategy to bring down the Cost per Subscription involved building a new Google App Campaign, optimising towards multiple goals, including; Registrations, Free Trials, and Subscriptions.
  • With this, came a need for a revised, appropriate Target CPA for the algorithm to optimise towards – accounting for the fact that we are now optimising towards multiple events. 
  • In order to come to this we divided the prior month’s total spend with the sum total of all of the actions outlined above. This target was much lower than the registration target and meant that the Google algorithm was seeing a significantly higher number of conversions a day for this campaign. 
  • We set this up and allowed the algorithm to use each of these event fires as measure of success, and ultimately predict what audience provided us with the most value. 
  • As we went on, performance began to improve and therefore so did the actual, overall CPA (cost divided by sum total of optimisation events) – therefore we began to slowly reduce the target CPA on the campaign in line with the actual CPA to maintain the downward trend on actual CPAs. 
  • Including each of these events allowed us to guarantee the minimum number of 10 daily event fires, even for a reduced daily budget, ensuring there were no negative constraints on the algorithm, but it also included Subscriptions (our ultimate end goal), as a measure of success.


  • April saw a slight reduction in budget from the month prior, and increasing market competition drove a 32% increase in CPM, both months served a similar amount of impressions in their respective German campaigns overall (roughly 200,000), which gives us a strong base for fair comparison of performance. 
  • The new strategy overall helped deliver the following improvements in all major KPIs; Cost per Registration reduced by 6.3%, Cost per Free Trial reduced by a massive 58.9% and Cost per Subscription was reduced by 4.6%.
  • We also know that as long as the campaign is not limited by learnings, Google App Campaigns do continue to improve with time as the algorithm gains more learnings from a continued influx of events.