Monthly Archives

April 2013

Optimal pricing for market places and platforms

By | Uncategorized | No Comments

A couple of weeks ago veteran Benchmark VC Bill Gurley posted a typically excellent and thorough analysis of pricing for market places and platforms. Bill calls the money that platforms take from transactions the ‘rake’ and you can see from the table above that there is a wide range out there. Rakes comprise a mix of straightforward take from transactions and other fees imposed upon merchants and/or consumers. In the case of eBay the rake comprises listing fees, fees to make listings more premium (e.g. more photos) and Paypal fees.

Bill’s central argument is that setting the rake too high is often a mistake. Short term revenues and profits will be maximised, but if the rake is too high marketplace participants will be constantly on the look out for other places to transact making the business will be vulnerable to competitors, and many potential customers may simply choose to avoid the platform altogether. He quotes Jeff Bezos’ famous saying ‘Your margin is my opportunity.’. To hammer the point home he quotes management guru Peter Drucker:

Number one on the list of Peter Drucker’s Five Deadly Business Sins is “Worship of high profit margins and premium pricing.” As Drucker notes: “The worship of premium pricing always creates a market for the competitor. And high profit margins do not equal maximum profits. Total profit is profit margin multiplied by turnover. Maximum profit is thus obtained by the profit margin that yields the largest total profit flow…”

Bill cites the example of oDesk stealing the market from Freelancer and others because their rake was 10% rather than 30% to argue that the optimum headline rake is around 10%. He also cites Apple’s 30% take and how that forced Amazon and Facebook to adopt non-IOS strategies as further evidence that 30% is usually too rich.

Finally, low headline rakes can be increased with mechanisms like additional fees for premium listings. The real trick, he says, is to have a model which makes people think ‘the marketplace is fair, but competitors activity on here makes me spend more than I would ideally like’. That way competitors get the blame and nobody leaves the platform. Google Adwords is a great example of a platform that is thought of in this way.

In Britain, online companies are growing 50x faster than the average company

By | London, Venture Capital | 2 Comments

CityAM reported today on research from Barclays:

Research from Barclays shows that firms that sell their products or services mostly online have seen revenues grow by on average 11.4 per cent in the last three years, over 50 times faster than GDP growth of 0.2 per cent over the period.

The research shows that the expansion is being driven mainly by small and medium-sized firms. Most high-growth businesses have fewer than 100 employees, while average revenues are £8.9m. Almost a quarter of firms – 23 per cent – are based in London, despite the capital accounting for just one-eighth of the UK’s population.

The dominant macro-economic story at the moment is one of ‘stagnation’ and ‘low growth’ which accurately reflects the headline statistics and situation at many of the world’s largest companies, but it fails to convey that there is a lot of volatility in the markets that add up to the flat headline figure. This data from Barclays illustrates the point well – average growth is flat, but UK online companies are growing at an average of 11.4%.

When we look around us this isn’t surprising. Some markets are collapsing – e.g. newspapers, much of high street retail, broadcast TV – and therefore for the headline figure to be flat some others must be growing fast – e.g. online retail.

This, of course, means that even in these difficult times there is opportunity for investors who are able to get money to work in those growth markets.

The other interesting thing about the Barclays research is that it shows much of the growth is located in small companies based in London – the sweet spot for venture capitalists.

A note on contrarian investing

By | Startup general interest | One Comment

Fred Wilson has a post up this morning title Return and Ridicule which talks about the dangers of herd investing. Fred quotes Bill Gurley saying that ‘you can only make money by being right about something that most people think is wrong’ which I think is spot on. If lots of people share your idea about what is right then they will all want to invest, the valuation will rise and it will be much harder to make money. The same goes with starting companies – if lots of people have the same idea competition will be fierce and success will be harder to come by.

In a nutshell this is the case for contrarian investing, and it is a driving force in my investment decisions. I look for companies that can be the winners in new markets and investing in those companies before everyone else wants to usually means believing that their new market is emerging when other investors still aren’t sure. Our 2011 investment in Conversocial is a good example. As I blogged when we made the investment our thesis was that they would become the leader in social customer service, but at the time most people thought that social customer service would only be a subset of the established social media software market and there wasn’t space for a new entrant. In the two years since then founder Josh and the team have done a great job establishing social customer service as a category and it is now getting recognised by the major analyst groups, but we all had to endure a lot of doubters along the way.

