How to Hack Correlation Between Share of Voice & Apple Search Ads Search Popularity?

Search Popularity & App Store impressions

Keyword selection is vital for your apps’ visibility in a competitive environment. To get the maximum Share of Voice, it makes sense to consider the App Store keyword search volume. Only if keywords are already searched by users can they potentially drive more traffic.

One of the ways to assess the App Store search volume is to check Apple Search Ads Search Popularity scores. This indicator shows the share of users who entered a search query in this or that storefront. 

In Apple Search Ads, the score is marked by the blue dots, where the more dots, the bigger popularity. However, with some tools like Keyword Popularity Checker by SplitMetrics Apple Search Popularity index is visualized with the real numbers ranging between 5 and 100.

Indeed, this range can help roughly estimate popularity. But can we use it to make more accurate predictions on the available App Store search volume for a keyword? 

We’ve teamed up with Merlin Penny, Data Science and Engineering Lead at Phiture, to estimate visibility for keywords. Based on SearchAdsHQ data, we’ve conducted research to answer the question: what is the relationship between Apple Search Ads Search Popularity and maximum impressions for the maximized Share of Voice (SOV) on the US App Store?

Here’s a short summary of the research findings for the US App Store:

1. Search Popularity and App Store impressions follow an exponential relationship:  

Max Impressions =  254.4443 * exp(0.0615 * Search Popularity)

For example, for a Search Popularity of 60 the maximum impressions are at least 10,000, while for a Search Popularity of 82 there are at least 40,000 impressions. In the table below, you can find some Search Popularity scores with the relevant estimated volume: 

Apple Search Popularity & Estimated App Store impressions2. Search terms with a Search Popularity of 99 can drive 500,000 impressions per day;

3. Apple’s Search Popularity lags behind impressions by about 4 days. In other words, the current status of impressions will be reflected in Search Popularity only in a while.   

The research is based on 30,805 search terms collected in the US App Store for the period from August to October 2019. We’ve gathered daily data with the search terms and Apple Search Ads impressions, and for each day and search term Search Popularity was added. All the above resulted in 315,993 observations that you can see below:

Search Popularity & App Store impressions_raw_data

This graph depicts the raw uncleaned data where:  

  • Most observations are distributed between the Search Popularity of 20 and 40; 
  • Over 90% of observations have less than 10,000 impressions;   
  • Impressions for the Search Popularity from 5 to 20 are close to zero.  

The graph doesn’t show the outliers with daily App Store impressions of up to 0.5 million. These are search terms with extremely high Apple Search Ads Search Popularity scores  – 98 or 99, such as “instagram” or “snapchat”.  

To get insights into how Apple’s Search Popularity index works, a qualitative analysis was performed. For three randomly picked search terms, daily data for impressions and Search Popularity were collected to get the graphs below:

2 - Search Popularity & App Store impressions_ST1
earch Popularity & App Store impressions_ST2
Search Popularity & App Store impressions_ST3

The observations above suggest that for each search term a spike in impressions is followed by an increase in Search Popularity shortly after, with the time lag of about 4 days. It is likely that Search Popularity is calculated as a moving average of up to 7 days, which is indicated by the adjustment period.  

The data obtained up to this research stage are unprocessed, so we can’t rely on them to draw conclusions. Before getting down to the analysis of the obtained data, the data cleaning was performed to eliminate outliers that can lead to a bad fit while building a model.  

Knowing about the lag, the Search Popularity results were shifted by 4 days:

Search Popularity_lag_removed

Next, for every Apple Search Popularity index unit the search term with the maximum number of impressions per day was picked. Taking the rolling median over three Search Popularity scores resulted in a very similar fit and removed the last observations that don’t follow the 4-day lag rule.

