Live Blogging A Webinar: Website Optimizer – What Should I Test?

By jameszol
Published on March 11, 2008

Webinar: Website Optimizer – What Should I Test?

Host: Tom Leung – Google Website Optimizer

Guest Presenter: Bryan Eisenberg – Future Now (blog is at GrokDotCom)

Slides will be archived and available within a few weeks on the Google Website Optimizer website in the resources section.

Tom Leung’s Presentation:

What is the website Optimizer? (For those not familiar with the product.)

It is part of an entire system of tools that Google provides to website owners that will help improve their website – AdWords and Analytics are a couple of these tools. AdWords brings the traffic in, Analytics monitors that traffic, and the Website Optimizer helps to improve conversion rates from AdWords and results can be measured in Analytics. They can be operated independently but using all three will bring a synergy to the table.

How does the Website Optimizer Work?

It essentially let’s you perform controlled experiments on your website by taking the guess work out of increasing your conversion rate with your website. Website Optimizer basically asks your users to vote with their actions.

Some features are A/B split testing, Multivariate Testing, Follow Up Testing, Works with ALL traffic – not just ppc, Statistical Analysis, Platform Independent, consistent updates, and more. (You do need an AdWords account to access the Website Optimizer…)

Benefits are that it is absolutely FREE, Easy to use, Does NOT impact SEO, Backed by Google, big increases in conversions are not uncommon, there is a variety of support in discussion groups, tutorials, consultant network.

Quick thoughts on Stats and Reporting

Combination reports show the customer’s favorite page combination. Take the debate out of website design and improvement by testing with the Website Optimizer and reviewing the reports/stats.

What To Test

This is a rendition I did a while ago on the graph about which pages to optimize used in this webinar (I think Google designed the original visual?):

Every Page Has an Optimal Recipe

Info Rich vs Short n Sweet

Left Layout vs Right Layout

Fact based vs Aspirations/emotions

Leads vs Sales

There are definitely more combinations and tests to conduct besides those listed above.

There isn’t ever really a set answer for each website. The only way to know which combination will work is to simply conduct tests.

More Ideas for Experimentation

Headline / Image/ Call to action




Embedded Widgets


RSS button clicks

Colors, Fonts, Sizes

Sales Incentives

Newsletter subscriptions

Form Fields

And many, many more.

Everything depends on your audience, a portion of your audience will react differently to each test. You can find out more about your audience with testing…do they like certain kinds of widgets? Do they like videos? Testimonials? Etc.

At the end of the day, there are an infinite amount of tests you can perform. Focus on the tests that are most likely to give you the results you are looking for.

Best Practices in Marketing Experimentation

Test a small number of variations.

-Rule of thumb is about 100 conversions per combination

Test bold changes

-If you can’t see difference between two combos in 8 seconds, visitors probably won’t either

Consider early indicators if you don’t have enough conversions

-E.g. if you have modest conversion volume, optimize for leading indicators such as request info, view product details, remain on page features page >5 seconds.

Don’t jump to conclusions

-Less than 2 weeks is no good, focus on absolute conversion differences, don’t get too excited by silver or green or red

Bryan Eisenberg Intro

Bryan is co-founder of FutureNow, Charter member of Google Website Optmizer Authorized Consultant network, Blog located at, Pioneer in persuasion architecture.

Bryan Eisenberg’s Presentation:

Case Study of (Multicategory Page)

Overstock was noticing a 90% exit/bounce rate

It’s about understanding people first

Test small variations so you can get the learnings and understand the variables

Secret of Online Conversion…

Dates back to 470 BC when four personality types were written down – the GrokDotCom translation gives us four types/patterns as displayed in this graph (a rendering based on Bryan’s slide/image)

Four types or patterns: Competitive (Fast Logical), Spontaneous (Fast Emotional), Methodical (Slow Logical), Humanistic (Slow Emotional)

*Holy crap Brian talks fast, flips through slides like crazy and bounces around sporadically (probably not Bryan but software/conferencing issues)…very hard to follow while typing so this is going to get choppy…

Eyetracking Slides (summary found on GrokDotCom)

Personality Preferences

Spontaneous people are looking for features

Humanistics care about reviews

Methodicals find by generally sorting

Competitives search by specifics that they have in mind

Went through 4 personality types for – Does it accomplish each type? Found that one single image needed replaced to target Competitives – and that gave them a lift of $25m!

HUGE gain on Overstock website by simply changing one image near a search box and it increased results by 5% or $25 million! The image was excluding competitive types of people due to search functionality/bounce rates on a specific page.

Opportunity Cost

Every day that Overstock did not fill that hole, they were losing $70,000 a day!

Framework & Optimization Process

1. Know your profiles – Personas, Four Perspectives, or simply emotional and logical

2. Define the conversion goals – what action do you want your visitor to take and what is the success page going to be

3. Do the Creative – based on the first two perspectives

Heirarchy of Optimization (detailed blog post can be found here)

Functionality first – Does it work?

