Saturday, 11 December 2010

Are Posterous Fudging Visitor Statistics ?


I've recently started using Posterous for the blog of my new developer recruitment startup and while Posterous is a great product I'm somewhat concerned that the readership counts generated by Posterous are significantly higher than what Google Analytics show.

Earlier this week I wrote an article on What Software Developers Watch on TV, here are the stats from it:

Posterous Stats:



Google Analytics:



As you can see Posterous reports three times as many visitors as Google Analytics (3000 vs 1000). Googling reveals many people seeing the same issue.

I also uncovered a response by Posterous co-founder Sachin Agarwal on the topic:

First of all, when people read your posts through the Posterous reader (like I do) that counts as a view for that post on Posterous. But that would not hit google analytics. Same for RSS feeds.

If people hit your site and have javascript disabled, that would still count as a view on Posterous, but would be ignored by Google analytics

If I go to the main page for your site that will count as a post view for *each* of the posts on that page. Google would count that as one view for that page, and no views for each post.

So the fundamental difference here is google analytics is counting when that particular page is loaded with javascript, while we count anytime that *post* is loaded, on any page anywhere.

However I don't believe that this response explains the discrepancy. My Posterous is relatively new so almost no-one is reading it via an RSS reader or via the homepage (well 27 users are according to GA). Javascript is only disabled by between 1-3% of internet users. While these factors might explain a 5% discrepancy in the stats, they come no-where near to explaining a 300% difference.

I've run a fair number of blogs and websites over the years and there's only two things I can think of that would cause such an inflation of numbers:
  1. If Posterous are counting each image download as a page view (which while it would explain a doubling of the numbers, I don't believe this is what is happening)
  2. If Posterous are counting page views generated by bots and crawlers.
I suspect it's almost certainly the second option. I've certainly had websites where 2/3rds of the traffic was caused by automated crawlers.

But if this is the case then the visitor numbers that Posterous are displaying to their users are completely wrong.

Unfortunately it's also in Posterous advantage to show inflated numbers. The biggest threat to a blogging platform isn't it's users defecting to another platform, but rather it's users getting bored and giving up blogging. And lack of audience is the major factor in causing a blogger to get bored and give up. By posting inflated numbers (whether intentionally or otherwise) Posterous is likely getting users to keep blogging when they might otherwise have given up.

I may of course be wrong; but unless Posterous can show that those thousands of extra viewers are genuine users, I'm going to feel slightly morally uncomfortable relying upon their (otherwise awesome) product.

Update: From the discussion taking place on Hacker News where one of the Posterous founders responds evasively to the issue it seems that automated bots/crawlers are actually responsible for the majority of these "phantom visits".

Sunday, 28 November 2010

Dear British Government: You've Screwed Up Visas Badly


Last week the British government announced they're effectively wiping out the Tier-1 visa. For those not familiar with it the Tier-1 visa is the points based visa that most non-EU software developers and startup founders use to operate in the UK (it's also used by academics, scientists and engineers).

The announcement this week reduces the number of Tier-1 visas from 14,000 down to a mere 1,000. To put that in perspective there are 330,000 professional software developers in the UK. Many of the most talented developers I've had the pleasure to work with in the UK have been here on Tier-1 visas.

Whereas before any talented software developer with a track record could come to the UK on a Tier-1 visa to start a company, now to enter the country they'll have to get a specialist "founder visa" which is much harder to qualify for.

The UK government is shooting itself it in the foot, they're introducing a hard to qualify for "founders visa" while removing the visa option that most foreign startup founders actually use in practice.

To make matters worse the government is actually using the cut in Tier-1 visas to increase the number of Tier-2 visas. The difference between the two categories is that a Tier-1 allows the holder to move freely between jobs (or start their own company), while Tier-2 ties the holder down to a particular company.

The Tier-1 visa didn't surpress salaries, if an employer didn't pay the holder a competitive salary then the visa holder could just go and work for another employer in the UK.

However the Tier-2 visa does surpress salaries. With a Tier-2 visa the holder is tied to their employer (as long as they want to remain in the country) and can't change jobs, so the company knows that it doesn't have to pay a competitive salary. That's the reason why employers wanted Tier-2 visas rather than Tier-1 visas.

In one week the UK's gone from having one of the most talent and entrepreneur friendly visa systems in the world to one of the worst.

