Wednesday, 2 December 2009

Analyzing Y Combinator


[Author's note: If you're just interested in the end result, feel free to skip to the conclusion at the bottom.]

This winter marks Y Combinator's tenth investment season, yet even with their reputation as the world's leading early stage venture fund their success rate (much like that of their competitors) has remained for the most part private.

The lack of transparency among early stage venture funds is bad for competition and for the global start-up scene as a whole. One of the most frequent criticisms placed against such funds is that they loudly broadcast their successes and brush their failures under the carpet. Even if done unintentionally this behaviour has a strong "selection bias" tendency which many have argued can make funds seem more successful than they actually are.

To combat this I'm undertaking a detailed analysis of these funds with the aim to publicly produce data showing both the success rates and the failure rates of firms invested in by these funds.

My original plan was to complete my analysis before posting about it, however, due to the number of requests I've had for the data I've decided to publish it as I go along.

The first step of my analysis is found below: An Analysis of Y Combinator.


Methodology


I've attempted to track down every start-up Y Combinator has funded from a wide range of sources including CrunchBase, news articles, and discussions on Hacker News.

While YC does not maintain a public list of start-ups they've fund they do however maintain a count on their about page and have done since the Summer 07 season. Prior to this it seems even their internal tracking was somewhat lax.

Using archive.org I was able to retrieve historical counts and from these deduce the number of start-ups financed each year. I was able to use this information to identify the seasons where I had missed start-ups and attempt to track them down. Using this information I believe I have tracked down all but 3 of the Y Combinator start-ups (1 from S07 and 2 from W08) from those seasons prior to S09. I've made the assumption that the 3 I've been unable to track down failed to launch.

I've avoided naming those companies from S09 who are still in stealth mode.

There has been some inconsistency in the usage of the common YC terminology of Sxx and Wxx when referring to summer and winter seasons of funding. Some sources use W08 for the Winter 07/08 season while others use it for the Winter 08/09 season. I believe the official stance is the earlier example and that's how I'll be using it here.

This convention has only been widely adopted since 2007, which has meant prior to that point it was often unclear which season a start-up belonged to. I've tried to use funding dates, demo days, and blog/news articles about YC funding to associate start-ups with the correct season. While I believe most have been associated correctly some of them may be off by a season.

Following S09 YC's current about page has a count of 144 start-ups that they've invested in. My final list has 145 (taking into account the three I know I'm missing). I suspect the reason for this inconsistency is Kiko/Justin.tv (both of which were founded by the same team but both companies took separate YC rounds of funding) which I've counted as two companies but YC might have counted as one.

While I've listed further funding rounds for many of the companies, this is based upon public data, it's likely that many of the other active companies have received further funding privately in order to be able to continue operation.

The criteria used to decide if a start-up has failed is the following: if it's been publicly declared as closed, if the founders now list a different employer (on linkedin, HN, etc) , if there's been no activity in the six months and if their sites show signs of abandonment. In the case where it seemed ambiguous I've marked the start-ups status as unknown.

Any suggestions to improve the methodology or corrections for the data are of course very welcome.


Overview of YC Companies

  • 82 - Active (24 having received further public investment rounds)
  • 33 - Failed (25 closed after launching, 7 failed to launch)
  • 14 - Acquired (2 of which are suspected to be low value <$250k share deals)
  • 9 - Stealth
  • 5 - Unknown (in a stage where it's unclear if still active or failed)
  • 2 - Other (1 merger, 1 where YC sold it's stake to a private investor)

Summary by season

  • YC S05 - 1 Active /3 Failed / 2 Acquired / 2 Other
  • YC W06 - 5 Active /1 Failed / 2 Acquired
  • YC S06 - 3 Active /6 Failed / 1 Unknown
  • YC W07 - 5 Active /4 Failed / 4 Acquired / 2 Unknown
  • YC S07 - 12 Active /4 Failed / 3 Acquired
  • YC W08 - 9 Active /10 Failed / 1 Acquired / 1 Unknown
  • YC S08 - 17 Active /3 Failed / 1 Acquired / 1 Unknown
  • YC W09 - 14 Active /1 Failed / 1 Stealth
  • YC S09 - 16 Active /1 Acquired /9 Stealth

Full Breakdown


YC S05

  • Reddit - Acquired by Conde Nast.
  • Kiko - Acquired by Tucows.
  • Loopt - Active. Raised 15million in Series A and B financing.
  • Simmery - Closed.
  • Infogami - Merged with Reddit.
  • MemAmp - Never launched.
  • FireCrawl - Never launched.
  • ClickFacts - Ycombinator stake acquired by private investors.

