All posts filed under “Analysis

comment 1

$20 Billion – and a relaunch of Seed-DB

Just 11 years after Y Combinator funded the first handful of companies in the first seed accelerator, over $20 billion has been raised by accelerator graduates.  For those keeping track at home, this is just 16 months after accelerator graduates passed the $10 billion raised milestone.

Over $5 billion in exits have already been achieved by accelerator graduates.  Companies that have yet to exit are collectively valued at over $80 billion.

Early stage startups continue to be a power-law phenomenon.  Despite funding over 6,000 companies in nearly 200 programs around the world, 75% of the investment dollars have gone into accelerator graduates of just four programs: Y Combinator, Techstars, 500 startups, and Angelpad.

That said, this stat isn’t as dramatic as it might appear.  Those four programs have collectively funded over 40% of the 6,000+ graduates.  Essentially, they’re prominent because they figured out ways to scale effectively, either through bigger class sizes or more frequent programs.  Between these programs’ alumni and mentor networks, and reputational effects, they’re able to consistently find, fund, and mentor a higher-achieving tier of startups.

And while you obviously don’t have to go through an accelerator to succeed; it helps.  Pitchbook found that one-third of startups that raised a Series A round in 2015 went through an accelerator.  But only about 1,200 companies per year went through an accelerator in 2013, 2014, and 2015, and there were far more than 3,600 companies started each of those years.  So while companies that go through an accelerator are a small portion of early-stage startups (likely 10% or less), they are a much larger percentage of successful startups (33%).

Why Seed-DB?

One of the big reasons I created Seed-DB is because the world of accelerators is plagued by anecdata.  It’s easy to remember the accelerators that helped the B2C companies that you may use today; it’s harder to know about accelerators behind the B2B hard-tech companies that don’t get a lot of press but are growing like crazy.  I also believe there are some great accelerators (or at least accelerators that have found great companies) that don’t get the attention they deserve.

For example, did you know the Flashpoint program at Georgia Tech funded two companies that have both gone on to raise over $100million each?  Did you know the third biggest exit of an accelerator company (for $350million) came from AngelPad?

Seed-DB exists to give entrepreneurs the data on which companies have been through which programs, in order to make more informed choices.  To answer the questions:  is an accelerator right for me? Which accelerator is right for me? And why?

Seed-DB – relaunch

Today also marks a re-launch of Seed-DB!  While the user interface hasn’t changed substantially (I’m not a strong front-end developer), the data structures behind the scenes have changed substantially.

Charts & Tables

Tabular data is valuable, but charts bring data to life.  Seed-DB now has a dedicated “Charts & Tables” page to showcase this information.  There are four key charts:

  • Total number of accelerator companies over time, updated monthly
  • Total funding (in $) of accelerator companies over time, updated daily – now > $20 billion
  • Number of funding rounds over time, updated daily
  • Number of accelerator cohorts/batches over time, updated monthly
  • Log/Log chart of company total funding (for companies that have raised >$500k)

You can see from these charts that there was a significant change in trajectory with more companies going through accelerators starting in 2011, which increased further in 2012.  The chart of total funding has a significant trajectory change in 2014, which increased again in 2015.

The same page also has some of the most popular tables:

Focus on Cohorts

The biggest data structure change has been a pivot on accelerator cohorts or batches.  Previously each accelerator was a flat list of companies they had funded, though Seed-DB did store the month they started with at the program.  Now each accelerator shows the highlights of each individual cohort, and then you can drill down further to see individual companies in that cohort.  (You can toggle back to the old view if you want, though.)

This is significantly faster for most users, but also shows a new layer of detail.  It’s clear to see that one of the most successful YC companies to date (AirBnB) was in one of the smallest YC classes ever, in the middle of the financial crisis in Jan – Mar 2009.

Additionally, this cleans up the user experience for accelerators that run multiple programs in different cities or verticals.  (Specifically, programs like Techstars, DreamIT Ventures, Startupbootcamp, Wayra, etc)  Instead of each of these programs getting listed as separate accelerators, the various cohorts are all grouped together in one overall accelerator.

