All posts filed under “Analysis

comment 0

Serial (podcast) – who(ithink)dunit

If you haven’t heard the hype already, the “Serial” podcast that spun off from “This American Life” this fall has become ridiculously popular.  With the final episode of the season airing this week I thought I’d post a few thoughts about what I think happened.


The Serial podcast team is essentially re-investigating a murder from 1999 in Baltimore.  A young Korean girl goes missing after school one day, and her body buried a month later.  While there’s no physical evidence, an anonymous call kicks the police off on a thread that leads to the girl’s ex-boyfriend, Adnan Syed.  He’s convicted largely on the testimony of an acquaintance/friend named Jay, who also led the police to the location of the girl’s car (which had also been missing since the day she disappeared).

Needless to say, there’s a lot more to the case… that’s why there’s been over 10 hours of a podcast to listen to!  But it’s the biggest brush strokes.

My thoughts

There are a few major points that I think are true:

1) It’s 100% clear that Jay was involved in the murder somehow.  (He did plead guilty to “accessory after the fact” in return for testifying against Adnan.)  After all, the only physical clue that ties any of the people involved to the crime was Jay leading police to the victim’s car.  The open question is exactly how much was Jay involved?

The crazy thing about Jay’s testimony was that it changed so substantially between his 1st and 2nd interviews, and his 2nd police interview and the actual trial.  Oh, and key pieces of it were essentially impossible.  (Key events could not have happened like he said they did.)  I understand that in real crime details don’t always wrap up neatly with a bow, but when major details don’t match key events and testimony, that’s a problem.

2) The case should not have been prosecuted.  The evidence was incredibly weak to begin with, and I have to believe some of the lawyers and detectives that the Serial team consulted with, who universally said that there wasn’t enough to make a case.  Adnan’s lawyer wasn’t particularly good however, and was disbarred within about a year of Adnan’s trial.  I have to believe with better counsel the case was easily beatable.

3) I don’t believe that Adnan had a strong motive to kill the victim.  Yes, they had been dating and intimate, but they both seemed fine and had moved on since.  Neither of them seemed hung up on the other, and I just can’t figure out the motive for Adnan to commit anything like a murder.  The police really played up the stereotyped conservative Muslim attitudes toward dishonour, but these were just high school kids living in a far more real/modern America than their parents would have liked to believe.

Unfortunately, there doesn’t seem to be anyone else that had a motive to kill the victim.  Jay (who was involved somehow) didn’t seem to have one, and there didn’t seem to be anyone else who could have.  This strikes me as the biggest missing piece of the puzzle.


I don’t think Adnan committed the murder.  But there’s nothing pointing to a) any evidence that he didn’t do it, or b) any evidence that someone else did it.  So unfortunately I think he’s going to be stuck in prison.

It’ll be interesting where the Serial team finishes this podcast… I strongly suspect they’re going to come down like I did.  The balance of evidence should say Adnan did not commit the crime, there’s no evidence to date of anyone else doing it, but since Adnan has already been convicted there’s really nothing anyone can do.  The bar is set so high for appeals that the status quo prevails.

comment 0

Trending on Product Hunt during YC’s Demo Day – the Seed-DB experience

I had a bit of a weird experience on August 19th this year.

And the circle of self-promotion closed, as Seed-DB was promoted on Product Hunt, driving people to see the YC listing on Seed-DB, which showed users that YC’s class had Product Hunt in it.

What does this mean for stats?  Well, let’s go to Google Analytics.  Here are the hourly pageview stats for the week on either side of trending on Product Hunt:

Analytics hourlyAug

I think the first spike came after initially trending on Product Hunt, and the second spike came after the Product Hunt daily e-mail was sent.  How does this compare to Seed-DB’s traffic across the year?  Well, it was by far the biggest spike:

Analytics dailyYear

And here are the stats from the “referral” acquisition channel for a week on either side of August 19th.  Product Hunt drove over 2000 people to Seed-DB:

Sources Aug

Since then…

Product Hunt has grown massively since then, and would now probably drive an order of magnitude more traffic for the top trending products.  For me, August 19th was a bit of a surreal day but was pleased about the coincidence in the timing.

I was recently going over the statistics, and Seed-DB has grown virtually every metric by 30-40% compared to 2013.  While I could look at that in startup terms and say “meh”, because I don’t do any marketing of the site I’m pleased that more and more people are finding it and finding it useful.  I’m going to be adding a few key features in the coming months, so please stay tuned!

comment 0

Accelerators and the focus on Demo Day

If you check out seed accelerators for long enough, you’ll come across one relatively consistent criticism.  (Particularly for the lower quality programs, I have to say.)  That criticism is that accelerators focus way too much on Demo Day.  I believe that founders that say this don’t understand the real “why” behind the preparation.