I'm making an open plea to the government: When people complain about immigration, they're not complaining about software developers, engineers, scientists and academics. Please don't ruin our international competitiveness for the sake of appearing to do something about immigration.

Wednesday, 24 November 2010

Ideas For Startups: Advertising Ideas


In the spirit of my previous post on the value of ideas I've decided to write up a series of posts (each post following a theme) covering some of the ideas I've had for startups but haven't got around to work on (yet).

In this post I'll be covering ideas for advertising related startups. So without further ado, here they are:

1) A Real Demographics Ad Network

Plenty of websites (small social networks, forums, dating sites) have lots of information on their individual users in term of age, gender, interests, etc. But the only way they can monetize this information is by building their own ad network (as Facebook, Linkedin and PlentyOfFish did) that allows custom targeting. A third-party network that sites like these could easily integrate into would mean more money for the sites, more relevant ads for the end users and an alternative to Facebook Ads.

2) Facebook Ad Tools

Facebook's own ad tools suck if you have a large number of ads, there are a large number of ad optimizations which you have to do manually (adjusting prices to be competitive, keyword splitting, A/B testing images, demographic tightening) and for sophisticated analytics (tracking Unique CTR over time, etc.) you need to download the CSV data on a regular basis and manually process it in Excel.

The only people who have tools to automate these activities are large ad agencies. A tool for everyone else would be nice. Facebook's api is in closed beta at the moment, so you'd probably need an inside contact to get access.

3) Ad Network with Feedback

At the moment if you're running a website with ads (or affiliate links) you only know if someone clicked on an ad, not how valuable that click was. So you don't know which of your users are "high-value" (so can't attract more of them or give them extra benefits). A CPA Ad/Affiliate network that fed back the the value of the ad-click (in a way that could be matched back to the user) to the site running the ad would solve this problem

4) Realtime Adsense for Dynamic Websites

Adsense only works on sites that the Googlebot can scrape, if a site uses user specific content or generates content on the fly Adsense becomes mostly useless. An alternative that could process content in realtime using embedded javascript to provide context-specific ads would likely be much more profitable than non-targeted banner adverts (which is what sites tend to use currently).

5) Correlated Interest Finder

On sites like Facebook where you can target users by interest getting a large enough audience can be tricky as people often don't list all of their interests. However if you can figure out complementary interests you can use those to widen your target audience.

Figuring out complementary interests can be non-trivial though, but it should be possible to automate the process. Maybe by data mining a bunch of social network and dating websites to find correlated interests. Facebook has one of these but it's terrible. It tells you everything is correlated with "music" because it doesn't discount the fact that everyone has music as an interest.

Sunday, 14 November 2010

The Value of Ideas


Every business starts with an idea. One of the most frequently asked questions by first-time startup founders is "How much is my idea worth ?"

Unfortunately most are surprised to hear the reply "nothing".

Paul Graham wrote a good article about the philosophical argument for this viewpoint but I wanted to present an alternative argument focusing on the underlying economics of the value of ideas.

It essentially comes down to the question: Does the idea give you a defendable competitive advantage ?

There are some industries such as cleantech and pharmaceuticals where ideas (new technologies, drugs, etc.) are patentable, and as such give you a competitive advantage which your competitors can't copy. In these cases the idea does have value, and you can often raise investment, etc. solely on the basis of your idea.

However in other industries such as in the web and mobile space, ideas are not patentable, and thus theres nothing to stop a competitor from copying a good idea. Hence ideas in these areas don't give you a defendable competitive advantage. It doesn't matter how good or unique the idea is. If you can't stop someone cloning it the day after you launch then the idea itself has no intrinsic monetary value.

A good idea is the foundation of a successful business, but like the foundation of a new building, it's purely the basis for the work ahead.

Wednesday, 20 October 2010

Low Valuations on Dragon's Den


Every once in a while the TV show Dragon's Den comes up in discussion among startup founders, most recently on a thread on Hacker News, and typically someone brings up how low the valuations are and how much equity the Dragon's take arguing that the Dragons are exploitative and taking advantage of entrepreneurs.

(For those not familiar with the show it involves entrepreneurs, typically with an early stage business either with a new physical product or service, pitching a panel of five angels for investment)

The average investment is typically in the region of 100k for between 30-40% of the company.