YC W06

  • TextPayMe - Acquired by Amazon.
  • Inkling - Active.
  • Flagr - Active.
  • Infinity Box / Wufoo - Active. Raised 100k in angel funding.
  • i'minlikewithyou - Active. Raised 6.5 million in Series A and B financing.
  • Sprout / Clustrix - Active. Raised VC financing from Sequoia, USVP, ATA Ventures.
  • AudioBeta - Closed.
  • Web Shaka / YouOS / Project Wedding - Acquired by eHarmony.

YC S06

  • Jamglue - Active.
  • Scribd - Active. Raised 12.7 million in Series A and B financing.
  • Xobni - Active. Raised 16.4 million in Series A and B financing.
  • JumpChat - Closed.
  • UZ Labs / Pollground - Closed.
  • Shoutfit - Closed.
  • Talkito - Closed.
  • Thinkature - Closed.
  • Mod Four - Closed.
  • LikeBetter - Unknown.

YC W07

  • Parakey - Acquired by Facebook (July 2007).
  • Heysan - Acquired by Good Technology (May 2009).
  • Zenter - Acquired by Google (June 2007).
  • Boso / Auctomatic - Acquired by Live Current Media for 5 million.
  • Octopart - Active.
  • Virtualmin - Active.
  • Snipshot / Pixoh / Tripfly - Active.
  • Buxfer - Active. Raised 300k in angel funding.
  • Weebly - Active. Raised 650k in angel funding.
  • View3 - Closed.
  • Whitenoise Networks - Closed.
  • writewith - Closed.
  • Socialmoth - Unknown.
  • tsumobi - Unknown.

YC S07

  • SocialPicks - Acquired by FinancialContent in (May 2009).
  • Anywhere.FM - Acquired by imeem in January 2008.
  • Clickpass - Acquired by SynthaSite.
  • Adpinion - Active.
  • Bountii - Active.
  • Disqus - Active.
  • Faux Labs / Splashup - Active.
  • AppJet / EtherPad - Active.
  • Zecter / Versionate / ZumoDrive - Active. Raised $1M from Tandem Entrepreneurs(November 2007).
  • Songkick - Active. Raised 1 million Angel round, 3.5 million Series A financing.
  • Dropbox - Active. Raised 7.2 million in Series A and B financing.
  • Justin.tv - Active. Raised VC financing from Alsop-Louie Partners.
  • Biographicon - Closed.
  • Blue Frog Gaming / DraftMix - Active.
  • Habit Industries / Fuzzwich - Active.
  • Reble - Closed.
  • SlapVid - Closed.
  • iJigg - Active.

YC W08

  • Omnisio - Acquired by Google for $15 million (July 2008)
  • 280 North - Active
  • BaseShield / Ninite - Active
  • Insoshi - Active
  • MightyQuiz - Active
  • Wundrbar - Active
  • Heroku - Active. Raised 3 million in Series A
  • WebMynd - Active. Raised 400k Angel round
  • RescueTime - Active. Raised 900k in Series A
  • 8aweek / SocialBrowse - Active
  • Chatterous - Closed
  • Kirkland North - Closed
  • Mixwit - Closed
  • Snaptalent - Closed
  • Tipjoy - Closed
  • YumDots - Closed
  • Deluux - Never launched
  • Joberator - Never launched
  • Addmired - Unknown