Accessibility second – Can everyone access it?

Usability third – Is it “user” friendly?

Intuitive fourth – Does it feel natural and doesn’t “make us think”?

Persuasive final – Do people really want and understand what their buying?

The higher levels take the most time and effort but give you the greatest return.

5 Formulas to Online Success

1. Product Images Tell A Story

2. Test Your Headlines & Copy

-Test fractions vs Percentages

-Test Asking Questions

-Self Focused vs Customer Focused words (I am, we are vs you can)

-Switch Paragraphs (ordering of copy)

-Different words/meanings

-Informal vs formal (Attitude)

-Font sizes

-Wording on call to action

-Wording on images

3. Forms & Point of Action Assurances

-We value your privacy (next to action button)

-Guaranteed Response time

-Return information

-Reviews next to add to cart button

4. Calls to Action – Get Them To Click

-8 variables for cart buttons (wording, shape, size, style, icon, color, legibility, location)

-Emotional vs Logical (learn more vs help me choose)

5. Don’t Make Them Wait

-Fast load speeds

-Optimize image file sizes (58k vs 7k)

Golden rule: he who has the gold, rules. Customers have the pocket books so they rule. Also, another golden rule to online marketing is not like the biblical or religious adage of do unto others as you would have them do unto you…but the golden rule on the web is “Do unto others as they would have done unto themselves.”


Q: What if you have a small amount of traffic but you don’t want to wait 6 months for results?

Bryan: Focus on the top half of the hierarchy.

Q: Ecommerce merchants who have long funnels – where do you begin with a multi-step checkout process?

Bryan: First, worry about the point of action assurances. Other things to do would be to add progress indicators and try to reduce the multi-step checkout process as soon as you can. Can you combine one or two pages?

Q: Do you begin at the beginning or the end of the checkout process?

Bryan: It depends on your analytics – target the biggest bottleneck and work from there.

Q: What are some things that you can do after you feel like you have done the basics like Headlines, copy, etc?

Bryan: Start dealing with persuasive and intuitive things like images, words in hyperlinks, words in point of action assurances, and more. Work up the hierarchy if you already have the basics down.

Q: What goes through your head during the first 30 seconds of reviewing a website?

Bryan: If I were a customer and I were spontaneous/methodical/humanistic/competitive, what would I get out of this page? Try to see what questions I have while reviewing the page as a customer.


-Controlled content experimentation is critical to maximizing your conversion rates

-Google Website Optimizer is a free tool that lets you do unlimited tests on all your traffic on any page

-Personas and scents can dramatically increase the likelihood of improving your conversions

-Almost everything can and should be tested

-More resources at

Bookmark or Share It Here:

If you enjoyed this post, make sure you subscribe to my RSS feed!

7 Responses to “Live Blogging A Webinar: Website Optimizer – What Should I Test?”

  1. Bryan Eisenberg (2 comments.) says:

    Let me know if you want me to fill in the slides you missed. Sorry Webex had lots of technical issues.

  2. jameszol says:

    I thought it was a Webex issue. :)

    Yes, please fill us in on the slides that I missed! Thank you!

  3. Dr. Pete (11 comments.) says:

    Funny, I was wondering the same thing: was that slide flipping Webex, or is Bryan a madman? :)

    Other than that, a lot of top-notch content in the presentation, and yet more reasons why I need to dig deeper into Google Website Optimizer.

  4. jameszol says:

    @Pete: I agree – I enjoyed Bryan’s speed because he packed a TON of excellent actionable advice and education into a tiny amount of time. I thought he was a ‘madman’ at first because the last few webinars I attended were so dry and boring that his enthusiasm caught me off guard!

  5. Bryan Eisenberg (2 comments.) says:

    Well actually, I am a mad man :-) But my idea of driving a few thousand people crazy is not flipping between slides at a schizophrenic pace. Even though I should probably try it some time at a live event so I can see people’s eyes bug out.

    I might need one of those disclaimers in front of the presentation like the ones video games have about how this presentation may cause seizures. The good news is Webex Support was working diligently to try and figure out what happened and the archived version will not have any of these issues thanks to the great folks at Google responsible for it.

  6. jameszol says:

    @Bryan: A dance beat and strobe lights that matched the schizo pace surrounding the slides would normalize the “effect” without having to fix the software…but you’re right, you would need a disclaimer in front of the presentation…haha :)

  7. links for 2008-04-02 » says:

    [...] Live Blogging A Webinar: Website Optimizer – What Should I Test? | semvironment Nice write-up on the website optimizer webinar. (tags: websiteoptimizer, web+analytics) [...]

Leave a Reply

About Semvironment

Learn more about Semvironment, one of the world’s top pay-per-click marketing management companies.

Press Center

Feature articles and noteworthy instances of Semvironment in the news.

Hall of Fame

Read our most talked about news stories and PPC management training guides.