Marc Andreessen on Iteration in Technology Startups

Marc Andreessen co-founded both Netscape and Opsware, so he’s one of the few people to found not one but two billion dollar companies. Recently, he launched his new venture fund, Andreessen-Horowitz. (That is actually their website. I kinda like it.) So he knows a thing or two about startups. I found a video where he shares his thoughts on iteration and agile business for technology startups. The relevant section starts at the two-minute mark, but the video is good.

“If you look at the history of the major technology franchises, most of them started off doing something very different from what they were ultimately successful in doing. So Intel started out doing memory chips, and ended up doing CPUs. Microsoft started doing programming tools and ended up doing operating systems. Apple started out doing the Apple II, and ended up being most successful with the Mac, which had no bearing, no relationship to the Apple II. Silicon Graphics, my partner Jim’s company, started out making graphics cards for mainframes and ended up making supercomputers. Somebody the other day said that trying to understand what is going to happen – how a tech startup is going to be successful – is a little bit like looking at a sonogram and then trying to predict the baby’s hair color. It’s incredibly hard to tell.
The idea really matters, and the products really matter, and that all matters, but you know so little about the adventure you’re undertaking when you’re starting a new tech company that you, in our view, have to assume that things are gonna change.

- Marc Andreessen, October 2009 Interview
What do you think? Leave a comment…

Popularity: 2% [?]

how salesforce.com uses customer development

salesforce.com is one of the great enterprise software companies of our time. It’s one of my favorite companies – I love the product, the philosophy, and the philanthropy model.

Because I was so impressed with the company, I ordered Behind the Cloud as soon as I could.

I eagerly tore the book out of the Amazon box and devoured it, reading all of Marc Benioff‘s 111 business plays.  It’s consistently great advice from one of great Founder/CEOs of our time.

As with most start-ups, salesforce.com started with a vision – a web-based CRM application, available inexpensively.

As with most successful start-ups, salesforce.com also modified their vision according to customer feedback – both qualitative - gathered by talking to their customers, and quantitative – from web analytics on all salesforce.com accounts.

How salesforce.com Iterates on Customer Feedback

-Get a product out there and get people using it.

-Gather qualitative feedback from individual customers and quantitative feedback from the wisdom of crowds

-If you’re trying to move up/down market, spend time with the customers you want to understand your needs.

-Iterate on your service, being careful to add the minimal amount of features possible.

Marc Benioff on Customer Development

From ‘Behind the Cloud’

From the very beginning, we initiated a dialogue with these users about what was missing in the Sales Force Automation (SFA) application.  Salespeople routinely asked prospects about features they wanted and relayed their feedback to the product managers.  We queried prospects on why they decided not to go with us, and we spent time with large enterprises – customers that we weren’t initially able to serve – to learn what additionally functionality was required to make them consider our service.  We heard about missing features, such as the ability to track multiple products or a way to manage price lists.  We might have thought of these on our own, or we might not have, but we certainly wouldn’t have known which were in the greatest demand.

Thanks to our “no software” model, we have another way to listen to customer response.  The on-demand architecture offers us the opportunity to “watch” how users use the application.  We don’t do this in a Big Brother way, where we can see data or information about a company.  Rather, the system simply counts broad patterns anonymously and notifies us if there are issues.  These insights into how users are using our service allow us to learn about what they use and what they don’t.

As we evolved our service in response to customer’s needs, we faced the risk of changing our service too much – and making it so specific that it couldn’t commonly serve all customers.  If we wanted to continue to have mass appeal, we needed to apply changes with a broad brush.

Do you use the wisdom of crowds or interviews for customer development?  Which one works better?  Leave a comment…

Popularity: 1% [?]

How Zynga Uses Ghetto Testing and Minimum Viable Products

Zynga is a social gaming giant.  Farmville,their largest game, has more than 69 million monthly active users, making it larger than Twitter. Unlike many companies, Zynga also monetizes these users effectively with virtual goods and advertising. Zynga makes a huge amount of real money – while the company is private, 2009 revenues are projected to be around $100 million.

Today I listened to a podcast of a talk from Mark Pincus, CEO/Founder of Zynga and Bing Gordon of KPCB.  In this podcast, Mark shared some of Zynga’s methods of creating phenomenally successful social games.  As with almost all successful start-ups, Mark uses an appropriate version of customer development and rapid iteration.

Mark talked about how they assess demand for new products and features without taking up engineering hours.  All of his methods are pure customer development/agile start-up.

How Zynga Assesses Market Demand

- Create a 5-word pitch for a new product or feature

- Put it up on a high traffic webpage

- If it gets clicks, collect the emails of interested customers

- Build a ‘ghetto’ version of the feature

- Test everything

- Iterate constantly

Mark Pincus on Ghetto Testing and Minimum Viable Products

(Words are by Mark Pincus, transcribed and slightly paraphrased for clarity and grammar.  Emphases are my own.)

Mark Pincus:  We do something at Zynga that I call “ghetto testing.” I like to take someone who has a gigantic idea, usually a game designer, and they have some gigantic idea that this would just be great…  Maybe they really want a hospital simulation game…

We want to ghetto test it.  Again, we have so many bullets(engineering hours) we can fire, and we’ve got to just treasure and honor our engineers.  If we do our job right, they don’t get burned out.  They have a great life and we have successful products, so that’s what we want.

So I say to the marketing person or the product manager, “Describe it in five words.  It’s built.  If six months from now we built every dream you have, how are you going to market it?  Give me the five words.”

We’ll put that up.  We’ll put up a link for five minutes saying, ” Hey!  Do you ever fantasize about running your own hospital?” (laughter)  And, well, maybe you have!  In this economy, it’s the only growth area.

We’ll put that up for five minutes, and the link will maybe take you to a survey, where you give us your email and we say when this comes out we’ll contact you. If you’re really doing ghetto, it says ’404 not found’.  That’s bad.

So first you try to get the heat around it, you see how much do people like it, then…

(Brief discussion of usage metrics… they’re huge!)

Once we get to the point of actually building a game, or building a new feature, which we love Bing [Gordon's] idea of golden mechanics.  You should take away and steal it from us, the idea of not a game, but a feature that you can deconstruct and see that this interactive feature – a way to do a gift will drive virality or retention or revenues. So we put it in a feature we can build in a week – it’s a ghetto build we AB test it, we flow test it, we put it out to one percent.

We built a data warehouse with a testing platform so we’re running several hundred tests at any given time for every one of our games.  And no single user has more than one test.

So, we love tests.  When we see that it moves our metrics in a considerable way, that’s when we take it to be a full feature roll-out, and then we do the whole 2.0.

So, one example, we just turned on flowers in Farmville.   So now you can plant and grow beautiful flowers.  There’s so many place you could take flowers.  The holiday season is coming up – what happens if we let you level up your flowers and create your own custom bouquets?

But we don’t want to go down those paths until we test them with our users.

What’s amazing, and this is a feature that you will all have available as you enter this third internet era, is that you’re going to run a service, and you’re going to test things every week with your users, something that I never had available to me at previous companies.

It won’t be ‘build it for three months and hope and pray.’




Do you use ‘ghetto testing’ at your start-up?  How does it work? Leave a comment…

Popularity: 100% [?]