Author note: the article below was originally published on the Richmond Times Dispatch. As one of my most shared articles to date, I’m reposting it here on the WebStrategies blog.

Have you ever gone through an expensive web redesign only to see lackluster improvement? You were convinced a cleaner, more contemporary design would bring your website into the modern era, while showering your call center with a steady flow of leads.

But that never happened…

Instead, you were left with a hefty bill and results that come far from justifying the cost.

In a time when we’re showered with data about our customers, businesses are still leaning on personal opinions to build their most important online asset: their website. Unfortunately for the business owner, those opinions don’t always get it right.

It’s time to consider an alternative way of taking your website to the next level. Instead of just replacing your website, you should evolve it; one piece at a time. In digital marketing lingo, this is called conversion rate optimization, or CRO.

What is it?

Conversion rate optimization is exactly what it sounds like. It’s the process of improving the rate at which website visitors are converted into customers.

Instead of growing sales by growing website traffic, CRO is about doing more with the traffic you already have. It’s about improving efficiency and effectiveness, without ever spending another penny on website traffic.

Sound appealing?

How it’s done

Maximizing your site’s conversion rate is part art, part science (but mostly science). The basic steps involve the following:

1) identify a specific area you want to improve

2) develop an alternative you think will perform better

3) run an a/b test of the old version versus the new version

4) analyze results

5) repeat

1) Identify a specific area you want to improve

The key to accomplishing any goal is to break it down into manageable stages. Instead of saying “lose weight,” say “eat a smaller lunch.” Instead of saying, “become smarter,” say, “read a new book each month.” Instead of saying, “improve conversion rate,” say, “improve my contact form.”

If you want to improve your website conversion rate, start by identifying one critical area that leads a user to conversion. This could be a key page, a contact form, or a promotional offer. Whatever it is, make sure it’s specific.

2) Develop an alternative you think will perform better

Once you’ve identified the one area you want to improve, it’s time to figure out how it could be better. Different colors, form fields, layouts, and graphics could all result in better performance. You could go deeper by changing the messaging or incentive based on customer persona analysis. Use the input of your co-workers, web designer, and mother-in-law to come up with a solution you’re confident will blow the old version away.

(If we weren’t truly sincere about improving conversion rate, we would stop right here. But we want to know definitively that our new version is objectively better. Now it’s time to test.)

3) Run an A/B test

A/B testing splits your website visitors into two groups. One group sees the original version of your site (version A) while showing the other group the new version (version B). We then see which group was better at completing your website’s goal.

Most tools that run A/B tests (Google Analytics and Optimizely are two popular options) will manage all the complicated math to make sure your sample size is large enough and that a winning version can be confidently declared.

4) Analyze results

Because version A and version B only had one small difference (the one thing we identified in step one and changed in step two), we can attribute any difference in behavior to that one modification.

Not every experiment will confirm your awesome new design outperforms what you had before. Some experiments will show improvement, but only marginal improvement. Some will blow the results out of the water. Some will crush your ego into one-thousand pieces (don’t take it personally).

5) Repeat

See a huge improvement? Great, move on to another test. Didn’t get the result you want? Try again. CRO is a game of incremental improvements. Sometimes that means 3 steps forward, other times it means two steps back.

Each experiment is a chance to learn a little more about your customers preferences. Each experiment can settle an internal design debate. Each experiment gets your website one step closer to optimal performance.

Now let’s look at these five steps in a real-world example.

We want to improve the response on the “Pricing and Options” page of your website. Our web analytics tells us it’s one of the most popular pages on the entire site. On average, 5.0% of visitors who see this page ultimately fill out the contact form.

We want to increase that number.

Our research suggests the “Pricing and Options” page is lacking some key language, so we create a new version of the “Pricing and Options” page and a/b test it alongside the original version.

After running the experiment for a few weeks, we see a conversion rate of 4.3% on the original page and 7.6% on the new page. That’s an improvement of 77%!

To put that into more practical terms, if 1000 people saw the “Pricing and Options” page, the new version would mean 33 more leads than the original. That’s a huge win.

This example demonstrates the importance of split-testing two versions as opposed to just swapping out one version with the other.

Our original page averaged a 5% conversion rate, but during the few weeks this experiment ran, the original page’s conversion rate just happened to perform below average (4.3%).

If we compared the 7.6% conversion rate of the new page with the historical 5.0%, we would walk away thinking we only improved performance by 16 additional leads for every thousand visitors. That’s still a nice improvement, but not nearly as impressive as what our split-test showed (a 33 lead improvement).

Running the two pages side-by-side gives us a much more accurate look at how the two pages compare and gives us the confidence we are improving our site, once piece at a time.

If you want to know the true impact of your website changes, never compare it to past performance.

Always test. Always evolve.