With the release of the book Appillionaire, the debates as to whether appstores are more of a casino than a goldmine are coming back to the front pages. This media craze is also fanned by mobile web advocates stating that appstores are now too full! “Crowded”, however, means different things to different people. For some it’s just the number of apps; for some it’s the number of developers, the chance to stand out and the success ratio; for others it’s the ability to “search” through the massive number of applications; and for others still, the ultimate touchstone is revenues! In this month’s in-depth analysis we’ll look at these various points and establish their veracity.
Numbers
The total number of apps in various stores are always fluctuating, but here’s a snapshot as of mid-November 2011:
Sources: WIP Appstore Catalogue, Royal Pingdom, FB data and estimates of monthly app add. 1 Figures vary as a Samsung representative recently announced 40k apps when the generally admitted figure is nearer to 13k 2 Website rather than apps 2010 figures
While the total number of apps within Apple and Google are impressive relative to other appstores, they are dwarfed by numbers from the web:
However there are two different mindsets when it comes to releasing products: the web mindset (quick development of a beta product, immediate publishing on the web, constantly update, fail fast) and the mobile mindset (long creation process, long publishing cycle through appstores, difficult upgrading system and keep products on the shelf). The Apple Appstore does typically fit in the mobile mindset with its painful approval process. The Android market though with its absence of approval clearly fits in the web mindset. This would explain why so many apps are taken out of the Android market, it’s the fail fast. Developers realize that the application is not good enough or get negative comments, which leads them to remove their app from the market, as can be seen in the chart below.
In terms of growth rate, the mobile app sector seems to be growing much faster than the web. Summing up this section it would appear that the mobile app space is still pretty empty compared to the web space. However absolute app or website numbers are meaningless, one needs to look at downloads and traffic to qualify crowdedness.
Average download
Figures in this field are rare and always subject to interpretation. WIP was one of the first companies to present average download data (i.e. downloads per app) as a way to compare appstores, but the practice has become more prevalent.
Research2Guidance, for example, proposed the following chart showing the average download per app on different platforms:
Many have judged this data to be dubious, as it differs from some of the public data available:
1 No public figure on the subject but anecdotally the Facebook app has just reached 1M downloads so we believe that the store has barely reached 1M daily downloads. 2 The total shipment of Bada phones has been 5M devices, Samsung announced 5.3M downloads from 300k devices in the UK. Optimistically this would give an average 9M monthly downloads over the 9 months of existence of the platform. 3 The previous assumption of 40k apps is maintained despite our doubt about the figure. 4 The situation is actually much brighter for Android developer as multiple app downloads occur outside of the Android Market. 5 2010 figures, 2.6T Page Views (PV) per month is the estimated worldwide web traffic. This gives an average per site of 11,000 monthly PV. Considering an average visit to be slightly below 3 PV, this gives roughly 4000 Visits a month, in with a 50% new visitor ratio and 2 visits a months for existing visitors gives 1300 new customers a month. 6 FaceBook claims 20M new app installs a day.
If two appstores were to have the same amount of users then intuitively the more crowded appstore will be the one with the larger number of apps. That is crowdedness correlates to the average number of download per app: the larger this number the less apps are competing for attention, the emptier the store. These stats, all based on publicly available data, give a completely different perspective on the crowdedness of various appstores. They show that the BlackBerry and Nokia stores are the least crowded, and that Windows Marketplace and Bada are the most crowded of all stores. Comparing these figures with the web is where the real surprise comes. Based on this study you have more chances of being spotted if you launch a new app than if you launch a new website, and even more so compared to launching a Facebook app, with apps gathering 2000 new users on average per month in the Apple Appstore vs 1300 new users a month on average for a website.
Most people though would believe the contrary and would say it’s easier to launch a new website. The main factor we would use to explain this is the fact that SEO and social media are relatively well known tactics used by agencies to help brands / entrepreneurs boost their online traffic or their Facebook app downloads.
Confronted with this data, any analytical person would object that studying averages are meaningless, and that what’s really telling of the developer experience is the median (in statistical terms this relates to the scenario experienced by half of users).
Median downloads and usage
In addition to median number of downloads, it would be useful to investigate the distribution of downloads among developers, aka how fat is the long tail? This is one of the hardest fields to investigate as data is rare and anecdotal evidence noisy. Even more complex to measure is the notion of usage and how much time people actually spend on the app. However there is probably more comparable data in this field than there is around download data, so we’ll use usage as a way to determine the median.
Nielsen usage data (for Android) suggests that the top 10 apps make up 43% of the time spent on apps. Which means that from a time usage perspective, the famous “long tail” is actually pretty short compared to the web. According to Alexa, the top 10 sites worldwide get less than 20% of the worldwide traffic (Google gets 5% of total internet traffic (based on page views), Facebook 5%, YouTube 4%, Yahoo and Baidu 1.5% each, Wikipedia / blogger.com / Windows Live / Twitter / Amazon each >0.5%). Similarly in Facebook, the top 10 external applications (not counting the Facebook mobile apps) make up ~20% of the total application MAU.
But time spent is probably not the best measure for mobile applications. After all, games were removed from the Nielsen usage stats above, as they consume much more time than other applications. In addition, the best mobile applications will specialize in doing one activity in a timely fashion (checking train times, for example) and would therefore be badly served by a time-based measurement. Similarly, mobile widgets make data available from the desktop and thus do not count into the app time.
Much better criteria would be the number of apps opened per month and frequency of opening. Here again, data is lacking:
The average iPhone user had 48 apps on his phone and 35 for Android users Average Internet users visit 89 websites a month (source).
This again confirms that the web has a much wider tail than mobile apps.
Revenues
However if usage is a much more accurate measure, both downloads and usage figures could be translated into revenue figures. Revenues could act as the ultimate touchstone whether they originate from sales or from advertising.
Anecdotal evidence has most mobile developers not able to generate enough revenues to sustain their business, despite the high eCPM and PPC touted by mobile advertising companies. But web startups are also numerous, and developers becoming millionaires are the exception rather than the rule. The fact is that there is no data available to judge whether a mobile startup has a better chance of success or failure than a web startup. This kind of data is not likely to emerge anytime soon either, as most technology startups will happily target web and mobile from a very early stage.
Discoverability
A last perspective on the question is whether crowdedness is not actually relative. Who cares if the web is crowded? What’s of real importance is the number of good results coming from a search. Taking a position on this subject is much more empirical. Here are a few emerging facts that can bring light onto the problem:
Increasingly people vilify web search as being less and less relevant despite Google’s effort to make it more local and more social. However, the principles behind SEO and getting a site to a high rank are pretty well known and many companies can help you do so.
Mobile appstores have traditionally relied on curation rather than search to promote apps. This is mainly due to the lack of metadata carried by each app, which. This results to app ranking algorithms being much more simplistic than a search engine algorithm.
As a consequence, the techniques used to help an app appear in the top rankings are pretty limited, not well understood and rather expensive (advertising, reviews by magazines, app aggregator, reviews by appstore etc.).
Therefore, the number of independent appstores and amount of website-curated app content is increasing as the major appstores fail to cater for specialist subjects (e.g., the enterprise) or geographies, whether because of a poor app offering or lack of payment options.
With the increasing number of curated stores opening, one could imagine a system similar to SEO that uses backlinks to an app to judge the actual quality of the app, pointing to an app search future that would resemble the one of web search.