The problem with contrarian investing and believing in things that most people think are wrong is that those people generally think you are stupid. Fred Wilson wrote:

I have found that return and ridicule are highly correlated over the years. We have made more money on things that were highly ridiculed than any other cohort. When I see people laughing at ideas and companies we have backed, I smile. It means we are going to make a lot of money on that investment.

I hear this, but it is only true if the belief you have that everyone else thinks is wrong is actually correct. As I’ve said, I think finding projects you believe in which others don’t is a great way to make money, but it is critical that you remain open to your investment thesis being wrong, or in need of tweaking. When people ridicule my investments I want to know why it is they don’t believe what I believe, and to make sure that they haven’t got a point. When they laugh I look forward to the day when they’re not laughing any more.



Netflix’s big numbers

By | TV | No Comments

Netflix announced Q1 results yesterday, and they topped 1bn in quarterly revenues for the first time and reached 33m subscribers. No mean feat. But the numbers that really caught my eye are that Netflix will spend $350m this year delivering its service, and a further $2bn on content rights. Those are some punchy expenses and respresent a significant barrier to entry for anybody else looking to get into the TV streaming game.

I think that leaves the opportunities in TV squarely in the discovery space. Alongside their quarterly results Netflix published a paper on the future of television and one of their predictions is that we will soon see the end of linear TV. I buy into that for lots of reasons, not least the impact that Netflix had with House of Cards because the whole series was available to watch from day 1, but it begs the question of what experience will replace the ‘hit the couch, put your favourite channel on and sit back’ mode of engagement that is so prevalent with TV today. The obvious answer is a clever combination of playlists, social inspiration and algorithmic recommendations.

Netflix, Amazon, Google and Apple will try to own this layer as well, but there maybe space for a startup here, particularly given that consumers may want their discovery service to cover more than just one silo. In the long run it makes sense to me that these discovery services will link directly to the content owners, maybe handling billing and rights management. In that view of the world Netflix’s emerging content business will be much more important to them than their distribution business, which would be cut out of the equation. 

Extolling the virtues of the slow road

By | Startup general interest | 2 Comments

This morning I read a good post from YC Alumn Vibhu Norby who describes the dangers of a big product launch. He had everything set up for success – 25k people signed up to be notified about the launch, forthcoming Techcrunch and AllThingsD articles, a great social media plan, etc etc but his initial bang was quieter than anticipated, they didn’t get top of the app store charts and then new user sign ups trended down over the next few days not up. The lesson that Vibhu took away is that it’s best not to have a launch at all. At Origami, his new startup, they simply put the service live and are focused on making their small number of users happy, and letting the user base grow by word of mouth.

Big launches are fantastic if they go well, and if you are hell bent on taking the quickest path to wealth and fame then the lure of the big launch will probably be irresistable, but most big launches fail, and that failure comes with a cost. Big launches come with high expectations and disappointment is hard to recover from – employee morale will suffer, investors may lose faith, you may pick up a hard to shake bad reputation with the press, and possibly worst of all, you may lose confidence yourself.

Why do most big launches fail? Because very few companies are clever enough, or lucky enough, to have a good product market fit on launch day. For most it takes a period of iteration as customers use the product before they find the sweet spot.

You could say that choosing to have a big launch for a startup is adopting a strategy of ‘be lucky’, and as we know, luck isn’t a strategy.

Next a caveat. If you are operating in a known market then the dynamics are different. In this case you have a much better chance of good product market fit out of the gate and a big launch might make sense. The new Samsung S4 is a good example here.  Also, companies operating in very crowded markets may not be able to think of another way to rise above the noise (although that raises the question of whether they should find another market….).

Wrapping up – the ‘big launch’ vs ‘build slowly from a passionate user base’ is one of many areas where startups can choose between being aggressive about getting as big as they can as quickly as they can or focusing on sustainability first and fast growth second. I’m all for shooting for the stars, but for me the second path is the better one. It locks down value more quickly and relies less on luck. It’s absolutely right to keep open the option of being lucky, and if a lucky break comes to double down, and double down hard, but going to early and relying on being lucky just isnt sensible.

