Emerging TechnologyAIHow machine learning can help you optimize your website’s UX

How machine learning can help you optimize your website's UX

For ecommerce businesses, your website’s customer journey is critical to your success. So how can you make sure it is optimized for maximum conversion?

For ecommerce businesses, your website’s customer journey is critical to your success.

According to one study, humans now have a shorter attention span than goldfish. Which means if your website’s UX isn’t clear, smooth and simple, customers simply won’t stick around.

In fact, 38% of people will stop engaging with a website if the content or layout is unattractive.

Source: Microsoft

So what can businesses do to make sure they achieve maximum revenue from website visitors?

Thus far, this series has covered the first two stages of conversion rate optimization (CRO). The first part of this series explained how and from where to collect data – using technologies like website analytics, heatmap tools and session replays.

The second showed how to use the data to generate a testable hypothesis, as well as tips and user experience (UX) red flags.

This article will look at best practices for testing and optimization, and explore how machine learning can improve efficiency throughout the process.

Content produced in association with SessionCam.

Testing and optimization

Now that you’ve done your research, collected data, analyzed it and identified areas for improvement, it’s time to get testing.

There are two ways you can approach testing. You can either:

  • Make changes speculatively and hope it improves conversion, or
  • Try and fix a known problem on your site.

If in doubt, start with the latter. Avoid testing whatever UX trick is doing the rounds on the internet. Don’t waste time building something that might not work; fix problems you know are there first.

Optimization is a gradual process. Each fix should have a measurable impact on your conversion rates.

How can you test?

It’s also possible to test more the one version simultaneously, by showing 33% of visitors the original, 33% version A and 33% version B.

This is called multivariate testing – useful in early stage evaluations where you want to identify the strongest approach before optimizing in more granular detail.

Tip #1: For big changes, start small.

If possible, start by rolling out changes to a small percentage of visitors and measure the results. Define which metrics or behaviors you’re looking to measure before you set out so you understand what success looks like.

For example, if you see that visitors are not scrolling far enough down a product page, you can test new versions, one which makes it clearer that there is information at the bottom of the page and another which brings it further up the page.

In this instance, an increase in the number of link clicks, time spent on the page or distance scrolled would constitute ‘success’.

Tip #2: Only test one variable at a time during an A/B or multivariate test.

If you change two or more things simultaneously, there’s no way to be certain which had an effect on the outcome.

The exception to this is for large-scale changes such as a site redesign where testing each individually detail becomes inefficient. A better approach in this case is to create three or four completely different designs to establish which path to focus on for further optimization.

Make sure to collect data for at least two or three weeks before selecting a winner – don’t be tempted to jump to conclusions based on a few days of improvement. It’s impossible to control for every variable, so time is the best way to reduce the chance of anomalies.

The best metrics to measure will depend on your specific business goals, but conversion metrics (number of registrations, number of purchases) and engagement metrics (time on page, pages per session) are a good place to start.

Tip #3: Test everything.

If you’ve fixed any glaring issues and have the resources, try testing other page elements to see if they have an effect. Images, image placement, headline copy, call to action placement, font size.

The sky’s the limit. Just make sure you’re not ignoring any actual bottlenecks in the visitor journey.

Machine learning and CRO

This series has covered the key steps to optimize your conversion rates. However, the process remains complex and time-consuming for marketers. Achieving positive results requires a significant amount of manual adjustment and implementation.

That’s set to change: innovative marketing technologies already use machine learning to make the optimization process more efficient – from data collection to analysis and testing.

analytics

Reviewing qualitative data such as session replays is time-consuming. However, a machine learning algorithm removes the need to manually trawl through data to discover why your website visitors are becoming frustrated. It can show you which errors are leading to drop off and uncover the most valuable routes to conversion.

It’s an affordable, efficient and effective way to identify your most costly website issues and deploy changes that quickly improve UX.

For more advice on how to optimize your website for conversion, register for our next webinar with SessionCam entitled How to Convert Your Website Visitors into Customers.

Related Articles

How artificial intelligence is changing email marketing

AI How artificial intelligence is changing email marketing

2m Al Roberts
Five ways fashion brands are using AI for personalization

AI Five ways fashion brands are using AI for personalization

3m Clark Boyd
Robots vs. humans: Will AI bring the advertising apocalypse?

AI Robots vs. humans: Will AI bring the advertising apocalypse?

6m Daniel Surmacz
The shock of the new: Artificial Intelligence and the future of digital marketing

AI The shock of the new: Artificial Intelligence and the future of digital marketing

7m Clark Boyd