While this may seem high our view of valuations get skewed because our background in startups for which exponential growth is feasible.

A Worked Example: Sheldor Industries

Let's look at an example with a fictional company:

Sheldor Industries have come up with a funky new gadget and have some initial sales. They have a pre-money valuation of 250k and are seeking an investment of 100k for tooling.

Say an angel took a 10% stake in Sheldor Industries, then the angel would need Sheldor to grow to a 10 million pound company in order to achieve 10x returns. That is the company would need to grow 40-fold.

Now if the angel took a 40% stake in Sheldor Industries, Sheldor would only need to grow to a 2.5 million pound company to achieve the same returns. Note that still requires 10-fold growth

Now while for tech startups which by-and-large are designed to scale and suffer less physical restrictions than traditional companies 40-fold growth can be perfectly feasible, for traditional product and service companies that sort of growth is completely unrealistic. Even 10-fold growth may be optimistic.

So to realistically achieve the 10x return on investment angels are looking for they have to take significant equity stakes in companies.

Hang on 10x...

While going for 10x return on investment may seem on the greedy side, it's actually the angels only option.

The average angel gets a return on their portfolio of between 2.4-2.6x after 3.5 years, but they need to invest with the objective of getting 10x return in order to achieve these numbers. Most angel investments don't pay-off, therefore the returns on the ones that make money have to cover not only their own cost but also that of the failed investments.

Typically 9% of angel investments produce 80% of the returns (this 9% generally being companies that do provide a 10x+ return on investment). Therefore it only makes sense for an angel to invest in a company if they can aim for a 10x return. If a company has no chance of being in that top 9% then the statistics say it's simply not worth investing in.

(Image Credit: Babbletrish)

Monday, 26 July 2010

Is Groupon Sitting on a Legal Timebomb ?

(disclaimer: I'm not a lawyer and this is my personal opinion. I may well be wrong)

Recently Andrew Mason gave an interview on Mixergy where he described how they went about obtaining the domain name groupon.com from it's previous owner who planned to launch a similar service. Essentially it boiled down to Groupon getting a trademark and using it to stop the owner of the domain from launching his service under that name (at which point he sold the domain to Groupon).

This raised a discussion on Hacker News about the legality of registering a trademark to pressure a domain owner to sell up. While there doesn't seem to be a clear answer on the topic, perhaps more worryingly is Groupon's trademark filing which includes the following declaration:

The undersigned, being hereby warned that willful false statements and the like so made are punishable by fine or imprisonment, or both, under 18 U.S.C. Section 1001, and that such willful false statements, and the like, may jeopardize the validity of the application or any resulting registration, declares that he/she is properly authorized to execute this application on behalf of the applicant; he/she believes the applicant to be the owner of the trademark/service mark sought to be registered, or, if the application is being filed under 15 U.S.C. Section 1051(b), he/she believes applicant to be entitled to use such mark in commerce; to the best of his/her knowledge and belief no other person, firm, corporation, or association has the right to use the mark in commerce, either in the identical form thereof or in such near resemblance thereto as to be likely, when used on or in connection with the goods/services of such other person, to cause confusion, or to cause mistake, or to deceive; and that all statements made of his/her own knowledge are true; and that all statements made on information and belief are believed to be true.

Note this trademark was filed in March 2009, a few months before they bought the domain from the original owner but after they were aware (going by the dates in the interview which may be inaccurate) of the original groupon.com. So it's surprising they could make such a declaration at that point.

While this might not be an immediate problem, it likely adds to the potential for legal issues in the future, especially given groupon.com's trademark lawsuit against groupocity.com.

Thursday, 10 June 2010

The London Startup Event Guide

I've recently been asked by a few people about the startup scene in London, and figured it'd be a good idea to write up a summary, especially for those people interested in starting a startup.

I think I've got most of the major regular events listed below, but please post in the comments if I've missed any. You can also follow me on twitter where I'll try to post info on more irregular startup events in London as and when they come up.

Weekly events:

OpenCoffee - OpenCoffee tends towards the more practical business side of the startup scene. They meet weekly, but the real value is on the OpenCoffee mailing list which is a positive treasure trove of practical information. Most one-off startup events tend to get announced on the list as well.

