A cambrian explosion in AI – but how will new services surface?

Dag Kittlaus, founder of Siri, wrote on Techcrunch yesterday to predict that A Cambrian explosion in AI is coming. He notes that there has been “massive uptake of assistant services spurred by Apple’s Siri, Google’s Now and Microsoft’s Cortana” but says these services are still in their infancy. The missing piece is an ecosystem of services that work with these assistants. That would enable the Holy Grail of an assistant that intelligently finds and uses whatever apps or services we need for a given task – just like in the movie Her from earlier this year.

Need a babysitter for tomorrow night in a new location? The assistant should know the age of your kids, find a service that matches your price profile and then make a reservation, maybe asking you to confirm first.

I have just been looking for a babysitting service in a new location, and it wasn’t easy, so that that sounds very exciting, and I very much look forward to it becoming a reality.

However, it’s not clear to me how the discovery process will work. Healthy ecosystems have some way for quality to float to the top. Examples include ranking or voting systems as we see in Product Hunt and Stack Overflow, algorithms that incorporate user signals e.g. Page Rank, and manual system that takes user feedback as a core input e.g. the app stores.

The whole point of assistant services is that they choose services for us. In the babysitting example above I don’t want the assistant to come back with three options and make me choose, I want one option that I’m happy with. Before I trust the assistant I might want to hear about second and third options to make sure I’m getting the best, but I imagine I would stop bothering with that pretty quickly.

However, if assistant systems make the choice for us then user signals are limited to feedback given on the service. New services, by definition have little if any feedback, making it unlikely that assistants will recommend them and that innovation would suffer. That’s one nightmare scenario. The other is that the assistants only recommend services which have a relationship with whoever wrote the assistant – i.e. Siri only recommends services that have built a relationship with Apple.

In my view app store owners already have too much influence over which apps we use, and that runs counter to the original promise of an open internet. My fear is that assistants, wonderful though they will be, will worsen this problem.

App store discovery a little less broken?

It’s a common refrain that the process by which apps are found or discovered is broken. Discovery and hence download volumes are driven more than anything by ‘app store placement’ and by being ‘featured’, both of which seem to be more down to the whim of Apple and Google than the merit of the app. What we need is an equivalent of Google’s Page Rank, but for apps. That way good apps would float to the top and discovery would be more meritocratic. That would be better for startups who often have great products but lack the resources or the networks to curry favour with Google and Apple.

The current discovery process isn’t completely broken, in that Apple and Google do take the quality of the app and it’s popularity into consideration, but it isn’t right. Consider these stories. Two similar stage startups that we are close to have recently been playing the App Store game with Apple. They both networked hard to get close to the right people at Apple, developed features that Apple suggested they should and then held back release of those features in the hope of getting promoted. One got promoted in a big way (Stylect) and the other got only a low placement in an App Store category with little traffic. Neither knew until the day of the promotion. That can’t be the best way to do things.

However, Apple and Google are both heavily invested in the status quo. Their app stores earn them a lot of money and are a protective moat for their mobile phone businesses. So I’m not expecting things to change quickly. Thus I was surprised to read this morning that app store competition is increasing. Tomasz Tunguz has found that app store volatility has increased substantially over the last twelve months which indicates that new entrants are doing better and that discovery is getting less broken.

That’s a little bit of good news for startups in an area where they don’t usually get much. I like to think that one day we will have an open system on mobile, but until we do life will be harder for young companies than it needs to be and we will get less investment and innovation in mobile than we could.

An ‘iPhone moment’ for wearables may be close

The wearables industry at the moment can be likened to smartphones in the pre-iPhone era. As it was with Nokia Communicators and Blackberrys, quite a few people have them but the experience is generally not very satisfactory and the mainstream isn’t buying yet. In large measure I think that’s because most of them don’t do much more than count steps and step count data isn’t that interesting. Heart rate, galvanic skin response, and accurate motion data are on the roadmap for all sorts of wearables companies and they hold out the promise of realtime data that is useful on an ongoing basis – e.g. to monitor stress levels and alert you before you realise yourself. These could be game changers in the way the iPhone and Apple App Store changed the smartphone game.

