Getting someone to click on your search ad is only half the battle. Once visitors arrive, the landing pages you display to them are a crucial component in converting them into buyers. This session looks at ways to test and tweak your landing pages to get that conversion. NOTE: The session is designed for those who are already familiar with how paid placement works. If you are new, be sure to have attended Search Advertising 101 on Day 1.
Moderator: * Allan Dick, General Manager, Vintage Tub & Bath
Speakers: * Scott Miller, CEO, Vertster * Tim Ash, President, Site Tuners * Jamie Roche, President and CEO, Offermatica * Tom Leung, Product Manager, Google Website Optimizer
Tim Ash - Conversion tuning crash course - ½ day workshop condensed into 20 minutes. - What is tuning? -- Follow the 3 main online marketing activities --- Acquisition --- Conversion Retention -- Increase conversion through tuning - Many companies drop the “part in the middle” -- Don’t neglect the landing pages - Economics of Conversion -- CPR = CPC/CR -- Fix your site to lower costs -- CPA is going up if your conversion rate does not improve. Who should design your site? - Agency, marketing dept, IT, boss or nephew? - The right answer is the customer!! - The only thing that matters is the process the users are going through. People with wallets win! What can you tune? - Landing pages that lead to trackable actions. - Price of product or service - Elements -- Headlines -- Layouts -- Navigation -- Color Scheme -- Form Layout -- Button text -- Sales copy -- Graphics -- Calls to action - There are no “Universal Truths” - Do not copy your competition - test your customers instead. Types of testing - A-B Split -- Test one at a time, send equal traffic to each. -- Easy to track and implement -- Typical Test Size: 1-10 - Multivariate -- Tests several variables at once -- Compress your data into one test across all the variables. -- Typical Test Size: 100 - Proprietary Tuning -- Infinite Mistakes to Avoid Do not ignore the baseline - Compare to a baseline - not just week to week -- Beware that things could get worse and it could be hidden in a week to week analysis -- Always measure relative to the baseline Not collecting enough data - Collect more data to increase confidence in the numbers - Get enough data! #3 Ignoring Variable Interaction - It’s not the picture or the headline - it’s the interaction between the variables Google sees strong effects between elements.
Get the combination of elements that gives you the best conversion rate.
Testing Theme - Less is more -- Remove the clutter - leave the persuasion and trustmarks. -- Shorten the forms - remove unnecessary elements.
Tom Leung Google agrees with everything Tim said.
Why is testing important? - More than just usability and conversion - It’s about users Economics are tough - you spend a lot of money to Google, SEO’s and your visitors still leave! The goal is to get you more green. More non-bouncers.
Tested the Picasa homepage and found a 30% increase in download rate. Tested 200 versions of the page using Google Website Optimizer.
Technology is only a tool - plan and manage the process! Know which pages to test and why, what variations you want to test. Which parts of the process - form, download, click, purchase. Know when the test is complete and know what you’re doing next. Options - Do it yourself - Hire an expert - Talk to design agency or SEM
Instantly create variations by adding them to the tool. The Google dashboard will show you the chance each test has to beat the baseline. The best combination is often not the full combo page, but a mixture of elements. The only way to know is to run a test.
6 Tests that you can run - A/B - Multivariate - - Split path - Multi-page multivariate - Time-based - Do Anything
Scott Miller You can use their software to test yourself or let them run the tests.
Tested Hacker Safe logo - Simple A/B test site wide - Objective -- Conversion -- Revenue Per Visitor (RPV) - 14.8% conversion lift - 15% lift in RPV Takeaways - Balance risk reversal, value and scarcity. - The only way to get there is testing
Test 2: Getting Attention Elements which can influence bounce rate - Logo - Headline - Tagline - Imaging - Audio/Video Ran an L16 Algorithm Taguchi test Ran against all PPC traffic Setup - Hero Shot - Call to action - Button color - Low Banner - Guarantee
Of the 16 variations, the Hero Shot, Call to action Header and Low Banner were the most important.
Anything underneath an image naturally gets attention. We look underneath to find out what it’s all about.
Test Results - 40% lift on lead generation through 3 rounds of testing Takeaways - Focus above the fold - Conversions ,mirrors eye tracking study - Blink of an eye mentality - Before you can convert, you must get attention
Jamie Roche - Offermatica
“Personalization and the value it brings” Normal people can use the tools.
As different people require different experiences, we are moving toward personalization.
Is the best page the best answer? No, different people require different experience to bounce less.
Reduce bounce by increasing relevance. Increase relevance by user type and bounce will drop.
Relevance is the key to targeting customer categories or customer stage. Make it relevant through personalization.
Showing content of interest has improved RPV and conversions.
Myth - Personalization is very difficult.
Dynamic insertion of search keyword on landing pages has shown increased conversion.
The most important thing to do when you start personalization is to group your customers into big buckets.
Grouping your audience - Behavior - new vs. returning, category - Time-Based - time/day/season - Source - Paid, organic, email - Environment - geo, IP, resolution - Registered Customer - CRM, analytics, demographics.
Steps to Personalization - Think about where you are going to start -- Focus on wrong areas. -- Focus on areas where you cannot make the change. -- Upper medium landing pages from paid are a good place to start. - What can you remove when you test personalization? -- Look at the site and remove elements - registration solicitation - Start with the big stuff -- More fun to see big results -- Once you start showing success, attention will be drawn and budgets will be opened. - Two types of optimization -- Evolutionary - it always seems to start this way. They are less likely to produce results. -- Revolutionary - Eventually have to do something drastic. It mail fail but you’ll learn. --- Make this test as extreme as you can! Stretch it out. Barriers - Change is blocked by -- Brand people stop the change and they fight you. -- IT gets threatened and wants to keep it to themselves. -- Standard operating procedures. - Overcome Obstacles -- Trojan horse - simple change. Go around IT. Host the page -- Fight for branding changes - get 5% of the traffic and win them over. - Benefits to working through barriers -- Brand can evolve as they see results. -- IT is happen not to serve the whims of Marketing Do Right Now - Do a keyword repeater - Get into category affinity - even if you have just 2 groups. Create versions based on interest. - Change story for repeat visitors - if they’ve seen it change it.
The key to performance is relevance. The key to relevance is showing the right people the right things.
Q and A Q - When does your test become your baseline? A - Tim - When you have enough data to support your results. The answer changes over time.
Q - Am I safe to consistently use testing tools in a natural search experiment? A - Tom - Google published a help topic with guidelines for testing. What you show Google bot should honor the spirit of what you’re testing. Don’t show babies and sell online poker. If you are testing clean then you will be fine. It is in all engines best interests to improve user experience so please test. A2 - Tim - Balance between stripped down page and crawler friendly pages.
Q - How do you set up a test and maintain the user experience page to page? A - Scott - Has the ability to define testable objects within a template and they can be maintained throughout the user experience. A2 - Jamie - Optimize the experience and not just the page.
Q - When we run multivariate test how do we know which variable worked? A -Tim - Run a confirmation test when you have the results to confirm the results.
From Steve Krull