Presentation by Kristen Mirenda (Lead Product Manager) and Karl Schultz (Senior Interaction Designer) to the Lean Startup Circle Meetup in San Francisco, December 13th, 2012.
There's a video of this presentation at http://justin.tv/leanstartupcircle/b/347873496 (we start at 1:06:00).
"Inspired to apply Lean Startup techniques at their decade-old company, Hotwire.com is using an 'innovation sandbox' to explore potentially disruptive changes to its hotel booking product, without putting its existing business at risk. Members of the hotel product team will offer a case study in using lean methodologies at a well-established company – how they are making it happen in such an environment, and why they want to experiment boldly on an already successful, market-leading product."
3. team
• Dedicated to product
• Embedded analyst
• Connected with users
• Ideas come from anywhere
photo credit: karlsbad (Flickr)
4. • Hotels provide unsold rooms
• We sell with huge discounts
• Posted price – no bidding
• Learn hotel after you book
opaque
photo credits: karlsbad (Flickr)
8. “hate it - can't even
sort anymore”
“I don't like it because you
cannot filter the results or
even sort them.. What
were you thinking?”
feedback
“absolutely blows...pure
garbage. need to be able to sort
asap. i'll come work for you and
help you figure it out. wtf.”
11. Kristen Mirenda
Lead Product Manager
Karl Schultz
Senior Interaction Designer
thanks
photo credit: karlsbad (Flickr)
Notas do Editor
KRISTENI’m Kristen Mirenda, product manager from HotwireThis is my colleague Karl Schultz, interaction designerWe want to thank David for bringing us into the Lean foldHe’s been an amazing coach this past year
KRISTENHotwire is a discount travel booking website.By traffic we’re the #3 online travel agency after Expedia and PricelineKarl and I work on the consumer-facing hotel booking productIt’s Hotwire’s flagship product and the star of our tv commercials
KRISTENA few words about our teamWe’re a 16 person Scrum team dedicated to the hotel product.This includes an embedded data analyst.We started exploring Lean techniques about 9 months ago.And we really try to involve the whole team.Stating hypotheses is really helpful.The analyst presents test results at our sprint demos.I’ve seen the team dynamic evolve quickly.Especially with engineers.They started asking to go to user labs.They propose their own product ideas.They’re even building POCs in their slack time.Generally the team has become more open and fun.I credit Lean thinking for a lot of that.We spend 40-50% of our time doing Lean stuff.The rest of the time is spent keeping the lights on.
KRISTENRaise your hand if you’ve ever booked a hotel without knowing exactly which one you’ll get.That’s our core business model. It’s called opaque booking.It’s a way for hotels to sell their extra empty rooms at big discountsBut we hide their names so they can still offer low price guarantees on regular bookings.This was disruptive in its day – that day being the year 2000.Now it’s an established product representing a significant portion of Hotwire’s revenue.So why mess with it? I’ll let Karl explain.
KARLwe started sitting with potential customersbeen doing user testing – but only new designsbook a hotel room – on any sitedo naturallygo to Hotwireafter 5 minutes, looked like thisflow was out of synccomplicated business modelcouldn’t figure out where the hotel wasa COHORT “confused users”big opportunitykept it working for everybody
KARLtest and iteratefreedom to experimentvalidation in labs, even other siteshow do we protect this business?prior we made small tweaks, how to test big?Test big against the main site (control)you want live data.strategy straight from Lean Startup, the Innovation Sandboxshadow site where we can test hypotheseslive data and real trafficworking code – we can easily productizewe can restrict and control exposuredivert traffic within 24 hours
KARLCan we help the confused user?This was our MVP which rolled out to 1% of trafficIt addressed the pain point where users had trouble orienting the hotels geographicallyMaps are downplayed on our current siteWe wondered, would they want to use a big Google map in the middle of the page?You can’t sort, filter, or even change your searchBecause none of those things mattered for our hypothesis about map usageIt was definitely way out of our comfort zone putting something like that out into the worldWe did build in the ability to switch it off on a few hours notice if something really blew up.In the upper right you can see we also solicited feedbackInitially it was pretty negativeThough indifference would have been even worse
KARLThese are a few actual comments from that MVPMany people did tell us they liked the mapAnd engagement metrics bore that outBut everyone yelled about the missing featuresThe majority of complaints were about sortSo guess what we built next – that’s right, sortAs soon as we launched sort, everybody complained about filtersSo we built filtersAnd so onWe’re still gathering feedback in this wayIt’s helping us validate our backlog of features
KRISTENThis graph shows conversion difference between the Sandbox and the original control site.Zero percent being no change.The orange lines are major iterations.As you can see, the MVP was way negative for conversion.Good thing we only rolled it out to 1%.We were looking for a baseline to measure the next iteration against.As soon as we got enough data to guide the next iteration, we turned it off.The data was an investment paid for in a few lost bookings.But as the gap closed, we could afford to expose the Sandbox to more users.And more data meant faster learning.There’s a period where things dip back down between iterations 3 and 4.We tried to do an MVP test of a totally different hypothesis, within the larger test that was still running.That didn’t work out too well. Measurement became very murky.In retrospect I’d recommend AGAINST testing inside of other tests.With our last iteration, we crossed the line of parity (happy face)We still have lots of features and optimizations in our backlogSo anything we do from here on should be a net gain over the old product.
KARLSo based on our experience, we say:If you are in an established enterprise, you don’t need to feel left out.There are opportunities out there for you to use the Lean approach.Don’t let the startups have all the fun