Endeavour Partners - Blog

We are doing some research on the future of the web, and one of the areas of interest is augmented reality. I’ve seen and played with some demos, but actually found two apps of real value to me over the next few days, when I’m in London:

• Nearest Tube does just what it says – finds the nearest tube stations and identifies their locations

When you load the app, holding it flat, all 13 lines of the London underground are displayed in coloured arrows. By tilting the phone upwards, you will see the nearest stations: what direction they are in relation to your location, how many kilometres and miles away they are and what tube lines they are on. If you continue to tilt the phone upwards, you will see stations further away, as stacked icons.

[youtube=http://www.youtube.com/watch?v=U2uH-jrsSxs&color1=0xb1b1b1&color2=0xcfcfcf&hl=en_US&feature=player_embedded&fs=1]

• WorkSnug finds WiFi hotspots in London, and good places to work

The latest [augmented reality app] is one that [is very] helpful indeed, to anyone who’s found themselves desperate for wi-fi and a coffee while on the road [in London]. Work Snug, an app being developed in conjunction with Plantronics, leads you to cozy workspaces, each of which have been rated on a list of qualities, from noise levels to seating to wi-fi.

[youtube=http://www.youtube.com/watch?v=0z_Q3yl4NjM&rel=0&color1=0xb1b1b1&color2=0xcfcfcf&hl=en_US&feature=player_embedded&fs=1]

David Pogue has a review today of Motorola’s Droid which includes some discussion à propos how to categorize and name devices of this type.

Motorola Droid

He promotes the noun ‘app phone‘ for them, attributing it to @mentalworkout.

[Cool app, BTW, for those who have fear of flying. I took the Virgin Atlantic flight to London earlier this week; if you're lucky enough to fly in Upper Class, it's such an extraordinarily soothing experience that you probably don't need the app.]

I really like ‘app phone‘ , and suggest that we all adopt it for this class of devices:

• Apple’s iPhone
• all current Android ‘phones
• most modern BlackBerrys – post Curve
• Palm’s Pre
• Nokia’s N97 and N97 Mini running the latest version of Symbian

This post re-surfaced for me, however, one of the key topics that we have found ourselves debating frequently over the last many months; what is a ’smart phone’, and what should we call it?

This is a common challenge in high-tech; how do you think about new phenomena? How do you build robust mental models? We believe that having the specialist expertise to do this, and the relevant experience of having done this, is one of the key things that differentiates Endeavour Partners.

First, what are the key criteria:

• downloadable applications – in which case do BREW and Java devices qualify?
• user interface, such as (responsive) touch screen or QWERTY+touch pad/trackball interface to allow easy navigation for the web and similar applications
• running multiple applications – which disqualifies the iPhone?
• great at browsing – typically with a full WebKit browser
• third party applications have to be available, affordable and accessible
• what about size – is there some constraint here, because otherwise a laptop could qualify?

And what about the additional capabilities that are now part of the competitive benchmark:

• fast graphics – for video, browsing and gaming
• accelerometers
• GPS – for location services
• WiFi

There are several specific devices or types of devices that illustrate this challenge, and the grey areas involved:

• older BlackBerrys with thumbwheels but without trackballs – great at e-mail web but suck at browsing
• the Nokia E71, a great (particularly when it launched) device handicapped by its click-pad for navigation (which on one occasion proved enormously frustrating as the cursor moved in clicks that circumnavigated a small target without ever being able to actually click on it, on a site that should have been designed with mobile devices in mind – Handango)

Nokia E71

• many of Nokia’s myriad Symbian S60 devices that have 12-key keypads, lacking either a touch screen or a viable navigation method for browsing
• and what of the forthcoming Nokia N900 – is this a smartphone, or not?
• and given how unresponsive the touch screen on the N97 and N97 mini can be, and some of the usability challenges that remain with Symbian, do the N97 and N97 Mini qualify?
• almost all Windows Mobile devices, that lack a touch pad, requiring a stylus or arrow keys for what is enormously painful navigation (Sony Ericsson’s Experia X1 is one of the few devices that overcomes this challenge)

On purely pragmatic grounds, and notwithstanding flame wars from some purists and Verizon’s new advertising campaign, clearly any definition that excludes the iPhone on the technically focused grounds that it does not run multiple applications at once, except for some built-in apps such as Mail and Phone, does not make much sense. Although this may be an important consideration, it clearly does not deter users, and the ease of switching amongst applications mitigates this significantly.