Sign Up for Updates

Interested in how accelerators progress over time?  It’s been just 16 months for accelerator companies to raise $10 billion; would you like to know how quickly the next $5 or $10 billion is raised?  You can now click Login, OAuth with Google or Facebook, and click one button to sign up for updates on when high-level milestones are reached.  (Your email address won’t be shared, and updates will be infrequent.)

You can also sign up for the Seed-DB newsletter, which will have more analysis and long-form updates, and is sent even more infrequently.

Better on Mobile

While I won’t say Seed-DB is truly mobile optimized, the tables of data in Seed-DB can be used far more easily on mobile than they ever have before, and the new charts work great on mobile, too.  (Thanks, d3.js!)

Patreon Campaign

Finally, I’ve kicked off a Patreon campaign to help support Seed-DB.  If you find Seed-DB valuable for yourself or the startup community, please consider supporting the campaign!  No funds will go to Jed; they will all be used to either pay monthly infrastructure costs, or go to contractors to help with data collection.  In other words, any contributions only go to keeping Seed-DB running and improving data quality.

Personal Disclaimer: I did my first research into seed accelerators in the summer of 2009, and created Seed-DB in the summer of 2012. Two and a half years ago I started working for Techstars as a Product Manager. This post represents my personal views, and not those of Techstars. All data comes from Seed-DB alone.

comment 1

$10 Billion

Ten years after the seed accelerator model was pioneered, Seed-DB has now identified over $10 billion that has been invested in accelerator graduates.  Over 200 seed accelerator programs around the world have funded nearly 5000 companies, and over 300 companies have already exited for a total of over $3.5billion.  The total valuation of companies that have come through accelerators reaches in the tens of billions of dollars.  If you don’t believe that accelerators are a relevant way for early-stage technology companies to get funding and started… you’re sadly mistaken.

How has this happened?

To quote Ernest Hemingway (via Chris Dixon’s great post): “Two ways. Gradually, then suddenly.”

The first accelerator, Y Combinator, only started ten years ago.  Techstars’ first class was in 2007, DreamIT Ventures in 2008, AngelPad in 2010, and 500startups in 2011.  Most of these only funded handfuls of companies to start.  Those early starts have compounded to create a juggernaut of high-quality startups getting funding.  And as mature growing companies are able to find plentiful later-stage capital in the current environment, large funding rounds are becoming commonplace.  Here’s what’s happened over time:


Top accelerators are now brands themselves, and their stamp of acceptance and access to their networks is self-reinforcing.  While $10billion in total funding is an impressive milestone, companies that have gone through accelerators comprise only a small portion of the total venture funded startup scene.  There is a lot of space for their influence to grow.

Top tier programs

76% of all venture capital funding of seed accelerators go to graduates of just five accelerator programs: Y Combinator, Techstars, 500startups, Angelpad, and DreamIT Ventures.  But there are three parts to this: quality, quantity, and longevity.

An accelerator needs to be of a sufficient quality in order to help their companies become investable.  That accelerator needs to fund a relatively high number of startups in order to have a meaningful impact in aggregate.  (Either by funding more per year, or steadily accumulating a portfolio over time.)  The longer a program has been in operation the bigger their companies can grow.

This is not to say that new programs won’t break into this top tier… just that they need more time.

Funding – a Power Law in action

If you search for [venture capital] and [power law], you’ll see that venture capital is an industry known to follow a power law… and power laws encompass the phenomenons of the “long tail” and the Pareto principle (aka 80-20 rule).  This holds true for the world of seed accelerators, too.  The following is a chart of the funding that the 939 companies that have raised the most venture capital after going through an accelerator (so ~20% of all accelerated companies).

Acceleratorfunding normal

This is a pretty extreme power law; Dropbox has raised over $1billion in funding alone, but hundreds of companies have raised between $1million and $10million in funding.  (And thousands have raised seed rounds of <$1million).  Let’s see what happens when we check for a real power-law relationship by plotting both axes on log scales.

Log log acceleratorfunding

The result is a straight(-ish) line, which means venture funding of accelerator companies is a power law relationship.  (The math to prove it is fairly complicated and outside the scope of this post.)

But… do accelerators accelerate?