The other thing that could change the game is a big push from Apple. Judging from their new ad for the iPhone 5s embedded below that support is upon us.

As an early stage investor I’m excited by the potential for new wearables startups. I think backing companies at the concept stage with enough money to get them through a Kickstarter campaign is an interesting play.

The Android vs iOS paradox facing startups

Benedict Evans wrote an interesting yesterday about Android fragmentation (tl:dr 75% of devices that hit the Play Store run Android 4.x meaning Google has reduced the impact of fragmentation, additionally they’ve sidestepped the issue for their own services by moving them out of the OS and into a software layer that can be updated over the air). However, the point I want to bring out is his list of the issues startups should consider as they choose whether to develop first for iOS or for Android:

  • Apple’s homogeneity means things behave in predictable ways reducing development costs
  • Android has a much larger addressable market – people who can afford $50 devices up to $600 rather than just $600
  • Anything on the bleeding edge won’t work predictably on many Android devices
  • There are more early adopters on Android than iOS

As Benedict notes, this leaves developers facing the paradox that the open platform is harder to hack and forces startup CEOs to make the trade off between keeping dev costs and time to market down on the one hand and reaching more early adopters and a larger market on the other.

Most startups we see opt to go iPhone first because that allows them to maximise the speed and efficiency of learning.

I love my Android phone and I’ve always thought that as Android gains market share over iOS more startups would start to develop first for Android and I would stop having to wait months for new apps to come my way. I’m now thinking my wait will continue.

Apple’s new product process is a long checklist

Yesterday I wrote about how building a startup is increasingly an exercise in disciplined application of process. Creativity and flair will always be important, but whereas that used to be most everything and charismatic sales driven entrepreneurs with huge personalities were common, these days implementation of processes like ‘lean’ and ‘customer development’ are increasingly important and the personality profile of entrepreneurs is changing. We used to get lots of Larry Ellison’s and these days we get more Mark Zuckerberg’s.

The reason for this is that many processes formerly regarded as the preserve of creatives have been broken down into process steps that non-creatives can follow. It turns out that Apple’s new product process is one of those. This is a quote from Leander Kahney’s recently published book on Apple design chief Jony Ive:

“In the world according to Steve Jobs, the ANPP would rapidly evolve into a well-defined process for bringing new products to market by laying out in extreme detail every stage of product development.

Embodied in a program that runs on the company’s internal network, the ANPP resembled a giant checklist. It detailed exactly what everyone was to do at every stage for every product, with instructions for every department ranging from hardware to software, and on to operations, finance, marketing, even the support teams that troubleshoot and repair the product after it goes to market.”

I remain in awe of Jobs’ ability to come up with products that hundreds of millions of people coveted – that was his genius, his magic spark – the point here is not to take anything away from that, but rather to point out that as far as possible everything downstream from the idea is engineered. The beauty of this is that it improves reliability and predictability of execution.

I’m thinking now that the venture capital could be similarly broken down into checklists and good process.

Nest’s Tony Fadell on the importance of focus and saying ‘no’

Tony Fadell is the man of the moment. Nest, the connected thermostat and smoke alarm company he founded in 2010 has just been acquired by Google for $3.2bn. I hear that in it’s three years of existence Nest got it’s thermostat into 1% of US homes and hit revenues of $100m (not sure if that’s trailing or projected). Before Nest Tony played a leading role at Apple brining the iPod to market.

At a Google Ventures event last year he has this to say about focus and saying ‘no':

I learned the power of ‘no.’ No is really important. Entrepreneurs are told to say ‘yes, yes, more, more.’ To help you focus, to help you really understand what you’re doing, you have to say no a lot. When you say yes to everything, you get distracted. When you say no, you have to get the one thing you’re doing really right.

Note the way he makes a positive out of saying ‘no’ – it forces you ‘to get the one thing you’re doing really right’. That’s important because focus often feels like a negative – choosing not to do something, and thereby shutting off an opportunity. At startups there is a constant tension between creating opportunities to get lucky, often phrased as ‘having irons in the fire’, and going all in on doing one thing really well. Nest’s amazing success shows the power of going all in, but that’s something you can only do once you have conviction that you’ve found the thing that’s really going to work for you.