Monthly(ish) events:

London MiniBar - probably the biggest of the events, normally the format is a few talks typically some from sponsors and some from local startups followed by networking. The talks tend to be very hit-and-miss and the acoustics of the venue aren't great. However the networking is often interesting, as Minibar draws people from a wide variety of backgrounds. @MiniBarLondon

uk.news.yc - Inactive since the start of last year the Hacker News meetup used to be the best networking event in town, full of highly energized developers getting products out there. The practice of getting everyone to introduce themselves to the group was a huge plus for finding people with similar interests without having to work the room. Sadly the guys at SongKick (YC S07) who normally organized it seem to have got too busy actually building a startup to keep running it. Listing it here though in the hope it might encourage someone else to pick up the reigns and bring it back to life. Update: It's back, organized at http://www.meetup.com/HNLondon/calendar/14910886/

DrinkTank - often rated as London's best startup meetup, it has a moderated members list to try and ensure quality. They didn't accept me so they must be good.

BootLaw - Run by the well respected startup lawyers Barry Vitou and Danvers Baillieu, BootLaw events focus on the law as it applies to startups. Between them Barry and Danvers somehow seem to be at every startup event in town so if you're active in the startup scene you'll undoubtly bump into them sooner or later. @BootLaw

AppFusion - Focusing on smartphone apps AppFusion normally mixes presentations by smartphone developers showing off their new apps with networking.

The following are events I've not attended but I've seen mentioned by people active in the scene:

ebizlaw - The other regular startup law event in London, this one run by the lawyers at Fox Williams.

OpenSoho - A tech/media startup meetup.

Facebook Garage - A Facebook app developer meetup.

London Tech Startups - A general purpose tech startup meetup.

MobileMonday - Another mobile meetup.

Less regular:

BarCamp London - The London branch of this global unconference. Didn't run this year, but was replaced by HackCamp. @barcamplondon @hackcamp

London Startup Weekend - One of two build-a-startup-in-a-weekend events in London (the other being Launch48 below). @startuplondon

Launch48 - Launch48 is a three day event with the first day being a conference on startup topics and the next two days being startup-building. @launch48

Both Launch48 and London Startup Weekend are great events, and I'd recommend anyone thinking about starting a startup to attend them. They both follow a similar pattern of people pitching ideas for what to build, followed by a vote to get the ideas down to a manageable number. Attendees then break up and join the teams for the ideas they like most.

LSW tends to have smaller teams more focused on product building, and L48 tends to have larger teams but tends to also cover the non-product side of startups (marketing, financing, etc.) An outcome of this is that L48 generally produces more serious startups (VouChaCha, protected.cc, Wraply are all startups that came out of previous L48's and are still active).

Overall both are great events for networking and learning.

Social Innovation Camp - Another build an app in a weekend event, except SI Camp focuses on apps that contribute to the social good of society. @sicamp

Techcrunch Europe Meetups - Techcrunch Europe runs sporadic networking events in London, the mix of attendees is usually different from most startup events and often attracts more later stage startups and investors.


Monday, 7 June 2010

Yes. Europe Does Produce Big Tech Companies.


Last night on the way back from this weekends launch48 I spent a bit of time thinking about Mike Butcher's tweet from earlier that evening:

"Name 10 *big* tech companies to come out of Europe. Let's see: Skype, Nokia, SAP... er, help me out here..."

After spinning through a few in my head I came to the same conclusion as he did following a few responses. That most of them are b2b.

There are the big enterprisey giants like SAP and Sage, and then there are the industry specialized ones. There's no shortage of big chip manufacturers (STMicroelectronics, ARM, CSR, Wolfson Microelectronics) and in the financial sector you can't throw a stone without hitting a British tech company (Reuters, ICAP and Markit all have billion dollar+ revenues from their technology arms). The open source sector also seems to have flourished in Europe (Canonical, MySQL AB, Trolltech). Yet apart from Canoncial, none are consumer facing, and none have large consumer brands.

However after much needed sleep I realized that there are indeed large European consumer tech companies, but we miss them because they're either so ingrained into our lives we don't think of them as tech companies or because they're so large we don't even associate them with a particular country.

Both IMDB (now owned by Amazon) and Gumtree (now owned by eBay) were originally British start-ups that built global consumer brands which fall into the later category.

