Reflections on startup life: Week 71

From data to instinct.

As a personal note, I really want to remember this last month. There’s an important lesson brewing here, I just haven’t quite put my finger on it. We are now in what I loosely think of as “The Valley of Death”. Charging forwards with no idea if we’ll make it out the other side unscathed.

We’ve experienced this at least three times before. Around twelve months ago, when we iterated from Twendly to Tribalytic. From week 36 to week 40 when we replaced the Tribalytic back end with Hypertable. Then again around Week 50 up until Christmas when we were trying to pivot to Trunk.ly and balance contracting.

So what is it? From a business perspective, it’s an untested hypothesis — some thing game-changing you’re doing that you find hard to tell if it’s right or not. I was chatting with Alex this morning about this and he summed it up really well — it’s where you’ve moved from a Data Driven phase into an Instinct Driven one.

It’s scary because there’s now no one to blame but yourselves. There is a lot of talk about Lean Startups and customer validation. One of the tenets of Lean is testing customer reaction with an MVP (minimum viable product), yet every now and again, despite the fact that I think we are very Lean most of the time, we end up in some “fat” phase, running on instinct striving for some BHAG.

Why is this? I think part of the answer is in this post here on Ending the Myth of Why Ideas are Worthless. It’s also worth pointing out that our idea of “fat” is something that takes four to six weeks to validate, rather than the usual two days.

Still, no matter how lean you want to be, I think there comes a time when you just grab the ball and run. You want to deliver a faster Tribalytic? You have to put your head down into engineering for a month or more to make it happen; and this is the key point, without knowing you’ll pull it off. Want to switch from Twendly to Tribalytic? You can show people pictures all you like, but there’s six weeks before you’re in San Francisco, and you just have to pull it off. We are here again with Trunk.ly.

It’s an interesting landscape — bookmarking related startups are struggling around us — Delicious is in the closing stages of a sale which looks to be for only $1M dollars, Reddit are down to only one developer (which doesn’t mean they are failing, but they’ll certainly struggle to innovate at the moment and run a multi-million user site) and Digg’s founder resigned to go build another startup. Although Reddit and Digg are somewhat different from Trunk.ly, there’s enough similarities to ask the question “just who do you think you are — building a content curation service when others have failed or are failing.” Rather than iterating just on features that users want (they want a faster horse), we need to step up and show the vision we’ve had cooking. If we’re to fail, then let’s fail fast. And that means running on instinct for bit.

We can play it safe and almost certainly go home (because gradual innovation won’t shake up the market and as I just pointed out, the market as it is, is pretty shaky), or we can run on instinct for a bit and show what Trunk.ly can be and how we envision this market to be different from where it’s at today.

Let’s review this in a couple of weeks and see how it all pans out! If it doesn’t, it’s four weeks sunk time for no net improvement, but on the it’s really critical to test this vision thing; Trunk.ly is a great start, but it’s time to have it really start to show where we want to take it.

For the record, we survived the valley of death the last three times. Here’s hoping we make it again.

Highlights

• Completed the contract work
• More user measurement in place
• Tony Leung offered to help out with some PR, that’s going well so far (a whole post on this next week I think)

Lessons Learnt

• I think we are in the middle of learning them! Still — funnel analysis is critical. Having (finally) got the funnel in place, there are lots of places we can improve to drive signups.

Goals this week

• Backend performance improvements needed to move to the user testing phase of the new vision, which is “coming”.