Android is on a path to total dominance

screen-shot-2014-01-08-at-3-19-49-pm

As you can see from the chart above (produced by Microsoft) Android devices have been outselling iOS devices for a couple of years now. However,  many of the Android devices were barely smartphones and device sales were only half the story. The app economy was the other half of the story, and iOS was stronger here. For most startups the app economy is the more important side of the story because it’s a better proxy than device sales for revenue and customer acquisition potential.

Note that I say ‘was stronger’. As you can see from the tables below the two app ecosystems are now comparable. Judging by review numbers Android users are doing more with free apps and iOS users are doing more with paid apps, perhaps reflecting the different demographics between the platforms and the higher proportion of low end devices on Android, but on balance there isn’t much difference between the two. (Slight caveat: it would be interesting to see this analysis for total app revenue, including in-app purchases.)

 

14.01.06-Top_200Given Android’s massive volume advantage, I expect that now the app ecosystems are at parity Android will swiftly move into a clear number one position on all metrics, particularly as low end devices become more app-capable. In my experience tech early adopters still prefer Apple, by and large, so we may not feel the shift anecdotally until it is quite advanced.

 

The inexorable rise of free – paid apps example

chart-of-the-day-app-store-revenue

 

Business Insider posted these two charts today under the headline Paid Apps are Dead. Looking at the data, it’s hard to refute that claim. App developers are still making money, of course, it’s just that they are doing it through in-app purchases rather than from an up-front charge. This solution is better for consumers because they get to try before they buy, and better for app developers who can increase their overall revenue by enabling their superfans to spend vastly more money than anyone would ever pay in an up-front charge. Supercell is the most recent poster child for this model. Only 10% of their users pay anything at all to the company, yet last year they made $40m in profit on $105m in revenues and earlier this year Softbank valued the company at $3bn (see here for more details).

As Nicholas Lovell explains in The Curve other companies should be copying Supercell’s model. The idea of The Curve is that the revenue potential of every user can be plotted on a curve – most will be at $0, or very close to it, but there will be a few where the number is much higher – often thousands of dollars. To make money in a digital age companies must organise their business models to exploit the curve and get the most value from their products and customers.

As Nicholas also explains, The Curve isn’t limited to apps, it applies to all industries where customers can be engaged for zero marginal cost. Zero marginal cost allows companies to build relationships by giving away small amounts of value in the hope of recouping much greater value down the line.

Business models built around the needs of companies don’t work

It seems trite and obvious to write that the best products are designed around what customers need and want, yet it is still surprisingly common for companies to make decisions based on what’s important for themselves.

Digital magazines are a great example. Physical magazines have been suffering from declining circulations for a while now and their owners have mostly struggled to charge for their content online. When the iPad arrived they and Apple saw an opportunity to reverse the trend by selling magazine subscriptions through the Newsstand app. Unfortunately they were solving for their problems rather than what users wanted. Magazine publishers saw a chance to start charging for digital content and Apple saw an opportunity to take a cut. Neither stopped long enough to realise that the great free content was still out there and would inevitably find a way onto people’s tablets.

I’m writing this post today because I’ve just read an article about the impending death of tablet magazines. It describes how many people are getting free realtime content in magazine format from apps like Flipboard whilst very few people are subscribing to digital versions of traditional magazines. Moreover, Apple has now made it so that the Newsstand app can be buried inside a folder whereas previously it had the prominence of a guaranteed home page spot. A lot of time and money was wasted by Apple and publishers in bringing the Newsstand model to market because they were thinking about their own needs rather than what consumers wanted.

The area where I most often see this mistake of putting the business model ahead of what consumers want is subscription ecommerce. Subscriptions are great because they bring predictable revenues and higher customer life time values (LTVs). High theoretical LTVs in turn make it possible to justify high customer acquisition costs (CPAs). For managers and shareholders in ecommerce companies the combination of high and predictable revenues, and justifications for  high CPAs is intoxicating and often results in a lack of discipline in questioning whether customers really want a subscription product.

You should be able to justify your business model by reference by what’s good for your customers,