[The contents of this page are from 2016. Many things have changed since then.]

10 Google Play Stats You Should Know #

For last couple of weeks I have been working on to crawl Google Play (US) to create a dataset of all the all the apps available in the platform. My intentions were rooted in App Store Optimization(ASO) and app marketing.

At first, I was scraping the data in multiple CSV files, but that seemed unmanageable after some time. So I switched on to PostgreSQL later. The second iteration of the dataset seems to be in a good shape now. No, it’s not complete yet! Nonetheless, it’s a good enough dataset of 1,118,620 apps available in Google Play U.S. store; good enough to get some insights from.

So, this morning I ran some SQL queries on the database to find answers for 10 simple questions, and here is what I found:

#1 How many developers/publishers are there on Google Play? The number seems to be 261,500+ in the U.S. store. On average every publisher has 4.3 apps on Google Play (U.S.).

#2 How many of the developers have secured a developer badge like ‘Top Developer’ or ‘Editor’s Choice’? Total number of developers with a badge is 494. Among them, 458 developers hold a ‘Top Developer’ badge each. Interestingly, there are 24 apps which sport a single character name!

#3 How many apps actually use any of the monetization techniques (that is, in-app purchase and premium app) offered by Google Play?

As you can see in the table, 79.4% of apps don’t use any monetization tool offered by Google Play (i.e. these apps are free and offers no in-app purchase). If you are thinking about why Google Play’s aggregate app revenues are still lower than iTunes App Store, then you should also take this into consideration.

And for the people who are still interested to play with paid apps, the median app price on Google Play is $1.50.

#4 How different apps use the 30 character limit for app name (since it’s deemed very important for ASO)? The median app name length is 18 characters. Around 60,250 or 5.4% apps use all 30 characters for a name.

#5 How different apps use the description field (character limit: 4,000)? In terms description, the median length is 566 characters.

#6 How many apps use video in Google Play to promote the app? In our database, only 118,751 apps use video in the app page, which amounts to just 10.6%. This is pretty low considering YouTube is deemed as the number two source for game discovery by U.S. mobile gamers.

#7 No of developers/publishers who have provided a physical address. 58,173 of the developers have provided physical address in the app page. This is approx. 22.2% in the U.S. store.

#8 How apps are doing in terms of user reviews? Approximately 193,900 or 17.3% apps have no user review on Google Play (U.S.). While there are 197 apps with more than 1 million user reviews. The median user review number for apps is just 8.

#9 How are the apps distributed in terms of filesize? 297 apps sports a file size of 1GB or more. Around 223,150 apps are less than 1 MB in size. Most of the other apps falls in between. The estimated median app size in Google Play is 2.3 MB.

#10 How many apps are optimized for different devices? Only 12,140 of 1,118,620 apps in our dataset is optimized for different Android devices. That’s just 1.1% of the apps.

Are You Targeting the Right App Category in Google Play? #

Few days ago, when I shared “10 Google Play App Stats You Should Know” in couple of app marketing groups, Per Haglund from the App Entrepreneurs and Marketers Group requested me to share how much apps are there in each category in Google Play. I believe he was trying to get an idea about the competition and lucrativeness in each category. If that’s the case, then merely knowing the number of apps in each category would portray half a picture. These numbers would only tell which categories the developers/publishers are interested in, telling almost nothing about user interests.

The question is, how we can measure user interest. We can use the following metrics.

Considering the data I have at my disposal, I’ve decided to use a mix of the following metrics.

The first part is concerned more about category lucrativeness. In this part I measured the relative distribution of apps in each category, apps from top developers in each category, reviews in each category etc. All of the relativity in this part is measured against the total number of apps or reviews in the dataset. For example, when I calculated the percentage measure of how much apps in ‘Action’ category are from top developers I divided the number of apps from top developers with the total number apps in the dataset (that’s 1,118,620). So let’s take a look at the table.

Did you notice in the above table that ‘Education’ category has the highest portion of apps in Google Play? About 8.9%! But user reviews in this category is only 1.1% of all reviews. This category is probably over-heated with developer competition, but not interesting enough to app users. Take a look at the “tools’ category, which have the highest percentage of user reviews (9.3%) and holds 7.1% of all apps in Google Play.

