Apple Search Ads Study: CVR vs CPT as a Prediction Statistic

Incipia here with another informal, yet still certified nerdy study on Apple search ads.

This time, we started off exploring the question of whether Apple search ads campaign TTR (tap-through-rate) and CVR (conversion rate) predict any other performance statistics, including TTR/CVR, CPT (cost-per-tap), CPA (cost per download, because Apple doesn't count installs/launches) and ROAS (return on ad spend).

During the study though, our findings moved us to investigate CPT and its correlation with CPA and ROAS, which helped uncover a few important conclusions:

1. CPT is more a more effective lever than CVR for influencing CPA.
2. CPT is also a more effective lever for influencing ROAS than CVR.
3. TTR shows little influence on CVR.

Study Setup

• We analyzed 141 exact match keywords.
• We did not control for bidding.
• We pulled data from 2 weeks' time.
• We looked only at keywords with at least 1 impression.

Study Findings

As mentioned above, the two most interesting findings were on how to influence CPA and ROAS.

Recall that CPA = CPT / CVR, so either decreasing CPT or increasing CVR should yield a benefit to CPA. That said, CVR is by definition limited to 0-100%, while CPT is theoretically limitless, but in practice will top out somewhere, depending on competitive pricing pressures.

The R-squared of increasing CVR explaining the decrease in CPA was .2284, with a slope of -.0746 (slope is negative because lower CPAs are desirable).

That compares to the R-squared of decreasing CPT explaining the decrease in CPA of .25095, with a slope -.0772x.

This indicates that, in comparative terms, CPT's ability to influence CPA is about 10% stronger and ~3% more effective.

Looking deeper into this trend of the cost per tap being more important than those of being more efficient at converting every tap, we looked into the influence of CPT and CVR on ROAS. Again, we found that CPT bested CVR for affecting the desired outcome.

The R-squared of increasing CVR explaining the increase in ROAS was .0445, with a slope of .0374 (slope is positive because higher ROAS is desirable).

On the other hand, the R-squared of decreasing CPT explaining the increase in ROAS was .31616, with a slope .0516x.

Even when excluding the two outliers of the study, the findings of CPT > CVR held true. 

The R-squared of increasing CVR explaining the increase in ROAS became .04979, while the slope of -.0041, which indicates a negative correlation.

The R-squared of decreasing CPT explaining the increase in ROAS became .02093, with a slope .0016x.

Other findings:

• TTR cannot predict CVR

• TTR cannot predict CPA

• CVR cannot predict CPT

 

• TTR cannot predict CPT

• Considering outliers, TTR correlates with ROAS

 

While this dataset is too small to be trusted to reveal trends with full confidence, it was interesting to investigate nonetheless and lays the groundwork for a larger-scale study – we'll try again later once we have more data! If you have suggestions, comments, concerns or ideas for future studies, give us a buzz @ hello@incipia.co!

That's all for now, folks! Be sure to bookmark our blog, sign up to our email newsletter for new post updates and reach out if you're interested in working with us.

Incipia is a mobile app development and marketing agency that builds and markets apps for companies, with a specialty in high-quality, stable app development and keyword-based marketing strategy, such as App Store Optimization and Apple Search Ads. For post topics, feedback or business inquiries please contact us, or send an inquiry to hello@incipia.co.