Look familiar?

Surviving the Climb

Part 1 of 3: How one startup is surviving the Trough of Sorrows, the Crash of Ineptitude, and the Wiggles of False Hope…


If I had to classify the relationship status of Shelby the company and Shelby the product on Facebook, I’d definitely say “It’s complicated.” So as we prepare to release new product, here’s what it’s been like in the last sixteen months since we “shut down” our flagship products.

First, a very brief history of Shelby.tv

Dec 2010: Shelby is born on a whiteboard.
Jan 2011: The first lines of code are written.
Apr 2011: TechStars demo day, a private beta launch, fundraising.
Oct 2011: We launch Shelby “NOS” (internal code name) on iPhone, iPad, and on web. [The Techcrunch of initiation!]
Jan 2012: Demo “TouchPlay” at CES. 
Feb 2012: Begin the UX design of Shelby “GT,” our second version.
Mar 2012: Play with GOATS @ SXSW.
Apr 2012: Demo “GT” and win Startup Rally @ TNW Conference in Amsterdam (but failed to actually launch the product at the time).
Jul 2012: Realize that building/maintaining “NOS” + “GT” simultaneously is too much. Shut down “NOS,” but raise additional $2.2M.

When we shut down “NOS,” we were both lauded and laughed at. The laughs came from those who read TC and thought, amidst the hype of the ‘social video spring,’ how had we not reached escape velocity with everyone else? We were obviously destined for the dead-pool… if not for that pesky extra $2M of runway.

The obvious answer was that we were pivoting [drink!]. Wrong again. Pivot product decisions, features, and design? Sure. Pivot our mission, market or core strengths? Negatory. We started Shelby with a mission of being a consumer video app. Period. Those who knew us understood that we were forsaking the battle so we could better win the war.

So what’s happened to the product since?

Well, the biggest reason for the shut down was our underlying technical infrastructure. The way we built NOS did not set us up for long term success. I’m not talking about scaling issues (a la Twitter). I mean that we learned a lot from our initial launch, and realized there was a much better way we could collect and structure data such that we would have a great understanding of what people watch, what they might want to watch, and how to make recommendations for these things in the future.

I’m more than proud to report that in the last year, our technical infrastructure has grown in leaps and bounds, enabling us to do more than we ever thought possible. It goes without saying, but I’ll say it anyway… strong fundamental technology is paramount to building a tech startup, and rushing decision making in this department will create more pain than its worth later.

We’re lucky to have a great technical leader in Dan… no matter where he does his work.

So what do I mean by “grown in leaps and bounds?” Well, we’re now able to process some 4000 videos/second from social sources, that enabled us to track ~600M sharing actions across our known users, and from that, we sifted out 30M of the best videos on the web, all explicitly shared by human beings. For a startup ‘without a product,’ that’s pretty neat.


But that wasn’t all we did. Like I said, we have this badass infrastructure so we got creative with how to use it. One teammate came up with the “Video Graph” which would allow us to do ‘serendipitous search’ as we lovingly called it. When it was announced that iOS6 would forego the Apple-designed (read: skeumorphic) YouTube app, we hacked up Shelby Genius. Genius was essentially a search app that was powered by our Video Graph, so you could search for “surfing” and Genius would turn up some surfing vids, but also stuff shared by people who share surfing videos… skating, snowboarding, music videos, stunts etc.

I’ll spare you the details, but we were unable to get Shelby Genius out ahead of the YouTube app and ultimately it was a cool hack that didn’t get the resources needed to make it a success. What it did do for us though, is prove that the Video Graph had some real value in it, and was the first stab at Shelby giving video recommendations. Unfortunately, the bad taste left in our mouths from Genius made us move on from the Video Graph (and all discussions about recommendations) for a while.

Team product session. Summer 2012.

At this point, we were pretty convinced that one of the issues with Shelby was that people wanted a way of “creating.” We heard it from users “I want to make a channel of just videos about X.” We’d also always been excited by the “.tv” extension, so it seemed like a natural fit… reece.shelby.tv == reecepacheco.tv == my personally curated channel. We weren’t the only ones who were excited about it. Our old friend and mentor Fred Wilson had once said “I could have FredWilson.tv!” So we gave it to him (Literally, we gave him an old TV on which we’d slapped a Shelby sticker and wrote FredWilson.tv).

Thing is, Fred’s a pretty unique case, and not everyone wants to do that much work with video. They just want to watch entertaining stuff… which led us to our next product hack, curated Shelby channels. We had a ton of ways to hack these together — whether true human curation or powered by the Shelby backend — and it was exciting enough of an idea that we started building an iPad app for it. Open the app, pick a channel, and you’d be watching video instantly. There were just two little problems.

  1. We are not an editorial company, and doing that kind of editorial work was not in our DNA.
  2. We’ve always held fast that the ‘paradox of choice’ presented by most media properties is overwhelming (I’m looking at you, Netflix).

The long term vision for Shelby has always been a single stream of video that is “instant-on” like your TV, plays video continuously and provides you with interesting content based on some relevant context to you.

How far we had strayed.

But all was not lost… We just needed to do a little spring cleaning.

So we dropped the channels. We cut features ruthlessly. We cleaned up our design. We re-prioritized our iOS strategy and laid out a very simple roadmap. Where a year prior we’d commit to three major features, this roadmap was all about saying “no.” The new iPhone app would be an MVP. A simple, personalized channel of video that you could take with you anywhere.

And very soon, you’ll have just that.

Spotted @Shelby, a little inspiration from Seth Godin.

Product Lessons Learned:

  1. MVP’s are great, but take your time thinking through your technical architecture. Think big vision creation and long term value.
  2. “Simplify!” — Thoreau. Big vision doesn’t mean overcomplicated. Keep things simple and say “no” to a lot of stuff. Like, everything.
  3. Beware false positives. Standing face to face, most people will tell you “Yeah, I’d use that.” They don’t know it, but they’re lying. You’ll know what real usage looks like.

Next up: Part 2 — What went on in our heads…

Everyone else has already written more, better posts about the head-fuck that is starting a company. I’ll spare you my $.02 and focus on what it’s like being a company without a product, because we’ve lived it… https://medium.com/@reece/surviving-the-climb-in-your-head-8350168e1e1d

Like what you read and want to sign up for our monthly newsletter where we talk about building startups and lessons learned? Sign up here.