YC S08

  • slinkset - Acquired by Posterous (June 2009)
  • Cloudant - Active
  • Anyvite - Active
  • CO2Stats - Active
  • Frogmetrics - Active
  • Jobsyndicate / Startuply - Active
  • Meetcast - Active
  • Poll Everywhere - Active
  • PopCuts - Active
  • Snipd - Active
  • Picwing - Active
  • ContestMachine - Active
  • TeamApart / Meetcast - Active
  • TicketStumbler - Active
  • BackType - Active. Raised 300k seed funding
  • Fliggo - Active. Raised 500k Angel round (as Vidly)
  • Posterous - Active. Raised 725k Angel round
  • Scoopler - Active. Raised 750k in seed funding
  • ididwork - Closed
  • Urbantakeover / lolcathost3000 - Closed
  • People & Pages - Never launched
  • Youlicit - Unknown

YC W09

  • CloudKick - Active
  • Echodio - Active
  • Foodoro - Active
  • Voxli - Active
  • AirBnB - Active
  • ReMail - Active
  • DivvyShot - Active
  • Picurio - Active
  • thesixtyone - Active
  • Wattvision - Active
  • FathomDB - Active
  • HeyZap - Active. Raised 650k in seed funding
  • SkySheet - Active. Unlaunched.
  • Nambii / pocketfungames - Active
  • Propable - Closed

YC S09

  • GraffitiGeo - Acquired by Loopt (October 2009)
  • Directed Edge - Active
  • FanChatter - Active
  • Highlightcam - Active
  • RethinkDB - Active
  • JobPic - Active
  • MixPanel - Active
  • JobSpice - Active
  • Olark - Active
  • RentHop - Active
  • WakeMate - Active
  • InstantQ - Active
  • Plurchase - Active
  • DailyBooth - Active. Raised 1 million in Angel funding
  • Listia - Active. Raised 400k in seed funding
  • FlightCaster - Active. Raised 1.3 million in Series A
  • Bump Technologies - Active. Raised 3 million in Series A

Conclusion

As Y Combinator is the first fund I've analyzed it's obviously not possible to compare it to other funds at this stage. However we can compare it to the performance of angel investors.

According to the NESTA report the average return for an angel investor after 3.5 years is 2.2x in the UK and 2.6x in the US. On this basis for Y Combinator to make an above average return they need to see an average exit of 650,000 per start-up (based on 15k invested for 6%).

There have been two seasons which are more than 3.5 years old S05 and W05. For these to have made above average returns exits would need to have a value of over 13 million dollars. It's probable that the acquisitions of Reddit, Kiko, and TextPayMe from these seasons comfortably surpassed this amount.

Looking over all 145 start-ups, total exits would have to be worth over 95 million to represent a return of 2.6x. Mid-point estimates of exits would place the value of existing YC exits in the region of 40-50 million (a follow up post will break this down in detail).

Given that three YC companies (Scribd, Loopt, Xobni) have raised 45 million in VC financing between them, it seems highly likely the current market value of the 80 active companies in YC portfolio is worth significantly in excess of the remaining 45 million required for YC to be performing better than the average angel investor.




Tuesday, 10 November 2009

Selling Tickets with Startup Megatrends (aka Yield Management 101)

The story

Imagine yourself as an airline owner who owns a plane with 100 seats, on a typical weekend flight you manage to find 60 customers willing to pay $500/seat. This is great, but what about those other 40 seats, as soon as the plane takes off those seats are worthless. What if instead of charging customers $500 for those remaining 40 seats we only charged them $300 ? - we could fly with a full plane !

So for a couple of months you try this and it works great, but then suddenly the number of customers buying the $500 seats drops dramatically. So you set off to speak to your customers and find out why, and they all come back with the same response: "we'd rather wait a while and get the $300 seats". By selling some of the seats at discount you've ended up devaluing all your seats. What's a poor airline owner to do ?

Welcome to Yield Management

Yield Management is the task of figuring out how to dynamically price a perishable resource (anything from fresh fruit to hotel rooms) without your discounted sales destroying your premium sales.

To quote Wikipedia:

There are three essential conditions for yield management to be applicable:

  • That there is a fixed amount of resources available for sale.
  • That the resources sold are perishable. This means that there is a time limit to selling the resources, after which they cease to be of value.
  • That different customers are willing to pay a different price for using the same amount of resources.
Who uses it

Let's look at some industries that have sophisticated yield management systems:

  • Airlines
  • Car Rental
  • Hotels
  • Online Advertising
In every one of these cases sophisticated yield management techniques have resulted in billions of dollars of additional revenue. Figuring out how to sell off excess capacity at discount without damaging your main sales is a huge business opportunity, and one which is barely exploited in most industries.

