How PBworks Used Lean Startup Techniques

PBwiki was a Lean Startup(but didn’t realize it)

AKA: An engineer’s journey into the wild world of business and some

lessons learned along the way

Lean Startup Circle Talk, 9/16/2009 by “David E. Weekly” <david@pbworks.com>

Executive Summary

• Startups are tough and fast.• You can’t just count on data. Use your gut.• You’ll need to have a lot of bad ideas.• The good ideas will often need to be

dramatically reworked.• ….• $$$ PROFIT! $$$

Business Is Not Sexy.(This is not you.)

(Lean) Startups Are Poor.

We’re short on cash, people, and time.

(Lean) Startups Are Ignorant.

We aren’t sure what to build or for whom.

(Lean) Startups Are Misguided.

• We don’t know which ideas are right.• When ideas are right, we don’t know which parts.

(Lean) Startups Are Prone to Failure.

Lots of experiments mean lots of mistakes.

But Why Does It Work?

• Humility w/Hubris– Having confidence in the correctness of a general outcome versus

making an idea your baby.

• Scientific Method– Form reasonable hypotheses given what you know, test them, iterate.

• Low Bullshit Factor– Ya gotta eat. Cash is king.

Don’t Go Tabula Rasa…

• “Blank site with just a UserVoice plugin”– Not so helpful.– First feature requests tend to be:

• more porn• more dinosaurs

Minimum Viable Product

• There’s a stone in stone soup.

• Start with the simplest thing that could possibly be useful to anyone.

• PBwiki’s first motto: “Ghetto But Useful”

Hunches & Creativity

• The market is an oracle but not a seer.

• Listen to customer problems more than customer-offered solutions.

• Then get creative about solving those problems.

So…• Deliver a lot of ideas.

• Count on most to fail.

• When you find one that is clearly not a failure, drop all the others.– (Even though it hurts.)

• When the market takes the idea a new direction, roll with it.

Ideally…

You can rapidly iterate on your ideas

to make sure you’re building the right thing

for the right market.

(but….)

Data Can’t Tell You Everything.

A/B Testing Low-Frequency Events

How About…

95% confidenceof a >10% uplift

for a 3x/day action (e.g. making a $10,000 sale)

?

If the proposed change is the only change tested against the baseline and it represents a 50% improvement over the baseline, you’ll still need over three months of data over three months of data to be statistically sure it’s made at least a 10% improvement.

It’s Time for a Fallacy!

Post hoc ergo propter hoc.

Optimize Successful Correlations

• Find variables that correlate with success that occur at high frequency.

• Judge which are likely causal variables.

• Optimize for these variables.

• Don’t game yourself!

Human Judgment Is Key

• Your gut provides the ideas, strategy, and approach hypotheses.

• The analytics judge if you were right.

• Started messing with private wikis (2002)• Started Coceve, Inc. (2003)• Got sick of setting them up by hand (2004)• Decided to let people set themselves up

(Drunken 1st SHDH, May 27th, 2005)• Hired help (November 2005)• Got financing + an office (September 2006)• Got VC investment (February 2007)• Hired a CEO (June 2008)

Was PBwiki a Lean Startup?

• Went for first three years without a salary.• Only employee for first two years.• $60k of friends-and-family cash.• LAMP + Danga• Bought servers off Craigslist.• Rapid-response Customer Forum– “Thanks for reporting this an hour ago; it’s fixed!”

The Products

• Hosted project management (2003)• IM Smarter (2004-2005)• BotBlock (2005)• SingleStat.us (2005)• Thwapp (never launched)• PBwiki (2005)– PBworks Project Edition (2009), full circle!

PBworks Evolution

• Ad-supported Free Wiki• Consumer Freemium Wiki ($5/wiki/mo)• Tiered Freemium Wiki ($10-100/wiki/mo)• Paid Content (FAIL)• Educational Wiki ($99+/wiki/year)• Business SaaS Workspace ($8/user/mo)• Enterprise Collaboration ($20/user/mo)• Secret February 2010 Awesomeness

The Model

• Rapid-release (Agile method)• Focus on being useful to users.• What’s the simplest useful thing we can do?• Ignore technical fanciness. Ignore competitors.• Note: simplicity empowers scale.• Ask your users what the heck they’re doing.– They may periodically surprise the heck outta you.

Pivot Questions• “You want to pay for an Internet service?”

• “You heard about us in Chicago?”

• “How big is the education market?”

• “Am I passionate about enterprise sales and marketing?”

• “What would this look like in a movie?”

Sci-Fi

• What would be ideal for the user?

• Backtrack to what’s possible, build a roadmap.

• Write out the user story, share it!– (we even made a short movie.)

Business Goals

• Don’t focus on acquisitions:– “It’s hard to sell a company but easy to get bought.”

• Forget about IPOs for now.• Acquisitions are real but rare.• Create value, not lottery tickets.– Craigslist, The Math Works, HotOrNot

• Focus on making money.– This leaves the most options open!

Executive Summary

• Startups are tough and fast.• You can’t just count on data. Use your gut.• You’ll need to have a lot of bad ideas.• The good ideas will often need to be

dramatically reworked.• ….• $$$ PROFIT! $$$

The Whole Point

Now go kick some ass!

david@weekly.orgtwitter.com/dweekly