How to optimise a low traffic website, Posted by Blog, Conversion rate optimisation, Steve's posts in
Steve Keightley, Head of Optimisation & Personalisation at Mezzo Labs, explains how websites with low visitor counts can still optimise their user journeys by using some simple methods.
At Mezzo Labs, B2B clients often ask us to improve their conversion rate through AB testing. However, they don’t have tens of thousands of visitors per month to experiment on, so each test could possibly take months to reach a conclusion.
Whilst traffic is a key factor in CRO, there are ways to optimise the customer experience on a low volume site. Here’s how…
1. Ask some basic questions
Firstly, before we worry about the practicalities of traffic volumes, let’s go back to basics, by asking a few clarifying questions.
You’ll have heard the questions many times before but they are still important to answer before you launch into an optimisation programme.
- What are the key objectives of the site? What are the key business challenges you are trying to solve through it?
- How do you measure those objectives?
- Who is the target audience?
- What are the main points of visitor frustration? Where are the main drop-off points?
I did warn you. These are classic questions, but it’s amazing how often the answers are different depending on who you ask.
2. Who is the website for anyway?
I would argue that these questions are especially important for a B2B website. With a B2C (Business-to-Consumer) website, the visitor is usually doing the buying.
With B2B, the visitor may not be a budget holder or decision maker. They may just influence the sale. The website needs to address the needs of each visitor type, so it’s really important to understand what those needs are.
3. Do your sums
Let’s assume that you’ve answered all those key questions and honed in on a hypothesis to test… Before you leap into the creative bit, finding your challenger design, you’ll need to do some due diligence.
You need to find out what sample size you’ll need in order to draw a valid conclusion to the experiment – to understand how long it will take to get a valid result.
You can do this using a sample size calculator (most AB testing software providers have one).
If you find your traffic volumes will be too low in reasonable timescales, you have a few options:
- Focus on “impactful” changes. By that, I mean when a challenger design is only subtly different from default it’s hard to see any difference in the results – that’s my general experience anyway – even on high traffic sites. Testing drastically different challenger versions is likely to have a big impact on the conversion rate. Therefore it’s possible to reach a statistically valid conclusion with a small volume of traffic.
- Test micro conversions. Micro conversions usually happen more frequently than macro conversions, therefore it’s easier to get a big enough sample size. Test on the click-through to the next page rather than purchase complete, for example. The downside is that these conversions won’t always shift the dials on Key Performance Indicators.
- Consider reducing the statistical significance. Results can be achieved with a smaller sample size. However beware, there will be a higher risk of false positives (a winning variable that won due to chance).
4. Blend quant with qual
If you have a heatmapping session-playback tool, such as Hotjar or SessionCam, you can validate a low-volume winning result by reviewing what actually happened on the page.
I have used AB testing tools which integrate with these in-page analytics tools so that I can replay each successful variant to assure myself that they were truly successful.
5. It’s not just about UX improvements
Finally, optimisation isn’t just about improving the web design to get visitors to complete the desired action. It can also be used to experiment with pricing and propositions and gain insight on the interactions with them .
It is possible to A/B test on a low traffic website. It just takes some thorough research up front, good planning and the courage to test bold UX changes. Finally, by using a blend of quantitative and qualitative analysis you can gain insights which will help to inform improvements to the customer experience.