Optimizely Roadshow 2014: Turning Data into Action (Part 2)

Dan Siroker, Co-Founder and CEO of Opti­mizely presented last week in Atlanta at the Opti­mizely Expe­ri­ence 2014 Road­show.  In his keynote, he posted a few ques­tions: “Would you want…

  • 764% more social follow­ers?
  • 23% more conver­sions?
  • 123% more content engage­ment?
  • Or 150% more customers?”

Does a one-legged duck swim in circles?   Of course we would!

Dan went on to show brand success stories in using Opti­mizely for a variety of tests to drive those results.  He presented testing types in 4 buckets:

  1. Capi­tal­ize on your traffic
  2. Opti­mize every channel
  3. Person­al­ize your expe­ri­ences
  4. Think like a scien­tist

Below is a recap of Dan’s exam­ples directly from his presen­ta­tion: here is a link

1. Capi­tal­ize on your traffic oppor­tu­nity

Don’t look at a single channel but look across chan­nels where customer activ­ity is growing.

This was a fasci­nat­ing way to show the lack of balance in indus­try focus… For every $92 spent on market­ing (getting someone to your site), there is only $1 spent trying to convert them! (Source: IBM TeaLeaf, 2012).

Don’t throw more money at media if you aren’t looking at the expe­ri­ence and improv­ing the backend!  A percent­age gain in conver­sion rate has a longer-term impact and will be more prof­itable at the end of the day.   Shown below, when PR or a promo­tion goes into market it will have a short lived bump in traffic or conver­sions, but if you focus on conver­sion rate improve­ments you get longer term rewards.

Another example was Code.org’s campaign testing differ­ent text and calls to action led to an extra 12MM addi­tional partic­i­pants to their “Hour of Code” project.

Code.org Optimization of Experience

1. Opti­mize every channel

Under­stand­ing your channel perfor­mance and behav­iors is impor­tant in testing.  Paid Search traffic will most likely act very differ­ently from email or social traffic.  There­fore you can utilize Opti­mizely to serve differ­ent landing page versions back on the refer­ral source.

Liftopia did this for Search engine traffic and simpli­fied the landing page to see a 20% increase in conver­sions.  Easy change, nice reward.

Knowing your audi­ence is key!  The oppo­site effect can also take place when testing if you forget to think about your audi­ence.  IGN wanted to increase video views and tested moving the video link in the navi­ga­tion to the far left (vs. far right).  Conse­quently the result was a 93% decrease in video conver­sions!  When digging into the data by segment the decrease came from return­ing visi­tors who simply didn’t see the video link where they were used to looking.  Know Your Audi­ence!

1. Person­al­ize your expe­ri­ences

Using testing tools like Opti­mizely for person­al­iza­tion of expe­ri­ences is done by a small percent­age of those using these tools, but it is a great feature to tap into.  Again, if you know your audi­ence and can tailor messag­ing and head­lines to that audi­ence it can pay off.   Iron Moun­tain person­al­ized content based on audi­ence segment and deliv­ered a targeted message to spark a 123% increase in content engage­ment.

1. Think like a scien­tist

Dan urged us to “Think like a Scien­tist” and be not only idea-driven (curi­ously creative and ques­tions conven­tional wisdom) but also data-driven (measure, analyze, opti­mize and iterate!)

Trunk Club is a great example of think­ing like a scien­tist and going against conven­tional wisdom of using the short­est purchase path flow.  As marketers and web design­ers we try to reduce the fric­tion in a conver­sion funnel and simply the process – usually by reduc­ing the number of steps of pages in that check out process.  Trunk Club went the oppo­site way in expand­ing the number of pages, yet in a simple way.

The Result was a 133% more customers.  At the new conver­sion rate of adding customers the new check out process has the ability to gener­ate $10–30MM in revenue the first year as Trunk Club customers.

Now that is worth testing for.

Happy testing, y’all.