The related question was what to call these things? We tried the term ‘smart device‘, to emphasize the that the capabilities went way beyond making a call. Unfortunately that promotes confusion as it embraces some very devices that do not have ‘phone capability at all.

So, let’s endorse the term ‘app phone‘ for these high end devices, and use the term ’smart phone’ for the broader group of which these are a subset.

[youtube=http://www.youtube.com/watch?v=4Fl718QO_xQ&color1=0xb1b1b1&color2=0xcfcfcf&hl=en&feature=player_embedded&fs=1]

James Geshwiler of Common Angels just opened for Mobile Monday here (yes, on a Wednesday) in Boston. Perhaps most interesting was how he opened, with an overview of the landscape for entrepreneurial investments in this space:

• the focus has shifted from carriers to apps and devices
• connectivity has become a commodity

This perspective is particularly interesting because it’s a good leading indicator of what sort of innovation by insurgents is going to be funded. To that extent, the perception of opportunity becomes reality.

Critically, it increasingly reflects an ‘over the top’ perspective, in which carriers face significant challenges in remaining relevant.

At 4G World here in Chicago, the debate is raging on about the potential value of femtocells.  One audience member asked yesterday, “If I already have a 20 Mbps broadband connection and a WiFi access point, then why do I need a femtocell?”  I think he was on to something, but it is interesting to go back in time a little and see how the relative business cases for WiFi and femtocells have changed as the market evolved.

This feels like a confession:  Three years ago, I was a proponent of femtocells as a solution for indoor coverage, particularly when compared to WiFi UMA.  Too few devices had WiFi, and it seemed unrealistic to expect many users to adopt a solution that required such a constrained selection.  Different family members or different personnel within an organization may have very different device preferences or be at different points within the replacement cycle.

By contrast, femtocells allow installation of customer premise equipment where it is needed, and the problem is solved.  Individual users don’t need to change their behavior, and if the costs were right, then it could be an attractive solution for many households and businesses.  But the costs were too high (I priced one at $300 at the time) and commercial availability was and still is limited.

Fast-forward three years and the market has changed dramatically:

1. Lead by the iPhone and RIM’s Blackberry, smartphones are rapidly gaining share – in developed markets, the majority of users will have one within two device replacement cycles
2. The challenge is no longer only about indoor coverage, but now also includes an immediate need for additional data capacity to support high bandwidth applications
3. Leading smartphones will nearly all be WiFi-enabled

The result is that homeowners and businesses no longer need to pay to install a femto cell, and users no longer need to be constrained to specialized, suboptimal devices.  For most people, WiFi coverage already exists in their home and office, and these same people are likely to carry a WiFi-enabled smartphone now or within two upgrade cycles.

These smartphones are also responsible for driving a massive increase in data traffic and clogging up 3G networks.  WiFi provides a mechanism for operators to offload 30% or more of this traffic without massive upgrades — and in advance of LTE networks coming over the next 3 years.

Device makers are scrambling to offer smartphones that match Apple’s iPhone as the key competitive benchmark (including WiFi).  Mobile operators are responding to increasing smartphone usage with investments in WiFi assets and improvement of seamless switching between WiFi and wide-area cellular to offload some of the associated explosion of data traffic.  The way the iPhone restricts high-bandwidth applications so they can only be used over WiFi is a good example of a first step in this direction.

Meanwhile, the level of investment in femtocells is small by comparison, involves a tough value proposition to customers, and is focused on solving a problem that is largely taking care of itself.

My money is on WiFi.

So, after all the discussion last week, how big is the App Store ecosystem?

