Copyright Lisa Clarke
There has been a lot of chatter on the Internets about unbundling of mobile apps. Tech media’s interest has been piqued by some larger brands jumping onto this bandwagon – examples include Microsoft, Twitter and Foursquare.
It’s not a new strategy. Featurephone apps companies used this for a long time. Several smartphone app publishers have had an owned & operated network of apps right from the beginning – examples include Zynga, King, Supercell, Smule and Outfit7. Incumbent media companies like CBS, News Corp, Viacom, Time Warner, Walt Disney and (in India) Times Internet have also had these networks for a while. Established Internet companies like Facebook and Google have had networks, built organically and through acquisition.
I think unbundling is a strategy that has not yet been applied with vigor in the emerging markets on smartphones. I think there are potentially disproportionate advantages to be had by unbundling in countries like India, in the short- to medium-term. Why is this? Because low device memory limits (typically less than 16 Gb), low bandwidth limits (mostly 2G) and relatively high bandwidth prices result in dramatic drops in conversion rates, download success rates and retention rates as app size increases. Also, in my opinion, discovery on the app stores is easier when there is a single focused value prop (kind of the approach that Whatsapp has taken with a singular focus on messaging).
Conversion rates drop with package size. Below is data from a global mobile analytics and advertising vendor. Data is global and is an average across all advertising products.
App Size | Conversion rate |
0-5MB | 1.00% |
5-50MB | 0.80% |
50MB+ | 0.42% |
Download success rate is nowhere near 100% (even on Android). The graph below is from China. I would imagine that the drop-off in India is steeper, given the greater prevalence of 2G and higher proportion of lower-end phones.
Retention rates drop with larger file size. Large apps are 33% less likely to be retained after 1 month although iOS users are 12% more likely to retain an app than Android users, according to Flurry.
App sizes vary by app type and platform. Below is some data I gathered from the iOS and Play stores. Basic utilities are 1-10MBs. Communications and social media apps are 20-30MB. Most casual games are 40-50MB. Most mid-core games are 300-800MB. From my not-so-scientific sample list below, Android apps are on average 37% smaller in package size than the iOS app from the same publisher.
So, what is an ideal app size, especially in markets like India with challenged infrastructure?
The ideal size is 10-15MB globally. Idea size for an app for tier 2/3 countries (like India) is below 5MB. 500MB+ is a non-starter. At 50MB+ the conversion rates fall off dramatically. On Android and iOS, conversion rates dip by 50% in tier 1 nations for non-game apps above 50MB. In tier 2 and tier 3 nations, conversion rates dip by 50% for games above 15MB.
To lower the cost of loyal customer acquisition (a function of conversion rate, download success rate and retention rate), unbundling in emerging markets makes positioning more clear and therefore discovery is easier, in my opinion. Unbundling also decreases the hits-driven nature of mobile apps businesses. And finally, cross-promotion within an owned & operated network of apps also dramatically reduces the cost of introducing a new app into the app network.
There is a cost to this strategy though. The engineering and products team now need to maintain multiple code-bases and roadmaps. Initially building out the network may required multiple marketing pushes and already strained marketing budgets may not be enough to get apps into into the high ranks on the app stores. Unpopular functionality, separated out into an app, will not get downloaded/used.
Some things to keep in mind if you are thinking about going down this route: You need to maintain a strong brand identity across all apps in the network to build company value and cross-promotion ability. Also need a common ID system to build and leverage customer data across multiple apps. And a good cross-promotion engine is needed. Tapjoy and Flurry are leaders in this category but there are lots of other options. To reduce app package size, you will need to rationalize your third-party SDKs, remove most heavy media files and reduce functionality dramatically.
10 comments
Comments feed for this article
January 23, 2016 at 7:32 pm
Anant Computing Platform » Hyperlocal: Building the ideal Hyperlocal solution (Part 3)
[…] storage space on their phones, by adding and deleting apps. I often quote from a blog post by Lightspeed Venture Partners who recommend that ideal apps size in India is 5 MB. But when we are talking about the dream […]
November 18, 2014 at 8:16 pm
Why the Internet’s Next Billion Users Will Be Mobile-Only - AcrossTheFader.BIZ
[…] very aware of how much data they are using and the size of the applications they are downloading.This is a fascinating quote in a post from LightSpeed Venture Partners, an investment firm focused on […]
July 14, 2014 at 11:56 am
badri sanjeevi (@badrisanjeevi)
Nice article. In Mobango.com – the independent appstore we run at People Group, we have spoken about this issue many times. The bigger challenge is also that Google does not make a number of APIs public. For instance, in the old Java world, there were detailed tracking for downloads was possible – delivery of Jad file, the delivery of jar file, the install status, the install success, failure reason codes etc. Google does not provide much intelligence on any of this chain of events between initiation of download and actual install happening.
Apps above 5 Mb are too tough to deliver good conversion on. This is largely depressing the overall app download advertising market in India as well.
July 14, 2014 at 12:09 pm
Dev Khare (@dkhare)
Thanks, Badri. Do any of the ad networks provide server-side as well as SDK-based functionality to measure the conversion funnel from click through activation?
July 9, 2014 at 6:00 pm
appcatalyser
Markets like india it is very tricky because people use budget phones with less specification but still want to enjoy everything. So having apps with huge memory needs won’t be a good choice.
June 16, 2014 at 1:56 pm
Roshan D'Silva (@roshandsilva)
Interesting that perception is that the only overhead is engineering. In my opinion the branding and marketing costs are more significant – imagine trying to create virality and word of mouth for multiple brands.
June 16, 2014 at 1:59 pm
Roshan D'Silva (@roshandsilva)
good example is Swarm by foursquare. How many people even know what it is?
June 22, 2014 at 12:34 pm
Dev Khare (@dkhare)
Fair point. Cross-promotion, common login/ID system and consistent branding between all apps helps.
June 12, 2014 at 4:59 pm
Gaurav Sharma (@IronLionGaurav)
Great article. One trend I have seen a lot recently esp. with heavy graphics games is that they are putting up installers on the appstore and then making the user download large files once the user opens the app. A great example is Gameloft’s Rival Knights. In the Google Playstore this game is highlighted for having killer graphics but the size is only around 25 mbs. Once I download the game a prompt shows up telling me I need to download another 80 mbs to be able to play the game. Its an interesting strategy but it could annoy a size conscious user.
June 13, 2014 at 9:32 am
Dev Khare (@dkhare)
Even the Twitter app (34MB) on my iPhone bulks up to 85MB with content. So for most content-heavy apps, this is probably the case; even more so for games.