FOWA: Launch Late to Iterate Often (Dick Costolo) [en]

Here are my live notes of this Future of Web Apps (FOWA) session. They are probably incomplete and may contain mistakes, though I do my best to be accurate. Chances are I’ll be adding links to extra material and photos later on, so don’t hesitate to come back and check.

Dick: a bunch of startups, last one successful (FeedBurner), so now people think he knows what he’s talking about. 😉

We hear a lot about how cheap it is to start a company now. Lessons learned that are somewhat counter-intuitive to what is usually thought in this industry.

It’s true that you can get a company started without much money, but it still costs a lot to scale.

Cofounders: unequal equals. Better to treat all cofounders as equal. Unequal brings problems (“yeah, sure you want to do that, you have 75%”).

Dick and cofounders never build business plans anymore. Business plans are things that people write to try to make things they want happen the way they want them to happen. Dick doesn’t think investors read them anyway.

Disagrees with trying to evaluate the size of the market. You can’t know. e.g. eBay.

Location: FeedBurner, everybody in Chicago. Believes there is no strategic benefit in locating a company in the Silicon Valley. Actually, better to be away, you’re distant from the echo chamber. Self-perpetuating myth. Benefit in buzz in being in the Silicon Valley, but do you really need buzz to be successful? For Dick, no benefit in the long-term success of the business.

Cash. You always need way more cash than what you think that you’re going to need. Estimate, then multiply by 2.5, and it was even a bit tight. The leading cause of companies going out of business is running out of money. So raise as much as you can. Don’t run out of business.

VC funding is great. Find the right investors. Raise money when you don’t need it. You can get better terms for venture investors. When you start raising a few millions from VCs, you’ll start seeing legal/jargon VC terms (preference, multiples, participation). They’ll tell you they’re standard deals, but there is no such thing. So learn to understand those terms. (e.g. on Dick’s blog, and other places).

It’s better to own a smaller piece of a bigger pie than the opposite. Everybody needs to be happy about what’s going on. Everybody employed needs the same kind of deal (options, equity etc.), keeps goals aligned, and everyone is treated the same way. Even if it’s the “only way I could get that guy”.

Hiring. Take the guy who runs the fastest and then figure out where to put him. Don’t go out to hire a VP of sales. Look for people who are best available athlete, well-rounded person for this kind of role, but able to zig if necessary. steph-note: …any startups looking to hire? 😉 Dick prefers flat organisations. Hierarchy begets bureaucracy. Problem with flat organisations: when there are under-performers. Replace hierarchy with tools. Deal with this by having employees come up with their own KPIs (measurable!)

Growing the team: mistake = hiring sales and marketing too soon. Once you start selling and marketing, things need to be cooked and ready to go. Without that, you can iterate rapidly. Speed of execution is a competitive advantage of small companies over big ones. Wait until you’re ready to go to market.

Product development and business strategy (1-4 years)

Visit to the eye doctor. Iterate on everything. Disagrees with “get a crappy version out there”, because then you have to iterate with that version that is out there.

Day 1: feed stats, but knew they wanted to do more later. They waited until they had a basic underlying architecture to be able to extend the service before they launched. It didn’t do much, but was ready for building more. So that allowed them to iterate very rapidly. “How are you guys rolling out features every month?!” Spent the first 5 months building that underlying architecture for extensibility.

Let the market tell you what the business model is (cf. Twitter). Open system with APIs, help the market tell you what the business model is. Lock-in is bad for business. APIs lower the barrier to entry and to third-party service development. Lock-in creates barriers to entry. steph-note: so does the fact you don’t own your data.

Revenue plan: don’t kid yourself. Goes along with “don’t run out of money”. You’ll never make as much money as you plan, or as fast. VCs don’t pay much attention to it. Those plans are always wrong and at least a year late.

Don’t spend months and months trying to get your pricing right.

Strong advice: don’t worry about your exit strategy, worry about everything else, and also be competitive on your merits, not on how much the other guys suck.

Let your company have a voice and a culture. It’s harder to make your language sound antiseptic.