You sit on the noisy paper covering the exam table and just as you begin to describe your symptoms, the doctor interrupts you.

He tells you exactly what you need: 50 mg of propylthiouracil and artificial tears, twice per day.

He leaves the room.

Or drops the mic.

The doctor made no attempt to understand the symptoms and causes of your illness. He skipped that whole diagnostic process, the part where he attempts to narrow down and identify the cause.

He winged it. Hopefully, you would be mildly alarmed (and mildly entertained).

Marketers take a similar approach to landing page optimization. We provide answers before asking questions, or propose landing page recommendations and cite best practices with as little information as, “improve this page’s conversion rate,” without looking at data or speaking to customers.

The cost of this approach is noteworthy. Marketers mistakenly optimize for clicks instead of revenue, miss the opportunity to provide a cycle of insights that fuel new hypotheses, and treat all experiments as if they’re equal (they’re not).

Successful landing page optimization, then, requires building a system. One that works. And that starts by asking questions.

Diagnostic questions provide you with a systematic way of identifying and prioritizing landing page optimization opportunities, and often reveal insights applicable to a range of landing pages instead of just one or two.

Here are four diagnostic questions that I use. Treat them not as prescriptive rules, but flexible suggestions that you can modify and use for inspiration.

1. Which Ten Pages Miss The Most Conversions?

Find the pages that have the largest potential to impact your goal of increasing leads, sales, or average order value. This group of pages will usually requires the least effort to generate the largest return and will likely be the source of your first ten tests.

Generally, the two metrics to focus on are conversion rate and unique visitors. Together, they tell you which pages have the highest count of non-converters:

# of visitors * (1 - conversion rate)

Consider the table below.

Landing Page 1 has neither the lowest conversion nor the greatest amount of monthly visitors, yet it is misses the most conversions, or bleeds the most value. This means that page with the lowest conversion rate or the most traffic is not, by default, the best place to begin your a/b or MVT test. Without additional context, Landing Page 1 is one of the first pages we should optimize.

This means that low traffic, low conversion rate pages are not top candidates for testing. If they generate a large amount of value (e.g. every visitor’s basket is size is 10x the average), then it’s worth investigating the source of this traffic and their personas so you can reach similar visitors.

Last, bounce rate and time on page are typically not indicators of high-leverage opportunities. Reducing bounce or exit rate by 10% does not translate to a 10% increase in conversion rate.

2. What is the Goal of the Page?

Each page of your website should have a clear goal. This is especially true for landing pages.

A landing page goal should answer the question: What action do you want the visitor to take next?

Before you begin to evaluate elements for testing – headings, form labels, call-to-action copy, etc. – define the goal of the page and the metric you will use to measure this goal. Write it out. Document it.

And share it with your team.

The goal of the page should inform the structure, style, and content of the page. A goal must be an action.

For example:

  • Goal: Persuade the user to sign up for the newsletter

    • Metric: successful newsletter signup form submissions

  • Goal: Encourage the user to explore a relevant use case

    • Metric: use case link clicks

  • Goal: Motivate user to buy product X

    • Metric: completed transactions

For example, the goal of Optimizely’s Free Trial page is to encourage non-members to try Optimizely’s tool. The success metric could be a successful form submission.

A better success metric would be a qualified form submission, wherein the visitor used authentic, identifiable information.

3. Does the Content Support the Goal of the Page?

The amount of elements and element interactions you could test on a landing page is endless, even overwhelming. Occasionally, it can seem that starting in one place is a good as the next; but, we can do better.

Instead of looking for what to add to the page, or what’s missing, begin with what can be removed.

Refer to the goal of the page and evaluate whether your content supports that goal:

  • Does this sentence support the goal of the page?

  • Does this navigational element support the goal of the page?

  • Does this form field support the goal of the page?

If an item does not support the goal of the page, it does not belong on the page.

Consider, for example, what Optimizely’s landing page could have tried to achieve had it lacked focus:

  • Introduce the Enterprise plan’s features and benefits

  • Resolve customer fears about a/b testing costs

  • Demonstrate success companies have had using Optimizely

Those are all ‘nice-to-have’ additions, maybe, but (without more feedback) they would only distract users from the goal of the landing page: to try Optimizely. The amount of information you need to provide to best help users accomplish the page goal typically starts with your best guess, then evolves through testing, using web analytics (site search, questions in comments section), and conducting usability studies with customers.

Compare Optimizely’s page to VWO’s Free Trial page, which does include distracting and extraneous information.

Do elements such as 50ms average response time, 99.99% uptime, or case studies help users to register for a trial?

Probably not, I would argue. They do the opposite. The text distracts and the case studies divert users to the blog.

A tactfully designed landing page is simple and prevents the user from having to wrestle with choices they otherwise would not know exist.

4. How Quickly Does Your Landing Page Load Compared to the Site Average?

Page speed, or the amount of time it takes your browser to load and render a page, has a direct effect on conversion and bounce rates. It’s also a factor in determining a page’s position on the search engine results page (SERP).

Walmart’s research, for example, reveals that the average site load time for converters (3.22) seconds was approximately half the time of non-converters (6.03 seconds).

According to KissMetrics, a 1 second delay in page speed can result in a 7% decrease in conversion rate.

Aside from abandonment, a laggy experience leads to user frustration, fewer returning users and – maybe someday, a rebellion against Time Warner Cable.

To determine your average page speed, you can use the Page Timings report in Google Analytics. Compare the load time of the top ten landing pages you defined in step one to your site average.

Once you have an idea of whether your landing page is above or below your site average, the next step is to consider how your page speed compares to websites other than your own, and what you can do to improve it using Google’s PageSpeed Test.

For more robust feedback, use Pingdom’s Website Speed Test, which will provide you with a list of the specific issues causing delays in your load time that you can then resolve yourself or pass along to your developer.

Often, the fixes to improve page speed may be quick and simple - such as compressing media; minifying HTML, CSS, and JS; or shortening a redirect chain.

So there you have it. I hope these questions help you to build a collaborative culture of testing where data trumps intuition (most of the time).

About The Author:

Vincent Barr is curious. He performs online marketing, web analytics, and split testing at MongoDB. He does not write about much of anything at