Hunter Boyle from Marketing Experiments gave an interesting talk at the SIPA Miami conference on ways to fix leaky sales pages.

He reminded marketers to put themselves in the mind of the visitor, who is asking three basic questions when he comes to your site or page.

• Where am I?
• What can I do here? and
• Why should I do it?

Hunter’s presentation was on “high-impact” changes to sales pages, so he pointed to some of the elements that are most likely to have a dramatic change in response.

• Headlines
• Testimonials
• Forms
• Navigation steps
• Price
• Product images

Some web pages have “related articles” links on the side, right? Well my mind is like one big “related articles” link while I’m reading or listening to a lecture. While Hunter was talking about these changes, I started thinking about some of the issues related to the examples he was sharing.

Hunter mentioned a marketing effort from a client (I believe it was an email) that mentioned “eight things” you could get. Unfortunately the landing page didn’t have the eight things.

So imagine an email campaign that promises eight things and sends people to an A-B landing page test. The test is set up to measure which page is more effective at driving traffic to the registration page.

For these purposes, let’s say the user’s action sequence would go like this.

1. Open email
2. Click on link to go to A-B landing page test
3. Click on “buy now” button to go to registration page
4. Register and finalize sale

The test measures how many people get to the registration page, but not how many people order. (Google’s website optimizer only allows one goal page, which is something they ought to fix. There should be primary and secondary goals. And maybe even tertiary.)

In this case, the email promises “eight things” of some sort. It’s an effective group of eight things, so it does what it’s supposed to do — drive people to the landing page.

But there are two landing pages. The A version has (or mentions) the “eight things,” and the B version doesn’t. Or maybe the B version mentions them, but they’re not clear enough to the visitor.

The users who go to the A page evaluate the offer, including the eight things, and some of them click through.

The users who go to the B page wonder where the eight things are and suppose they might be on the next page, so they click through — not because they’ve evaluated the offer, but because they’re confused. They get to the registration page, which also doesn’t have the eight things, so they bail.

What has happened? The B version — without the eight things — has won the A-B test, but it gets fewer orders. A marketer might look at his A-B test results and say, “Wow, this page increased my CTR by 75 percent. What a winner!”

Of course it’s not a winner at all. It only increased throughput in one part of the process, for all the wrong reasons.

There are two solutions to this kind of problem.

The first is to set up the experiment so that the “goal” page is the “thank you” page after a successful registration. That way you’re measuring which landing page got more orders, and not merely which landing page got more clicks to the next step.

This is not a good solution. It solves the problem of “which landing page gets me more orders,” but it makes it harder to optimize each piece of the funnel — mostly because it takes a lot longer to run the test.

Marketers should optimize each part of the process.

• Optimize the subject line of the email to get more opens.
• Optimize the email creative to get more clicks to the landing page.
• Optimize the landing page to get more people to the registration page.
• Optimize the registration page to get more completions.

Which brings us to the second way to fix the problem. Do the A-B test on the landing page only — with the landing page as the test page and the registration page as the goal page — but also set up a funnel in your analytics program so you can monitor every step of the process.

Then if the B page wins the A-B test but kills registrations, you’ll be able to see that in your funnel analysis, and you’ll still be able to run a quick, targeted test on your landing page.