This is a very difficult question to answer.  Luckily there are some very smart researchers trying to quantify this.  Ben Hallen, Chris Bingham, and Susan Cohen have done some great work in trying to answer this question.  Essentially they’re trying to determine if companies that have gone through accelerators reach key milestones faster than companies that haven’t gone through accelerators.

They haven’t yet published their paper, so I’m not going to steal their thunder.  But their work should put the analytical horsepower to confirm (and disprove) various theories on accelerators.

I want to note that Yael Hochberg is another researcher in this field to watch; she and Susan Cohen lead the Seed Accelerator Rankings project.

Data Disclaimer

All data on company funding comes from Crunchbase.  Some companies don’t enter funding information in Crunchbase, and others don’t even have Crunchbase pages; in those cases the total funding would be even larger.  Additionally, I know a number of accelerators have funded companies that aren’t yet listed on Seed-DB; I continually work with programs to help make sure data is accurate but inevitably the data for many companies will be missing.

Also, I pulled the data above earlier this week in order to write this post; it’s already out of date with the funding rounds raised this week.  (Three Techstars companies announced over >$100million in funding within 48 hours this week, for example.)

Personal Disclaimer

I did my first research into accelerators in the summer of 2009, and created Seed-DB in the summer of 2012.  One year ago I started working for Techstars as a Product Manager.  This post represents my personal views, and not those of Techstars.  All data comes from Seed-DB alone.

[Check out the discussion on Hacker News]

comment 0

The weird reporting of YouTube stats

Here’s a paragraph from a recent New York Times article on Susan Wojcicki and YouTube:

Smosh, a pair of 20-something lip-syncing comedians, have roughly 30 million subscribers to their various YouTube channels. PewDiePie, a 24-year-old Swede who provides humorous commentary while he plays video games, has a following of similar size. The list goes on and on. For the sake of perspective, successful network television shows like “NCIS: New Orleans” or “The Big Bang Theory” average a little more than half that in weekly viewership.

My problem? – Why does any reporter try to compare YouTube subscriber numbers with actual television viewer numbers??

To be fair, the author does mention this several paragraphs later:

But advertising on YouTube isn’t like advertising on television. Subscribers don’t translate neatly into viewers.

But that is a woeful understatement.  YouTube subscribers have simply pressed a button once.  TV weekly viewership actually has watched the content.

I just logged into my own YouTube account and saw that I’ve subscribed to 115 different channels, everything from TED Talks to the America’s Cup to CGP Grey (who’s awesome!) to Matthias Wandel’s woodworking channel.  I rarely watch the vast majority of the videos from any of these channels.  The exception would probably be CGP Grey, who’s videos are so awesome that I set e-mail reminders to make sure I don’t miss them.  So me being counted as a “subscriber” for any of them is essentially meaningless… a vanity metric.

So what’s my point?  Beware YouTube subscription numbers, especially when they’re being compared to TV viewers.  What’s the best comparison?  Actual views of video content…  It’s definitely not easy to make that comparison, but whoever said life (and journalism) was supposed to be easy?

comment 0

Product design courses & Kickstarter

I love Kickstarter.  In fact, I’ve had to develop a few new habits around Kickstarter because otherwise I end up purchasing a whole ton of stuff that’s really interesting, but that I don’t actually need.

Recently I was thinking about design courses that most universities offer (at least in engineering departments) in students’ senior years.  It’s been a few years since I completed my degree, but I remember there were always at least a few design projects that were really interesting.  One friend built an automated bartender; punch in a code corresponding to a drink, and a series of windshield washer fluid motors would pump the right amount of ingredients into a glass to mix your drink.  (You still had to add ice and do any shaking/stirring required.)

What I would love to see is universities encouraging students to try to scale these projects by getting them funded on Kickstarter.  Just going through the process of trying to scale a one-off project to something more, and the marketing skills involved in that, would be hugely valuable to most engineering students.  If the projects actually got funded, it would mean students would learn lessons an order of magnitude more valuable… in actually engineering and creating and fulfilling products at scale.

I’ll admit that this doesn’t work for everyone… the space system design course I took as an aerospace engineering senior is not well suited to Kickstarter… to say the least!  But there are a lot of students that would learn so much more by taking their academic designs and testing them in the market.  If universities aren’t trying to encourage this today, they’re doing a disservice to their students.