Clayton Christensen has similar advice when he talks about going for profitability first, and scale second, whilst the likes of Steve Blank and Eric Ries have a similar message when they say that finding product market fit is a startup’s first job. These are wise words from wise men, but it can be hard to know exactly when product market fit has been reached (how sure should you be?) and profitability turns out to quite complex too, as many startups reach the point when their inheret profitability is apparent before their management accounts are showing month on month profits. For me the time to invest hard to scale fast is when scaling doesn’t require any great leaps of faith, i.e. demand and marketing channels are proven at a small scale and there are no obvious barriers to growth.

Even the oldies are on Facebook now…

By | Social networks | No Comments

ofcom social networking dataThis chart from recent Ofcom research is up on Techcrunch today. It shows pretty clearly that an increasing number of people are social networking across all ages, with an especially large increase amongst those age 55+. I imagine this activity is largely on Facebook, with maybe a little on Twitter.

One intesesting thing to note is the length of time it takes older folks to pick up on tech trends that are popular with kids. Facebook was founded in 2004 and at that point social networking was already pretty widespread amongst students, so in this case it took eight years for the new technology to reach 25%+ penetration across all age groups.

Josh March on the power of social customer service

By | Conversocial | No Comments

Josh March, Founder and CEO of our portfolio company Conversocial has a great post up on pandodaily explaining why social is making customer service more important. In a single sentence it is because customer service becomes marketing too. On pandodaily Josh tells this story from Conversocial customer GoDaddy which illustrates the point:

A [GoDaddy] customer named Wes Tweeted that he was having problems with his service. He didn’t address the company directly, and he didn’t skimp on the salty language. Regardless, Waisman’s team, monitoring Twitter, soon learned about Wes’s problem, and tweeted back to fix the issue. Later that afternoon, Wes Tweeted again; “You want to know what great customer service is,” he wrote, “then talk to the people at @GoDaddy they tweeted me after seeing that I had a problem.”

Because social is public and because the convention is to make a comment or status update when notable service is received (good or bad) then good customer service delivers not only on customer service objectives, but on marketing objectives as well. As one of the commentors on the pandodaily suggests, the next step might be to divert some of the increasingly ineffective dollars spent on things like TV ad campaigns and use them to improve social customer service.

Hopefully this has piqued your interest to go and read Josh’s post in it’s entirety.


Behavioural economic tips for making consumers remember an app favourably

By | Startup general interest | No Comments

Pain chart

The power of behavioural economics is that identifies irrational cognitive processes to produce counter-intuitive insights which help us to better predict human behaviour. Consider the chart embedded above which shows the pain intensity over time experienced by experimental subjects. Patients in the B category clearly experienced more pain, but the surprising, counter-intuitive, result is that Patients in the A category remembered themselves as having had a worse experience.

In this case it is human memory which performs irrationally. Rather than accurately remembering and each element of an experience we are driven by what behavioural economists call the Peak-End rule which says that our memories are unduly influenced by the best and worst elements of an experience and by its final moments. In the ‘pain experiment’ which was conducted by Daniel Kahneman patients in Category A finished with much greater pain than those in category B and it is this worse ‘end moment’ which explains why they recall the pain as being worse than patients in category B (more details of the experiment here).

When I read about theories like this I like to test them on my own memories, and when I think back over the best and worst experiences of my life it is the peaks and troughs that I remember. For example – when I think about the best gigs that I’ve been to, I come up with a list where the best moments were AMAZING, not where the end to end experience was great. I can also see the importance of final moments, bands for example, routinely save their best tracks for the encore.

Turning to consumer internet, the peak-end rule tells us that to get consumers to have a great memory of an app it should deliver a real wow and maximise the peak, and that the impression in the moment when the consumer leaves the service is also important. Think about the Twitter ‘Fail whale’ and the amount of goodwill they got by having a bit humour having just delivered a service outage.

Interestingly, the amount of time spent using a service doesn’t impact how we recall it which implies that after a point engagement metrics are a poor guide to customer satisfaction (although they may be important for revenue). Google intuitively grasped this point in their early days when their goal was to get you off of Google and onto the page you wanted as quickly as possible.

