There comes a point in time where every marketer encounters a roadblock. Despite your best CRO efforts, you can’t quite figure out why a certain page or landing page isn’t converting as much as you had hoped.
You ran a test for a full month, made sure it started and ended on the same day, made sure to test two variants, and results even showed an increase in conversions from 5% to 10%. I mean that’s enough evidence to submit it to your boss for approval right?
With the okay from your boss, you decided to implement the new page. And within the first month, you notice the conversions are down more than they were before. Maybe it wasn’t enough time? You decide to let it run for a few more months. Now you notice there’s a decrease in monthly revenue too. What’s going on?
You are passed the CRO basics and have already experimented a fair amount with CRO, so here’s a rundown on 12 commonly made CRO mistakes.
1. You rely on A/B testing as your only method of CRO
Naturally, A/B testing will be the first of many methods of conversion rate optimising you will experiment with.
A quick rundown on A/B testing — also known as split testing — is when you create and compare two versions of a web page or landing page to determine which one is performing better. The one that has a higher conversion will be the page you implement.
Based on statistical analysis, a business is able to test various changes such as a single headline or button, simple colour changes to a call-to-action or a complete redesign of the page to track which variation has a higher conversion rate.
Unfortunately, basing most of your CRO on simply doing A/B testing is a one-trick pony that could cost you. Pages with low traffic can be difficult to run tests on because of how long it takes to gather enough significant data. Making decisions without enough data or having low conversion rates could potentially see a decrease in conversions.
In these situations, there are three things a business can do:
- Split test two largely different variations such as completely different design and layout, headlines, call-to-actions, and more rather than restricting yourself to one variable at a time.
- Run multiple tests more frequently to gather more data
- Run tests for longer periods of time to obtain significant results (traffic fluctuates at different times of the day, week, month, and holiday periods etc.)
2. You don’t have tracking set up properly
A successful CRO process requires you to get the right data and lots of it. Sometimes the right information could be as simple as having the conversion numbers, and other times you need to dive deeper into the behaviour of users on site.
Many marketers, even great ones – have made this fatal mistake. Not tracking the right things.
Conversion numbers tell you one side of the story. But, here are some thoughts – where are your users coming from? Why do they stay on the page? Why do they leave the page? Why didn’t they buy? Why didn’t they click?
These are the sort of questions you need to be able to form a hypothesis on to understand why a change on a website or landing page, did or didn’t work to your favour.
3. You can only manage what you can measure.
When you set up a CRO experiment, make sure you are:
- Measuring the right conversions (enquiries, clicks to a page, purchases or other desired action)
- Ensuring conversion data is ACTUALLY recording (test this in analytics)
- Confirming there are no duplicate numbers (happens more often than you realise)
- Tracking user behaviour, not just the final conversion
Here are some important metrics you’ll want to be tracking:
- Bounce rate – how quick a visitor leaves a page as soon as they arrive
- Average time spent on page
- Call-to-action (CTAs) – views, clicks and submissions
- User path – provides insights on visitors behaviours
- Page load speed
4. You run tests at inopportune times of the year
Ever ran a successful campaign that yielded a high conversion rate one month and was a complete dud the next, and you’re sitting there with a blank face wondering why?
Remember when we told you about tracking data? Well the day, time, week, month and season can have a huge impact on the success of your campaign.
Seasonality is an important factor that should be considered in the early stages of A/B testing as it can indicate the quality of traffic during that period. Your audiences might be more likely to purchase your discounted product on Boxing Day but come mid-June they may feel less inclined to whip out their wallets.
5. The sample size for your test is inadequate
What did we talk about at the beginning of this article? CRO is a science. This means you need an adequate sample size to experiment with in order to achieve meaningful results.
Same goes for testing on pages with low traffic. Basing your tests on a small number of conversions and low traffic does not give a clear indication of how it would perform if pushed to scale.
Size Guide
The size of audience you need to test a change will vary depending on the type of business or website you’re optimising. The rules of engagement for an E-Commerce website with significant paid traffic, will vary compared to a lead-gen business site selling services to other businesses.
However, here’s one easy way to have a reference point for how much traffic you need before gaining direction on your test. Calculate the amount of traffic you need to generate 50 conversions at your current conversion rate.
For example, if your current conversion rate is 3%, then this number would be 1,667 visitors (50 / 0.03) = 1,667
This is the number of visitors you should drive to the page to run the conversion test and see if your change had a positive impact or not. The disclaimer to this is these visitors have to be qualified visitors, meaning that they’re the right type of people, in order to get a true result.
If you’re driving traffic to a new page on the site, and it’s a low traffic website generally speaking. I would recommend reducing this number to 10 conversions at the estimated conversion rate.
For example, you run a B2B lead gen website and your estimated conversion rate on the page is 5% for qualified traffic.
You would want at least 200 visitors on the page (10 / 0.05 = 200) in order to make a call on whether your conversion test was successful or not.
These are only guidelines to use for deciding on a sample size that’s appropriate for your test. You need to consider the type of test you’re running, potential size of impact and complexity of the implementation to decide on the right sample size.