Before we look at industries where it could be used, let's have a look at how the four industries above do it so we get an idea of how we could adapt the techniques.

Airlines and Car Rentals have both taken similar approaches in becoming real-time pricing industries, with prices constantly changing, making it almost impossible for consumers to know if a price will increase or decrease. Price confusion works. Faced with uncertainty people are willing to pay the price offered rather than wait for a better price, people don't want to take a chance on the price dropping if it could also rise. There are however very few other industries where this technique would be accepted so let's have a look at the other models.

Online advertising typically takes a dual sales route, direct, where most major sites will have a sales team which tries to sell their premium ad-space, and brokered, where the left over "remaindered" ad-space is sold via ad brokers who resell it pseudo-anonymously. One example might be a site using Google Adsense for their remaindered ad-space, the buyers have no idea which site their ads will be featured on so there's no risk in the premium ad-buyers going that route. One of the key advantages of this approach is that the site keep selling the premium ad-space up-to the very last second, until a page is actually served when a brokered ad inserted instead of an empty premium ad slot. This means that unlike the other industries listed here, sites don't have to be able to predict in advance how much spare capacity they'll have.

Hotels have one the most interesting solutions, a few years ago Priceline came up with the idea of selling hotel rooms anonymously in a similar way to what's done with online advertising. Called "Name Your Own Price" it let customers specify a criteria for a hotel (4* in Manhattan) and a price, if it could find a hotel which matched the room would be booked (you commit to the purchase before finding out which hotel it is). By taking this approach hotels could discount their empty rooms without devaluing the rest of their rooms.

Who should use it

Let's take a look at some industries that don't regularly practice yield management but could benefit from using it to monetize underused capacity/resources:

  • Theatres
  • Cinemas
  • Gigs / Music concerts
  • Sports events
  • Museums
  • Theme parks
  • Trains
  • Restaurants
  • Fast food/lunch shops
  • Sporting facilities (gyms, tennis courts, golf courses, etc)
  • Equipment rental (music, sports, tools, etc.)
  • Venue hire
  • Hairdressers
  • Band hire
  • Services (Personal tutors, photographers, etc.)
  • Universities

A successful yield management system for any of these would easily be worth millions. However one of the reasons that they haven't adopted these techniques is they can't afford to invest in building models to predict future demand like airlines and hotels can. So a last-minute approach in a similar style to online advertising is critical.

So what would we need if we wanted to achieve the same results without being able to predict demand in advance ? - well there are two key building blocks that would allow us to build our new funky low-cost yield management system:

  • Ability to target customer individually based on market segmentation. Customers get annoyed if different people see different prices. However customers (for whatever irrational reason) are unfazed at other people getting discount vouchers. The ability to push out a "voucher" to an individual is extremely valuable.
  • Ability to wait until the last possible moment to offer discounts. Holding off as long as possible encourages people who already want to make a purchase to make it otherwise if they wait they risk losing out and not getting what they want. Plus the supplier gets to know exactly what their excess capacity is when discounting.
Historically achieving these two has been tricky, individual targeting is expensive and when holding off you end up with a lead time of a few days in practice (to allow time for people to see the offer, book and get to the location) which isn't late enough to avoid interfering with premium customer purchases.

In an ideal world imagine a scenario where a theatre realizes it's only 75% full an hour before a show starts. What if they could push out an offer to all students within a one mile radius offering tickets at 70% off and make immediate sales to anyone interested ?

Until recently such a concept wouldn't be practical, but with the convergence of smart-phones, social networks and pushed real-time data suddenly it's become almost trivial. Innovative just-in-time yield management pricing has become feasible for even small businesses. Something along similar lines would work for almost every one of the industries listed.

If you think the combination of these three (mobile, social and real-time) sounds familiar, it's what Fred Wilson describes as The Golden Triangle. And right bang in the middle of the triangle sits the gold mine of yield management. A huge opportunity for start-ups to ride the three megatrends of today and still be profitable from day one.