Публикация Школы траблшутеров

How to increase conversion rate to mobile app installation?

Время чтения: 6 мин 40 сек
26 февраля 2024 г. Просмотров: 116

Траблшутинг | Олег Брагинский, Константин Алексеев

Oleg Braginsky, founder of the school of Troubleshooters, and his student Konstantin Alekseev were faced with the problem of low efficiency in attracting users to install a fintech application during advertising campaigns.

The restriction stopped us from buying more traffic and slowed down our plan to grow the product. So, we revamped our improvement plan and made focus on this task.

Advertising auction formula

Before delving into our actions, let's first examine how traffic is distributed among advertisers. This process is explained using the CPM (cost per thousand impressions) formula:

CPM = CTR х IR х CPI х 1’000
or
CPM = IPM x CPI,

If we count Impression Attribution, where’s:

  • InstallsPerMille           – number of installations per 1,000 impressions
  • ClickThroughRate      – click-through rate
  • InstallRate                  – installation rate
  • CPI                             – cost per installation.

The solution was to increase the IPM (impressions per mille), as adjusting this parameter directly impacts the advertiser's position in the auction.

Analysis of creatives and pages in Google/Apple app stores

We examined video materials and reviewed app store pages. The creatives and click-through rates (CTR) were satisfactory. However, the overall performance seemed slightly lacking. With the absence of an ASO manager, we found a workaround by focusing on CPP (Custom Product Page), which was accessible through Apple Search Ads (ASA) sources.

We conducted a thorough review of app store pages to identify the most effective ones in terms of conversion. This helped us understand what aspects of the product were most important to the audience in specific regions.

Since there was no dedicated App Store Optimization Manager, we prioritized optimizing the Custom Product Page, leveraging its availability through Apple Search Ads.

Decision

We needed to choose a traffic source by conducting tests to determine if it could process the &af_ios_store_cpp= or &af_android_store_csl= parameter in the tracking link. Moloco emerged as a suitable option with its A/B testing tool and the ability to incorporate specified parameters into the link. However, we discovered that not all sources connected to Moloco could handle this parameter.

This realization implied that if the test proved successful, future advertising campaigns would only need to target sources capable of processing this crucial parameter.

Discussion of solutions, idea defense, and roadmap

During the initial implementation phase, we convened a team meeting to outline the concept, assess resources, and plan our approach. Despite expecting diverse viewpoints, the response was limited to questioning the necessity of testing when details appeared clear.

Emphasizing our commitment to data-driven decision-making, we underscored the importance of testing to validate our strategies and prevent unfounded assumptions from guiding our actions. By aligning our approach with the company's core values, we emphasized the necessity of rigorous testing to ensure optimal outcomes and mitigate potential risks.

Test preparation

After examining custom pages and adopting pre-existing ones, we assessed the results and compiled a categorized list based on geographical presence. Analyzing advertising campaigns, we pinpointed the region with the lowest IPM.

Audience size calculation

To determine the audience size and cost for the test, we employed a Bayesian calculator, inputting data derived from the metrics of an ongoing advertising campaign. Parameters included the daily population size, conversion rate for the metric under consideration, expected percentage increase, and the number of variations. Although the calculator initially suggested one day for achieving statistical significance, we opted for a two-day duration to ensure a more conservative approach:

Test configuration:

We proceeded with caution:

  1. It is important that the settings of both ad groups lead only to sources that can process custom links: Applovin, Adx, Fyber, Unity, Inmobi, Vungle:

  1. Duplicate the advertising group until it is identical to the one involved in the campaign.
  2. Add a new tracking link to the copied advertising group.
  3. Create a tracking link with the parameter &af_ios_store_cpp= or &af_android_store_csl=.

We attempted to conduct the test on a campaign with the broadest reach, free from restrictions on Impression Frequency, Limit Ad Tracking (LAT), or advertising format.

Result

There was no need to wait until the end of the first day to assess the experiment's outcome. Almost immediately, we observed a remarkable 100% increase in install conversion, with a statistical significance of 97%. This outstanding result was achieved with only half of the budget allocated for research.

Scaling

The successful approach was expanded to other campaigns and geographies, evaluating the effectiveness of custom pages. Acknowledging that not all sources can accommodate desired tracking links, we ensured continuity by launching campaigns targeting compatible sources without missing out on valuable traffic.

Conclusion

Understanding the functionality of resources and identifying potential structural formulas is crucial. By exerting influence on individual elements, we can impact the entire system. Diligent attention to detail will be necessary for success in this endeavor