Now, lets find out find which category attracts the highest number of top developers. It’s ‘Puzzle’. With 4.0% of apps and 4.4% of user reviews ‘Puzzle’ seems to be a nice balance. Immediately after ‘Puzzle’ there is ‘Arcade’. With only 3.1% of apps this category has 7.9% of user reviews and 85 top developers. This seems to be a lucrative category to explore. I think you have already got an idea about how to use the metrics. Now you can do your own analysis on this part. But you will also need to take a look at the second part before you reach any conclusion.

In the second part, I tried to analyze both lucrativeness and competition in each category. That’s why, in this part the relativity is measured against the aggregate sum of a metric in each category. For example, when calculating the number of reviews your app need to have to be above 95 percent of apps in a category, I used the review numbers exclusive to that category. So, let’s take take a look at the table first.

Let’s take a look at ‘Role Playing’ category. If you want to perform better than 95% percent of apps in this category, your app needs to have more than 33,306 reviews (highest in all categories).From the first table we know that this category sports only 0.2% of apps in Google Play. And, 44.7% of apps in this category offers in-app purchases. So, what do think?

Now lets take a look at the ‘Business’ category. Your ‘Business’ app will only need to score 82 reviews to be on the 95th percentile. But 34.8% apps in ‘Business’ category has no user reviews. And from the first table we can see that this category holds 5.8% of apps and sports 0.5% of reviews in Google Play. Seems to have a lot of competition with little user interest, right?

Mobile App Attribution for Indie Developers #

Earlier this week I shared some tips on how to fight fraudulent traffic in your CPI or CPA campaigns. The countermeasures largely depends on mobile app attribution data. Developers who are familiar with mobile app attribution will find these tips relevant. Only issue is, when I talk to indie developers, I still find many developers don’t have any clue about what mobile app attribution is, or why it’s necessary, let alone how mobile attribution can be used to harness their marketing campaigns. Hence, I decided to write a piece on mobile app attribution.

What is mobile app attribution? According to Tune, mobile app attribution “refers to the measurement of user events (such as an app install, repeat app launch, level completion, or in-app purchase) that are a result of marketing activity.”

Okay, but still not sure why do we need mobile app attribution? In simple terms, mobile app attribution is used to measure performance of user acquisition and re-targeting campaigns. Mobile app attribution lets app developers/marketers track where the users are coming from, what milestones are they achieving inside the app etc. If you want to measure the ROI of a campaign, or the LTV of users coming through a campaign, or compare across acquisition channels, then mobile app attribution is a must.

Let’s consider a scenario. You are running a paid app install campaign to get new users. How would you track whether a user came organically or through the install campaign? You are probably thinking about integrating the ad network SDK in the app for tracking installs. This will work only if you are running campaigns with a single ad network; when you run campaigns with multiple ad networks things will get messy. For example, if a user clicks on install ads served by multiple networks, then all the networks will claim the install and all of them will charge you for it. In addition, you need to integrate SDKs every time you decide to try out a new ad network. Hence, you need an independent 3rd party service to look over your app installs sources. That’s why mobile app attribution services came into being.

But why can’t we do it using app analytics? The cookie-based tracking technology that work in web doesn’t work in mobile apps in the same way. Mobile apps need specialized tracking solutions to make attribution work. Although the current versions of Google Analytics and iTunes App Analytics have some attribution friendly features, these are still far from achieving the kind of accuracy specialized mobile app attribution services provide.

Cool, but do we still need attribution if we don’t plan to run paid campaigns? The answer is, yes and no. If you are an Android developer, using Google Analytics, and never intend to run any paid campaigns, then you might get by without a commercial app attribution service. However, if you want to track user profile data with minimal effort and want to track your organic campaign performances easily, you should use a mobile app attribution service. (I would rather not go into the details behind my reasoning at this moment.)

How does it work? The best way to learn about how it works is to read through the documentation of mobile app attribution service providers. You can start here.

Which are the leading mobile attribution services?

