SES New York: Landing Page Testing & Tuning

Filed in Conferences/Educ. by Matt McGee on April 12, 2007 2 Comments

Landing Page Testing & Tuning

Tim Ash,
Tom Leung, Google
Scott Miller,
Jamie Roche, Offermatica

Tim Ash

Conversion: Getting users to take the action you want them to take

Acquisition — Conversion — Retention
— conversion is the weak link in the process

(Tim just gave out $20 to someone in the audience who answered a question.)

What to tune? Mission critical parts of site
— landing pages that lead to trackable actions
— price of your product or service

On a page…you can tune everything. Don’t copy what your competitor is doing. They’re audience may not respond to same stuff as your audience.

How to Tune:
1) A-B split testing
— test one variable at a time to see which works best; keep that one
2) Multi-variate testing
— test several things at once; e.g., test two headlines and two call-to-action buttons simultaneously
3) SiteTuning Engine
— larger scale testing

Three Mistakes to Avoid
1) Ignoring your baseline — need something for comparison

(Tim just gave out another $20.)

2) Not enough data. Making decisions based on too little data.

(Another $20 given out as Tim is asking questions to the audience.)

3) Interactions are important. Mistake is to ignore/underestimate these.

(And another $20 giveaway. I’m coming to this session at every SES.)

Tom Leung, Google

Will talk about Google’s Web site Optimizer product

White Hat Testing
— don’t show bots something different from visitors
— variations should not differ to the extent that a user would be surprised on reading your site
— test tags alone won’t harm you in engine’s eyes

How to Test w/Optimizer
— choose a page to test
— choose which sections to test
— wizard will give instructions for pasting javascript into pages
— review experiment’s settings & launch
— view reports (after a couple weeks of data are available)

Testing Pitfalls
— too many combos being tested
— conversion goal too far out; easier to test how many start a funnel than finish it
— variations too subtle
— stopping test too early

Scott Miller

Good presentation, but it was two case studies and so not many opportunities for note taking….

Jamie Roche

“Campaign Relevance Optimization”


Click: optimize to get people to your site; match keywords in ad to keywords on landing page

Stick: goal is to lower “bounce”
— pages must be relevant and engaging
— “you searched for KEYWORD on Google” – helps visitor know they’re in the right place; works well on some pages, not on others
— “would you like to search our site for KEYWORD?”

Close: repeat primary message throughout funnel, i.e. – “free shipping”
— optimize forms and cart

[tags]sesny2007, seo[/tags]

Comments (2)

Trackback URL | Comments RSS Feed

  1. tleung says:

    Great summary!

    One minor point: I think I should have been more clear at the panel about the viewing of reports.

    Generally, we don’t recommend making big decisions until at least a week or two of data has been collected and you see a clear winner in the test.

    Of course, if you’re curious, you can view the status of the experiment anytime since new visitor/conversion data is reflected in the reports within a few hours of any activity after launch.


  2. Matt McGee says:

    Tom – thanks for stopping by and sharing some more thoughts. I liked your points about not stopping the test too early and not jumping to hasty conclusions. The more data the site owner has, the better decision s/he can make. Hope you’ll stop by again in the future….

Leave a Reply

Your email address will not be published. Required fields are marked *