Our analysis suggests that total App Store ecosystem revenues will be between $700 million and $1 billion for 2009.  This estimate breaks down as follows (** see note for more details):

2.7 billion downloads * ~15% of downloads are paid apps * $1.75-$2.50 per paid app = $700 million – $1 billion

But, as Digital Clockworks points out, in all the debate about numbers and methodology some of the most interesting and important elements of the story are being missed.  First, this is a big, rapidly growing market.  The key difference between our estimate and some of the lower numbers we’ve seen is that we’ve accounted for continued growth over the second half of 2009.  The second half of the year will generate 50% more revenues than the first half.  Also, we see this growth continuing.  Over the next 12-18 months the iTunes App Store will become a multibillion dollar annual business.

More importantly, however, the iPhone and the App Store have revealed a new paradigm for how mobile users will interact with their devices and use the Internet.  The App Store represents one more example of how “over the top” approaches over IP networks are beating out purpose-built, vertically-integrated network businesses.

Most of us wouldn’t dream of paying for a customized Internet experience on a tailor-made device from our broadband service provider (*** see note).  But that is the way we used to buy telephone service, and it continues to be the way we do things for mobile and video services.  Over time, all of these businesses will follow a similar pattern, breaking down into their component parts so that the best adapted players win in each piece of the business.  The only questions are:  “Who are the best adapted?” and “How long will it take?”

The App Store represents a major 'over the top' incursion in mobile data. How long will it take for 'over the top' models to eat into the mobile voice market and the video market?

__________________________________

Notes:

** Total downloads for 2009 come from Endeavour Partners’ analysis and forecast of App Store downloads, tracking Apple announcements; ratio of paid vs. free apps comes from Pinch Media but is consistent across multiple sources; average price per paid download has been reported as low as $1.67 and as high as $2.87, so we used a range

*** We actually used to buy Internet services this way as well:  AOL, Prodigy, NetZero, and PeoplePC (which included a PC in the bargain!); there is a segment of users who appear to be heading back in this direction (see the success of Comcast.net or the recent push for connected netbooks with service from AT&T or Verizon)

Mobile Entertainment has spotted a fascinating development: the Apple’s App Store now apparently has a ‘top grossing’ ranking.

Why is this so important? One of the things that we think is critical is a ruthless, relentless focus on the data; what do the facts tell us? One of the most common metrics used is market share, but here’s the very first question you should ask yourself:

• volume (units)?
• or value (revenue)?

In mobile phones, for example, these can mean very different things: Apple has ranked as high as #3 by value (and #1 or #2 by profit), despite having only a small share by volume. Whenever you see market share data, ask yourself if it’s market share of volume, or market share of value.

In mobile applications, where many are free, it’s not enough to just know ‘Top Free’ or ‘Top Paid’, you really need to know revenues to understand the economics well.

Moreover, if rankings on this chart at least, correspond to market share of value, rather than market share of volume for paid apps, it enables developers to re-price to maximize revenues. Compare, for example, Top Paid versus Top Grossing for today; Madden NFL is #2 on Top Paid at $7.99, but #1 on Top Grossing, as AppBox Pro only costs $0.99. Moreover, TomTom at ~$100 was always going to find it hard going to make it into the Top Paid (it’s not there now, at least down to #50), is #5 amongst Top Grossing.

Top Paid

Top Grossing

This may help stall the rush to the bottom which has been the cause of so much unhappiness amongst developers.

AdMob posted some interesting App Store usage data last week, suggesting that the App Store ecosystem is generating $200M in revenues per month prompting at least one commentator to proclaim the App Store ecosystem to be a $2.4B annual opportunity:

If I were to tell you that Apple’s app economy was worth more than $2.5$2.4 billion a year, you would laugh hysterically, shake your head and walk out of the room, yes? Surf on over to some other web site? But here I am telling you exactly that! According to mobile advertising startup AdMob, there are some $200 million worth of applications sold in Apple’s iPhone store every month, or about $2.4 billion a year.

These numbers are significantly higher than our recent estimates, so we decided to look a little deeper. We’re not alone in being a little skeptical, however:

Estimates that the iPhone App Store is worth $2.4 billion a year are utterly ridiculous, iPhone developers say.