Influencer Marketplaces and Tips For Beginners #

Couple of months ago, I started working on influencer marketing and the ROI was staggering! In the past months, I have tried a number of marketplaces for influencer marketing and a number of models. Based on these experiences I have put forth a list of influencer marketing marketplaces and a few tips for people who want to promote their apps through YouTube, Facebook, Instagram etc. for the first time.

Below are some more platforms you should check out.

Tribe - Follows a project-based pricing model and charges 20% commission on top of each purchase.

Channel Pages - Channel Pages is another platform geared toward YouTube influencers and marketers.

Sponsored Mentions - A platform to buy Twitter mentions.

Izea - Izea supports multitude of social platforms and is focused on corporate clients.

Speakr - Targeted towards large brands and understandably you’ll need a deep pocket to work with them.

Most influencers in the above marketplaces like to work in the project-based payment model. The above marketplaces are also built to support this payment model only. If you want to establish a business contract based on the affiliate model, you should ask influencers after you have established contact.

10 Paid User Acquisition Tips for Indie Developers #

Indie developers who are keen to test paid user acquisition (UA) channels often lack the necessary background in mobile advertising. This post is for them. I have shared some tips which might come in handy for first-time ad buyers.

Prerequisites

1. Before you start paid UA campaigns for your app, your app must be integrated with a mobile attribution service. If you are not sure why you should follow through, please read my blog post on mobile app attribution.

If you are planning to run campaigns on Facebook, you might need to take extra measures for attribution. Facebook works with only a limited number of mobile app attribution services, and most of these are not economically viable for indie developers.

2. You need to allocate a budget of at least $500 to run a test campaign on most networks. Some self-serve networks will let you run campaigns for even less. However, if you opt-in for a managed campaign, $500 is the lower bound of the spectrum for many networks. This threshold can go as high as $50,000 for a test campaign on some networks. Don’t get disheartened after seeing this, there are all kinds of networks in the market. You’ll find one or more that suits your budget and goals.

Now that we have talked about the pre-requisites, let’s dive into the pool.

1. Go for affiliate marketing There are two distinct models of marketing in mobile advertising. One is affiliate marketing and the other is non-affilate. In simple terms, affiliate marketing is the method where you pay for the result, not for the efforts. The good thing about this model is that the risk is managed by the affiliate, not the advertiser. For example, Cost Per Install (CPI), Cost Per Action (CPA), or Cost Per Engagement (CPE) ads fall under affiliate marketing. Because the advertiser pays only for installs or pre-defined actions by users, not for impressions (CPM) or clicks (CPC). The affiliates take the risks, giving them the stimulus to bring users at low costs (to keep it profitable for themselves). In the non-affiliate model, however, the scenario changes. The risk is now on the advertisers shoulders. Ad networks charge advertisers for sending users in their direction, regardless of whether these users convert or not. While experienced marketers can make wonders with such high risk-factors, this is not a path first-time advertisers or indie developers should choose. Affiliate marketing is much more suited for indie developers.

2. Start with incent ads Contrary to popular belief, incent ads can actually drive a lot of high value users towards app. I have worked with both incent and non-incent formats, and I believe the argument that non-incent ads are inherently superior in acquiring quality users compared to incent ads is deeply flawed. There is no straight answer, and it depends on a lot of factors that varies from app to app. Most importantly, incent ads are probably the best option you have if you’re low on budget.

3. Go for a CPA or CPE campaign early CPI ads are the de facto standard in mobile app user acquisition. However, CPA or CPE formats are the ones that can drive your marketing ROI to green with the least amount of investment. Below are couple of things you should remember when considering CPA or CPE campaigns.

5. Install volume or install quality? Payout rates and install volumes are proportional. The relationship is, however, non-linear. Nonetheless, you will see increase in your daily user influx as your payout rates increase significantly. Increase in your daily installs means increase your app ranking too. The higher you go in the ranking, the more noticeable you are to organic users. So, you now have a dilemma whether to go for a quick rise in install volume, or to take baby steps. If you have a budget constraint and have no experience in burst campaign, I would advise you to take the second route. Go for quality, instead of volume. Burst campaigns are waste of money if you can’t support it afterwards with continual marketing efforts.