6. You treat all traffic the same
Your marketing message needs to be so powerful that your ideal customers can’t resist it. To create a powerful marketing message you need to know who your customers are.
Distinguishing between different types of traffic allows you to form customer segments, which can help you differentiate and find patterns in the types of people you should be specifically targeting — your ideal customer.
With a clear idea of who your target market is and who isn’t, you are able to craft customised unique web experiences and really hone in on what makes your ideal customer convert.
For example, making sure your website is equally compatible for mobile phone users as it is for desktops. If it’s quick to load, easy to navigate and provides a clear message combined with an irresistible offer and a cheeky form to fill out and you’ve got yourself a conversion.
Remember all of this is not possible without empirical data from testing, tracking the right key metrics, taking in timing, and understanding what variations to implement and be pushed to scale.
7. You don’t really understand the metrics
CRO is heavily dependent on metrics.
If you don’t really understand why, when and what you’re testing, then you won’t be able to run proper tests to obtain significant results, which could seriously crush your campaign.
This may be one of the biggest reasons that everyone should speak to a conversion rate optimisation agency or independent professional, for sound advice. Getting clarity on how to identify, select and report on the right metrics can save you years of strife and a whole lot of cash in failed experiments.
8. You don’t experiment on pages or campaigns that are already doing well
Just because a page is doing well doesn’t mean it can’t be improved. In fact, the pages that are doing well have the highest potential to perform EVEN better if optimised.
Don’t focus all your marketing efforts and resources on optimising pages that need the most work. More often than not, you’ll find a fountain of hidden wealth on pages that are already performing well, and if optimised, could bring in much better results.
9. You forgot about website speed
How many times have you closed a page because it took too long to load? Well you’re not alone. Studies show that 47% of users expect a page to load in as little as two seconds or less and that 39% of users will leave a website if it takes too long to load.
Page load speed is a crucial component when optimising a page and can be a major reason why your website is experiencing high bounce rates, which could ultimately cause a decrease in conversions.
There are however, a number of ways to improve your website loading speed.
- Evaluate your website performance regularly using tools such as Google PageSpeed Insights, Moz reports, or ahrefs website audits
- Run stress tests and load speed tests to improve server response time
- Ensure load times are at optimal speeds across all platforms and devices
- Compress CSS, HTML and JavaScript files larger than 150 bytes
- Compress and optimise images and combine images into CSS sprites
10. You forgot about mobile-first
One of the biggest mistakes you could make right now is disregarding how mobile-friendly your website is. By 2021, mobile e-commerce sales are expected to dominate the online world and account for 54% (or $659 billion) of total e-commerce sales.
There is a large chunk of people out there who go ‘digital window shopping’ on their mobile phones. I take the train twice a day, 5 days a week and I’m 100% guilty of doing this. What’s alarming is that 77% of people, like me, make impulse purchases when doing this.
It’s important to make sure when testing a page, that you’re testing its compatibility across all devices including desktops, tablets and especially smartphones.
11. You give up after a “failed” test
One of the biggest CRO mistakes any marketer could make is giving up after a failed test.
Tests provide results. You either get the results you wanted or you learned something new. Just because it ‘failed’ and didn’t provide the result you wanted, you obtained valuable data.
Go back to your process with this data, tweak your hypothesis and test again, and again, and again. The fastest movers in the industry have tested and failed countless times to achieve success-driven results and it’s their greatest advantage.
Success is driven by results. Results come from testing. Testing reveals failures. You need to fail to get winning results that drive success. Test more, improve faster and get closer to building a winning campaign.
12. You base your CRO tests on tactics instead of research
When a landing page isn’t converting well what’s your first instinct? A change needs to be made. So we make some changes based on what we ‘feel’ or think will work better and implement it.
What happens next? Conversions go down even further.
This is where things get frustrating. Clients are now on your back asking why they aren’t getting any leads, and your only response is we’re working on it.
Quite often, using individual marketing tactics and guessing what will work better without referring to the research can be a slippery slope to countless man-hours wasted on writing 2, 3 maybe even 5 versions of a page and still no reward.
This is where you need to stop and rewind it right back to the beginning. Before you start again, put your ego aside and throw your assumptions out the window.
Remember what we mentioned earlier? CRO is a science. You need to take a systematic approach. Create a process and base your decisions on the empirical foundations of data and research.
The data never lies.
What does a winning CRO process look like?
We learned this the hard way. We went through countless experiments over the years and came to the conclusion we had to stick to a systematic method to see things clearly and actually improve pages and websites’ performance for the long run.
Here’s what a winning CRO process should look like.
- Start with research. If a page is not converting, analyse the data and find out where the drop-off points are in your sales funnel.
- Create a hypothesis. Explore more than one individual tactic and come up with a solution for overcoming these drop off points.
- Define your control group and establish how and what you will measure so you can collect data properly.
- Conduct the experiment. Give the experiment sufficient time to gather enough significant data.
- Analyse the data. Find out where the weak points are.
- Test again.
Now what?
You’re ready to implement this CRO process and eager to see positive performance results. Now the next thing to think about is: what are the type of tactics you can use to improve your CRO?