“Do the math and that’s a ridiculous claim,” wrote developer Layton Duncan of Polar Bear Farm, an iPhone developer based in New Zealand.

Duncan did the math: $2.4 billion divided by the 65,000 apps in the App Store is $37,000 per app, per year. And while some developers earn that, many do not.

David Barnard of App Cubby, … says AdMob’s number is at least 5x too big. The iPhone App Store is worth $250 and $500 million per year, estimates Barnard….

Here’s what Barnard says are the average prices for Apps in the App Store:

• Top 10 = $1.99
• Top 50 = $2.23
• Top 100 = $3.18

According to Barnard, it takes about 400 sales per day to break into the top 100; and about 10,000 sales per day to hit  the very top of the charts. Assume the average sales in the top 100 to be about 1,000 per day. If the average price for an app in the top 100 is $3.18, that’s about $116 million per year for the top 100 apps.

“Most apps sell in the single digits per day, and quite a few don’t sell at all,” Barnard says. “There is a long tail, but it’s a very skinny one. I wouldn’t be surprised at all to learn that the top 100 grosses as much as all other apps combined.”

Comparison between our April 2009 estimates and Admob’s August 2009 numbers:

Admob data is significantly higher than our estimates

AdMob reports:

1. a higher number of installed devices
2. more than double the number of downloads per user
3. a higher percent of users buying paid apps
4. a much higher dollar value spent per revenue-generating user

Luckily, there is a rock solid number we can use to adjudicate these differences.  Apple has been periodically reporting total App Store downloads and we have been tracking and projecting this number:

April = 179 million, August = 260 million

Total downloads for April were approximately 179 million while total downloads for August were approximately 260 million.

Factoring this back into the AdMob analysis:

Based on AdMob's estimated number of August iPod and iPhone users, the number of downloads per user per month is 5.7, not 13.6

The AdMob data is significantly overstating downloads per user.

So, where is the disconnect coming from?  While the total downloads reported from Apple are “hard” numbers, the AdMob data is survey-driven and has two potential sources of inaccuracy.  The first source is sample bias:

All data in the feature section is based on survey results taken by users on their mobile device.

Respondents were sourced by responding to mobile ads throughout AdMob’s iPhone and Android networks.

There was no incentive offered to participate in the survey.

There were 1,117 total respondents:  390 Android, 380 iPhone and 347 iPod touch.  The survey was run from August 14th-August 21st

Based on the methodology described above, we believe sample bias could explain the entire discrepancy.  In essence, AdMob’s sample is skewed heavily toward active users (rather than “all users”).  This cleanly explains the higher number of downloads, the higher percentage buying paid apps, and the higher average revenue per user.

The other potential source of inaccuracy is that the survey captures self-reported estimates by users about their usage patterns.  Surveys of this nature are notorious for over or underestimating usage as people both lie to give a more favorable impression of themselves (putting forth who they want to be rather than who they are) and just have a hard time accurately remembering what they did over any given period of time.

Taking these together, it looks like AdMob’s results are distorted by a couple of half-truths:

• the people who responded to the survey are unrepresentative of the market as a whole
• they may also be exaggerating a little, or perhaps kidding themselves as well

One of the key trends that we’ve been tracking closely is the emergence of the smartphone as a handheld gaming platform:

• the threat it poses to specialized handheld gaming platforms, such as the PSP and Ninetendo DS, which it rivals or surpasses in performance
• how significant an opportunity this may become

Gameloft announced today that it has sold over 6 million paid games:

Gameloft(R), a leading publisher and developer of downloadable video games, today announced that it
has sold over 6 million [paid] games on the App Store.
From the distribution model to the types of games available, the App Store has completely revolutionized the way handheld games are played, perceived and received
The iPhone OS continues to be Gameloft’s number one platform

Clearly, as competition amongst smartphone platforms and smartphone vendors intensifies, one of the key factors that determines the outcome will be how good they are as gaming platforms, how attractive they are to games developers and the strength of their business ecosystem.