Having removed all the noise, we can see that the collection of data points follows an exponential curve relationship:  

Search Popularity & App Store impressions_final

From the analyzed data, Phiture has come up with the formula to calculate maximum impressions for a maximized Share of Voice on Apple Search Ads:  

 Max Impressions =  254.4443 * exp(0.0615 * Search Popularity)

By using this formula, it’s possible to estimate the available volume for a keyword if you have its Search Popularity available. This helps conclude that, for example:

  • The Search Popularity of 60 will result in about 10,000 impressions;
  • The Search Popularity of 82 will result in about 40,000 impressions.

The exponential relationship between App Store impressions and Search Popularity could be caused by the combination of the Search Popularity lag and other factors. To verify the results above, we’ve carried out a couple of robustness checks.     

Robustness check 1: using the median. 

Unlike the average, the median of a large data set eliminates the influence of prominent outliers. By using the raw data and taking the median impressions for every Apple Search Popularity index unit we’ve found out the exponential relationship as well: 

Search-Popularity-App-Store-impressions_median

Robustness check 2: test on another country.  

We’ve applied the same methodology to the data collected from the German storefront, which resulted in a strong exponential relationship as well. The curve, however, lies at lower levels of impressions, which is probably caused by insufficient data for the DE storefront. 

Search Popularity & App Store impressions_DE storefront

Apple Search Popularity index & Share of Voice: applying research findings in practice

1. The research has revealed that impressions and Apple Search Ads Search Popularity follow an exponential relationship:

Search Popularity & App Store impressions_final

From this graph, you can see that a strong exponential increase starts from around the Search Popularity of 40. So it makes sense to allocate more budget on keywords with the Search Popularity of 40 and above, as they can bring about a significant increase in impressions for a maximized Share of Voice. 

According to our findings, a keyword with the Search Popularity of 60 drives 10,000 impressions, while another one with the Search Popularity of 82 will result in 4 times more impressions – around 40,000. Thus, by choosing a keyword with even a slightly bigger Search Popularity (starting from 40) you can expect a considerable boost in your Share of Voice.    

On the other side, impressions for the Search Popularity from 5 to 20 are close to zero, and from 20 to 40 the growth in impressions is hardly seen. So, it’s reasonable to lower bids on keywords with the Search Popularity under 40

2. We’ve discovered that the maximum number of impressions (in the US storefront) for a Search Popularity score can be calculated using the formula: 

Max Impressions = 254.4443 * exp(0.0615 * Search Popularity)

By knowing the Search Popularity of Apple Search Ads keywords that you already bid on, you can understand the maximum possible number of impressions for a maximized Share of Voice. If you see fewer than the maximum, you can try increasing the bid depending on your budget. It can be that there’s more volume available for a keyword.

3. The research has revealed that Apple Search Ads Search Popularity lags behind impressions by about 4 days. In other words, the current Search Popularity may reflect the App Store keyword search volume of around 4 days ago, while its current impressions may have changed in either direction. 

This knowledge may be helpful to ASO practitioners. When you research keywords to put in the title, description and the other metadata, take into account not only Search Popularity but also keywords impressions. By doing so, you will make decisions based on the current state of things. 

Identifying search terms with a high seasonality is still important since these might bias the results.

For example, “FOX Sports Super 6”, a game that allows to predict the outcomes of sports events and win money, implies a high degree of seasonality. Temporarily, you may see high Search Popularity scores. This will make you conclude that the number of App Store impressions is very high, whereas the average impressions throughout the year will be relatively low. 

Here, scheduling bids for search terms with a high seasonality will definitely pay off.

Moreover, the peak period when you can benefit from search terms with a high seasonality may last under 4 days. If you rely on Search Popularity only, you may miss the opportunities to reach out to users. Pay attention to impressions as well, as they reflect the most actual status of popularity.

That’s it for now. We want to say thanks to Phiture for teaming up on this research. Hopefully, you’ll find the findings insightful, and they will come in useful as you do your Apple Search Ads and App Store optimization. 

If you have any questions or feedback on the research, feel free to share them in the comments. 

SearchAdsHQ is a platform to manage, optimize and automate your Apple Search Ads campaigns. Request a free demo at SearchAdsHQ.com.