Conversion rate optimization, better known as CRO, is the practice of improving the results of each element of your website (or web page). CRO, however, only works when you can measure it. Make multiple changes at once, and you won’t get clear results. Some changes could be wildly successful while others might flop, making your net CRO result a big zero.
When making website adjustments, test one change at a time. In our experience, this is one of those “Don’t Pass ‘Go.’ Don’t Collect $200” type of rules.
Even if your changes do lead to improved CRO, with multiple adjustments, you won’t be able to pinpoint which ones were the cause. Was it all of them? Was it one?
Accurate CRO measurements are essential if you want the most effective changes and more in-depth insight into what resonates and works with your audience.
To eliminate confusion and provide clarity on the impact of any change, we use the following list of 6 Dos and Don’ts for every CRO experimentation. It provides the clearest, best, and most profitable results for every website change.
#1. DON’T Test Multiple Elements at Once
Think of yourself as a scientist. Proper scientific research uses a control group and a test group. For a website change, your control is your current site. The test is whatever element you’re going to change. As marketers, we call this the A/B split test.
Take it one step at a time. Keep it simple. And make sure you have well-defined goals for the experiment and a process to implement, measure, and analyze the outcomes of each test.
#2. DO Get Specific
While this “DO” follows from #1, it gets more specific. Your goal needs to be much more than merely desiring more conversions or wanting to improve your website. You need to determine what SPECIFICALLY you wish to improve and, taking it one step further, specifically what that looks like!
When planning your test, you might find yourself specifying:
- What color will the button be?
- What happens if we add copy? Remove copy?
- Does this headline work better?
- Is this structure (specifically detailed) or that one (likewise detailed) the best for this product page?
Every detail must be absolutely specific if you want to measure the impact of a change. Sometimes, and this might seem odd, the smallest change, such as button color can have a wildly massive impact on the results. For example, once for a client, we changed a button from gray to orange. This little change resulted in an increase of 1000 sessions with more pages per session.
#3. DO Think About the Results You Want
You want to test like a scientist but think like a marketer. This means you have a specific outcome in mind. If it’s a sales landing page, you want the outcome to be more sales. The scientist aspect of it involves not caring whether it’s the gray button or orange button that produces the most significant results. The marketer part is testing to produce the results that move you closer to your goal.
(Scientists don’t have an expected outcome in mind when experimenting. Having the desired outcome is called ‘Research Bias.’ Marketers, on the other hand, better have an outcome in mind but be open to the facts that present themselves on the way, like preferring the gaudy orange button to the classic and refined gray button you may like.)
Thinking about results often requires understanding your goal in the form of a metric. For example:
- Do you want to improve engagement, meaning you want your pages per session to increase?
- Do you want more people contacting you through your website, meaning you want more contact forms submitted or calls made to your call center?
- Do you want a lower bounce rate, which means people stay on the pages they click to?
These are, of course, only a few examples of the myriad of site metrics you might want to improve. The point is, you DO want to identify each goal in terms of a metric and then make process-driven changes that would impact that metric.
#4. DO Test Your CTA Button
By now, you probably saw this one coming. We can’t emphasize enough how important buttons can be on a page. We’ve mentioned button, but there are so many other factors we’ve discovered can affect the impact of a button and should be considered, such as:
- Where is it placed on the page? What changes if it’s put on the left? Center? Right of the page?
- Is it hidden by its placement?
- Which button shape is best? Rectangle? Circle? Square?
- Does the button copy clearly state what to do and expect?
- How does changing the button text size change results?
Every facet of the button can be tested. To give you a head start on testing your button, we have found brighter button colors work better than subdued or transparent buttons.
#5. DON’T Get Off Brand
Being creative means getting outside the proverbial box and exploring all possibilities. While this is good for brainstorming sessions, we find it’s best not to get too far off-brand when testing and presenting to your audience. They like you for who you are. Experiment, by all means, but don’t get off-brand.
Revisiting your CTA button, you should follow, to some degree, the color scheme of the website. It may be that a dynamic contrasting color will work, but I once tried a blue button on a red background on a personal website and found that it was not pleasant to look at, at all!
Going beyond the button rule applies to everything from messaging, voice, tone, site structure, the information you provide, CTA, and more. Stay within your brand style guide.
#6. DON’T Give Up Too Early
A/B CRO experiments need to run. A lot depends on volume as a test with lots of activity could run two weeks while one for normal daily traffic may benefit from running two to three months. Ideally, you want to collect as much data as possible.
Often the more data you can get from testing, the more dramatic the results. Maybe you’re trying to increase pages per session (the number of people on the website). Short-term results may only show a 5-10% improvement when running the test longer will push it higher toward our recommended goal of a 100% improvement.
How long you test also depends on how much you are testing, which is determined by your goal. If you’re trying to address multiple areas of the buyer journey, you may find you want to run multiple tests, although doing this comes with the challenge of reducing the size of your test group, which limits your data.
Testing Your Audience Tells You How to Drive Your CRO
If you want to change your website, we recommend you experiment with your CRO. While website design may be part art and creative, you want to drive that with scientifically derived data. You may find yourself surprised by what works, but in the end, you’ll be thrilled by the results!