Rethink your Startup Website: Make a Net, not a Funnel

http://www.whattofix.com/blog/archives/2012/04/build-a-net-not.php

by edw519 • 14 years ago

52 10 14 years ago

Traditional A/B testing assumes that there is one goal, conversion, and you tweak one thing at a time to reach that goal.

This has the same end-point, make something people want, but the methodology is completely backwards. Cover all the bases, provide all kinds of chances for users to do stuff, then figure out what your goal should be.

Interesting to see what some more experienced start-up guys think of this.

harlanlewis• 14 years ago

I've most often seen this tactic in secondary products, with the goal of learning from (hopefully) representative visitors in a way that doesn't risk presenting an unfocussed core offering.

This sort of safety-zone data mining doesn't have to be limited to perfecting your pitch, either. I don't have any inside info, but Massive Health's The Eatery struck me as a very clever way to learn an awful lot about their potential users for whatever they're really working on behind the scenes.

edash• 14 years ago

This isn't an either / or decision for a company. It makes sense to create a net if you're early in the lifecycle of a company or product—before it's reached product/market fit. Once a startup has reached product/market fit, however, creating a funnel and A/B testing the hell out of it is probably a good idea.

To use an example from your post, now that you know what people are looking for (and how they're behaving) on your paystub page, you can begin to tweak the design, layout, and flow of that page to optimize whatever it is you want to optimize.

With the above in mind, "make a net, not a funnel" is another way of saying "don't optimize prematurely".

rs3123• 14 years ago

In the article, OP wrote about Facebook: "They didn't want you to do anything but what you wanted to do naturally, they just wanted to pay close attention and help." Does OP or anyone else have specific examples of what Facebook did in their early days?

hsuresh• 14 years ago

>> Confused about how so many people could visit and still I was clueless about what they wanted, I logged on to the Hacker News IRC channel and asked the room "Can anybody help? I have a site I don't know what to do with."

There is a Hacker News IRC channel? Anyone know the details?

astrofinch• 14 years ago

Google suggests it is #startups on irc.freenode.net

chrisacky2• 14 years ago

irc.freenode.net #startups

It's the "official" un-official channel.

It's easily my favourite channel. Currently has 458 users joined. Hit us up!

CCs• 14 years ago

From your web browser:

http://webchat.freenode.net/?channels=startups

wicknicks• 14 years ago

How do people react to fake buttons? I'd suppose it makes them more annoyed. Isn't that counter-productive? Or is it worth the risk?

chrisacky2• 14 years ago

This is actually quite a useful technique yet equally unknown method of figuring out product viability from your own user cohort.

The overall concept...

- Not sure if a some FizzBuzz feature is required on your site?

- Not sure if people are interested in your awesome newsletter series on Blue Widgets?

- And in any event, after you've created FooBar... What will the take up be if you add it to your site?

Well, before you spend 4 days coding your new and exciting product, only to find out that no one actually wants to use it... do some market research on your own visitors!

How?

Easy... put the feature on some of your sites pages and purposefully 404 it, and when they click through, record the event.

Just provide some message saying "Ooopsey, looks like we are having some problems!".

Run the test until you are sure that either: it's working, or it's a total flop.

Guess who's already doing this? TripAdvisor. Listen to Kaufer explaining how at TripAdvisor they purposefully 404 their users into figuring out if something is worth doing.

http://techcrunch.com/2011/12/22/founder-stories-tripadvisor...

Rethink your Startup Website: Make a Net, not a Funnel

Related Stories