Operating successfully in a world with lots of theories and little data

By | Startup general interest | No Comments

You may have seen the recent brouhaha in the financial press about a 2010 research paper by Carmen M. Reinhart and Kenneth Rogoff of Harvard which found a correlation between high levels of government debt and low or negative economic growth. This paper was important because it underpinned the austerity policies adopted by many governments around the world since the financial crash – not least here in the UK. The brouhaha has come about because economists from the University of Massachusetts have repeated their analysis using the same data but different weightings and found that the median growth at highly indebted economies was much higher.

In the meantime other studies of different data have apparently found similar results to the original 2010 paper from Reinhart and Rogoff.

This is a big deal because whether to pursue austerity policies is perhaps the most important decision facing most developed economy politicians right now.

The problem is that there isn’t much data to go on and the data quality itself is questionable. There haven’t been many periods in history when governments were as indebted as they are now, and it is impossible to know if economics have changed to a sufficient extent that historical data isn’t relevant. If we understood economics well enough to have a causal theory linking debt and growth then minimal data proving that theory would be helpful, but in the absence of that causal theory we are simply looking at correlations of limited data sets. In this situation people with convictions based on their political persuasion use data to press their case rather than to determine the right answer. A very dangerous game.

This problem also exists in business where most of our ‘understanding’ is based on observed correlations with little understanding of causal relationships. With the best intentions in the world this can lead to slavish following of trends to the detriment of business performance. The problem is, if anything, more pernicious in business because the advice of most business gurus includes is backed up by some level of causal logic as well as observed correlations with success. The 1970-2000 mantra that a simple focus on shareholder value was the best way for companies to generate success is a good example. It started with an observation that there was a correlation – successful companies like Coca Cola were shareholder value focused – and then added the causal theory to back it up (in simplified form a company’s shareholders want value, which can only come from generating good cash and profits, which are the essence of a sustainable business), but the correlation has recently fallen apart and the causal link between a focus shareholder value and long term success is now hotly disputed. The problem arose because the causal theory was based on an erroneous assumption that shareholders would distinguish between short term profit creation that destroyed long term value and truly building for the long term.

So what to do?

I think there is huge value in drawing insight from the limited data available but it’s critical to sense check the conclusions that others come to before following their advice. The most important sense check is whether it feels right – intuitively it makes sense to me that indebted countries will grow more slowly because they are having to use more of their resources to repay debt and hence have less to promote growth. In a business context this intuition often has its roots in the vision and values of a company. Advice that is inconsistent with that world view won’t feel right.

The next sense check is the credibility of the theorist – most importantly, are they impartial, looking at the data and searching for conclusions, or did they start with the conclusion and then go looking for data to back it up.

Thirdly – do research. If an idea is important then it is worth investing time to form a considered view.

Finally – stay open minded and be prepared to try things and change. Don’t make the oh-so-common mistake of ignoring everything that doesn’t fit with existing beliefs (behavioural psychologists call this the ‘confirmation bias’) and be prepared to disucss half formed opinions and lightly held convictions. In a fast moving world with little data it’s the only way to progress.

These ideas apply to many decisions that startups have to take on an ongoing basis – should I have a freemium business model? should I sell director or via channel? should I pursue growth over profitability? etc. etc. etc.

Data to inform hyper local advertising

By | Venture Capital | No Comments

Mobile ads GPS study: How far will you drive for a deal? (infographic)

I love it when I come across data like this. The chart shows how far people will drive to get to a business in a bunch of different places in the US. It’s no great shock to find that in major cities people are less willing to travel – they have more options close by – but I bet there aren’t many mobile ad campaigns that factor this insight into account. The data, collated by navigation services company Telenav and reported on Venturebeat, also found that people will drive further to visit restaurants and malls than sandwich and coffee shops.

The really fun thing about data like this is that you can extrapolate to form a bunch of other hypotheses. For example, I would guess that the closer you get to the centre of a city the shorter the distances that people will travel, and that in general distance travelled will correlate with time and money spent.

I guess I enjoy this sort of analysis because it informs my view of how the world works which helps with my personal life, but is a real help with investing where I need to form views about what products will work in what markets on the back of very little data.