I joined Techstars at the beginning of June this year, and in that time have seen the preparations for the Demo Days of the Techstars London 2014 batch of companies, as well as the first Barclays Accelerator batch of companies.  So I’ve already seen, up-close-and-personal, two cycles of companies spending time and getting ready to pitch at Demo Day.  And Demo Day is important: there are hundreds of angels and institutional investors there and it’s a once-in-a-lifetime opportunity for most companies.  They need to work hard to make the most of the opportunity.

But the subtle secret about preparing for Demo Day is that it’s not just about one 5-minute pitch, it’s a month of deep critical-thinking about how to communicate a product, a company, a market, a team, and an opportunity.  Sure, the direct output is that 5-minute pitch, but founders learn how to give a one-line description of what they do, an elevator pitch about their company, and how to talk about the company in ways that really resonate with a particular audience.  This process, and particularly the feedback from experienced entrepreneurs and mentors, is critical to founders.  (And while it involves the whole team, it should only be the day-to-day job of the CEO, leaving everyone else to continue working on the company.)

Let me give two examples from the Barclays program:

ClauseMatch – Evgeny from ClauseMatch was not a natural speaker, and his company (a platform for contract negotiation) is in the legal world, which tends to make peoples’ eyes glaze over.  And at times, he struggled to communicate how revolutionary their product is.  But he cracked it with a simple (and amusing) anecdote to start his Demo Day pitch.  He took the audience back to 1995, when Microsoft Word introduced “Track changes” and e-mail started to become widely introduced.  For the first time, instead of faxing manually annotated contracts back-and-forth, lawyers could e-mail Word files back and forth… it was a revolution.  Then he made a simple statement: after twenty years of internet and cloud technology development, lawyers are still working the exact same way.  It was a massive “a-ha” moment for the audience that grabbed their attention for the rest of his pitch.

GustPay – Werner from GustPay actually spent a bit of time at the start of his pitch talking about Disney… specifically about the NFC wristbands that Disney has developed for their theme parks.  He talked about the “magic” of the experience, in that the wristband becomes their ticket and their wallet and their room key, and everything they need for their stay.  Then he told the audience that Disney spent >$1billion in developing this technology, but GustPay provides the same experience for venues and events for just $1/wristband.  Again, it was an “a-ha” moment that got people to recognise what they did, and why it was important.

Being able to communicate your startup to a wide variety of audiences (investors, early adopters, sales prospects, press) takes a lot of hard work.  And while it may seem all that hard work is just in service of a 5-minute pitch, the real benefit is far, far beyond that.

comment 1

Frustration with a taxonomy for startups

tax·on·o·my – noun – the branch of science concerned with classification, especially of organisms; systematics

I’m frustrated with the current lack of any standard taxonomy for early stage startups.  Chalk it up to a little bit of obsessive/compulsive behaviour, the desire to better compare like-for-like amongst startups, my years of experience in industries where these taxonomies existed, and also trying to make better connections between the corporate partners and friends of Techstars to the Techstars portfolio companies.  Startups are messy and change ridiculously fast, so a taxonomy will never be as rigorous as what exists in the world of biology.  But it doesn’t mean we shouldn’t try!

Angellist probably has the best structure of markets that I’ve seen so far, but the way Angellist structures these markets behind the scenes is actually a fairly deep web of interconnected markets.  While that makes sense in that a graph represents how markets are related to each other, the way they’ve built the graph can make it difficult to analyse startups and markets more broadly.

Not only that, but I think startups are complex enough that there should really be multiple dimensions in building a taxonomy for them.  These are the dimensions I’ve been pondering recently:

  • Market (ie, FinTech, HealthTech, Advertising, Infrastructure, etc.)
    • This is where there needs to be multiple layers of “markets”
  • Revenue model (Advertising, commerce, subscription, etc.)
  • Platforms (desktop, mobile/iOS, mobile/Android, hardware)
  • Orientation (consumer, enterprise, marketplace)

Why am I posting this?  Frankly, I’d love any and all feedback.  I’d like to get to a point where there’s a taxonomy that helps people like me understand and directly compare and contrast startups that are doing similar things (in different markets), or different things (in similar markets), or any combination thereof.

If you’re interested in this project, or would like to help, please comment below or get in touch with me directly.