There are in-fact entire sectors which are European dominated. In the gaming (in both the video and gambling sense of the word) and telecoms sectors there's no shortage of hugely successful companies.

European gaming companies like Rockstar Games, Media Molecule, Lionhead Studios, Rare, Codemasters, Crytek, Criterion, Rebellion, Traveller's Tales and Eurocom (and many more) have produced a significant percentage of the best selling computer games of the last decade. Big online players such as Playfish and King.com are also European.

And when it comes to online gambling, pretty much every major gambling site in the world from Betfair to PKR.com is British, and those few that aren't are mostly European.

The European telecoms sector has produced Vodafone, O2, Orange, T-Mobile, Skype, Nokia, and Ericsson all globe trotting European giants.

Europe may still be well behind the Valley in producing large tech companies, but maybe not as far behind as some people think.

Wednesday, 24 March 2010

Journalism: The Advertising Business Model


A few weeks ago Y Combinator funded NewsLabs announced a call for journalists for their new news platform which seeks to provide a platform that journalists can use to publish their articles with NewsLabs handling some of the work traditionally done by newspapers (ad sales, hosting, etc.)

The basic business model they propose is selling ads on the content and doing a 80:20 revenue split with the journalists doing the writing while avoiding the high overhead costs that traditional newspapers have.

The obvious question is: Will the web advertising model produce enough money to make this financially viable.

Let's construct a business financials model and see if the figures add up. There are two key figures,
  1. How much it'll cost to pay a journalist for an article to make it a viable alternative to writing for traditional newspapers
  2. How much money an article can bring in.
If the second is higher than the first we have a workable business model.

For the first we can just use industry standard figures. Typically a journalist will get paid between $0.10/word (low-end) to $0.50/word (high-end). Most professional freelancers will get somewhere between those two figures, so lets take $0.30 (a typical rate for an article that would require some research) as our average cost and 1000 words as our average article size. From this we get the cost per article of around $300.

How much money an article can bring in is trickier due to huge variations in ad rates. So to get around this let's establish a range. To get sample price data I used the Federated Media rack rates, FM provides advertising for a wide range of popular blogs and news sites so is probably reasonably representative of the type of content we're interested in. Let's assume we'll run three ads per page, a Wide Skyscraper, a Leaderboard and Medium rectangle - a combination fairly popular on news sites.

From FM's catalogue I picked their most expensive ad prices for our best case, prices from the second highest quartile for the average case and prices from the fourth quartile for our worst case.

However before we can uses these prices we need to apply a couple of discount factors. Firstly because FM will take a cut of around 40% (which is typical cut for ad sales) and secondly because often a lot of ad inventory will remain unsold by direct sales. Typically for unsold inventory (called "remaindered" in the industry) those ad spaces are replaced with a generic source of ads such as Google Adsense. So I've added some estimates for how much our remaindered ad space will bring in, $0.50 - $3 CPM is typical for general content so those are the figures we'll use.

So now lets plug the numbers into a spreadsheet and see what we get (click to enlarge):

Best Case

Average Case

Worst Case

Before we look at these figures in details lets do a little sanity check to make sure our numbers aren't completely crazy. One way to do this is to run an alternative analysis and see if the numbers match up.

Luckily Dharmash Mistry (former executive at EMAP; now partner at VC firm Balderton Capital specializing in media) has done this for us. He did an analysis of online newspapers and came out with an estimate that a newspaper website with 20 million unique monthly viewers would likely make less than £20 million/year. To get a comparable CPM we need to convert monthly views into page views. As audited circulation figures are made public we don't have to guess this value but rather just look it up where we can see that when the Guardian had 20 million page unique monthly viewers they had 186 million page views.

If we crunch these numbers we get a CPM of $13.40 - so slightly worse than our average case of $17.40 but in the same sort of ball park, indicating our model is probably not unreasonable if a little optimistic.

Now we can combine our figures for advertising and the cost to pay a journalist for an article and come up with a figure of how many page views we need. At $17.40 we'd need 22,000 views per article (at $13.40 we'd need 28,000), while these figures aren't completely infeasible for a good writer, it seems unlikely that a significant number of journalists would be able to maintain that sort of readership for an individual column over a prolonged period especially for only "average" returns payment wise.

Overall it seems even with a "bare-bones" model, there just isn't enough money in online advertising to finance journalism.