If you decide to take baby steps toward a sustainable growth, you might want to optimize the user influx through paid channels by controlling your payout rates and by setting up a daily conversion cap.

6. Choose your target geos Don’t go on with a worldwide rampage if your target users lives in just a continent. Choose your target goes carefully before you start a campaign. Also remember to negotiate payout rates for each geo.

7. Track ROI and optimize your campaign(s) Regularly track your campaigns’ Return On Investment (ROI) and optimize your campaigns accordingly. Two simple measures that will come in handy here are Gross Profit and Average Revenue Per User (ARPU) for each campaign. Most affiliate networks support publisher or sub-publisher level optimization. Some would also let you apply multi-tier payout for different publishers. Use the above measures to optimize your campaign(s) on publisher or sub-publisher level.

8. Keep your eyes open for fraud Regardless of the ad format you choose, you might encounter fraudulent users coming through. Keep your eyes open for such activity and work with your affiliate network to prevent such fraud. You can read more about this on “Simple Ways to Detect and Fight CPI Fraud.”

9. Connect your accounts to App Annie If you are not familiar with App Annie, it’s a marketing data analytics service for apps. Connect your Google Play and/or iTunes accounts, along with your monetization and advertising accounts to App Annie. This will help you a lot in terms of easy access and analysis of all of your marketing data.

10. Test your campaign setup before going live Test your campaign setup thoroughly before going live with your campaign. This is actually a drill carried out jointly by the network and the advertiser. Pay close attention to the data points logged during the testing, and discuss with your partner network if you feel there are issues relating to the setup. This will help you avoid conflicts in future, if there are issues indeed in the campaign setup.

Simple Ways to Detect and Fight CPI Fraud #

Most of the people working the mobile marketing are aware of install frauds. This is a common problem for incent-CPI or CPA campaigns. Last year, a blog post by Mobile Attribution service provider AppsFlyer suggested that the estimated cost of fraud in mobile marketing is about $1.5 billion. Yeah, that’s right! Fraudsters rack up huge sums of money from mobile advertising.

But how do they do it? CPI campaigns are the de-facto standard in mobile advertising. And a significant part of installs come through incentivized traffic; i.e. the end users get incentives to complete install offers. These incentives come mostly as virtual currencies which the users can later use inside the originating app or can redeem for tangible rewards. Because of these, fraudsters target incent CPI traffic. They use a number of techniques to carry out these fraudulent activities, but let’s not delve into the technicalities at this point.

Does that mean incent traffic is bad? Not necessarily. We see a lot of fraudulent users coming through non-incent traffic as well. Some say this is because many publishers (affiliates and app owners) push incent traffic to non-incent traffic. In my experience, it’s not that straight-forward. Okay, but how do we track such fraudulent installs or users? The most simple method is - through a mobile attribution service. Look into the install logs and match the device carrier information with IP location info. See if you can find any anomalies. For example, you may find user IP location to be in North America and device carrier to be in another location. Once you identify such cases try checking some of these IPs randomly through who.is to find out if they belong to any VPN or proxy service provider. If the IP belongs to any hosting provider, then you can be pretty sure that the IP belongs to a VPN or proxy service.

But it may not be so straight-forward all the time. The device IPs may belong to mobile operators who provide internet services to the general population, as well as hosting services to enterprise customers. In such cases you might need another data source to confirm your hypothesis. If you are using a push notification service or a backend service which also tracks user location independent of your mobile attribution service, then match location data (along with device carrier data) from these two sources. It’ll tell you how accurate your device carrier data is.

However, the above methods have their pitfalls. What if the users are coming from a device without a SIM card or phone connection (such as tablets with only WiFi support). To address this issue, you can detect device IP vs install ratio. If too many installs are coming from a single IP, it might be fraud. You might to take some extreme measures as well, such as tracking user mobile number or device GPS data. There are a number of ways on how you can implement these, but be aware of not violating your users’ privacy.

Okay, great! We’ve learned how to detect fraudsters using some simple methods. But we still haven’t discussed the most important issue.

How do you we reduce such fraud? There are not that many ways to handle such fraud from the advertiser end, unless your ad network or ad exchange has proper countermeasures in place. Nonetheless, you can take the following measures to curb such activities.