By Matthäus Bognar, General Manager – EMEA & APAC, Nosto
Many leading online brands owe their success partly to their openness to experimenting. Amazon, for example, reportedly conducts more than 10,000 online controlled experiments (A/B tests) every year. Microsoft’s Bing makes dozens of monthly improvements, which have collectively boosted revenue per search by 10% to 25% a year.
A/B testing is one of the most effective approaches to check if changes to a website perform as hoped. It enables brands to see if improvements actually do optimise the customer experience or increase conversions or revenue, for example. After all, even small adjustments to the position of elements on an e-commerce website, such as changing banners, rewording a CTA, incorporating social proof, or reshuffling steps in the checkout process, can often make a significant positive impact.
Yet we still come across many Ecommerce brands that shy away from A/B testing new ways of doing things. Or if they do test, they don’t always act on the results, suggesting that they may not be as data-driven as they might claim to be.
Here are six ways that companies can overcome any concerns and challenges around experimentation:
Make testing part of the culture
Many digital-first brands are founded on a culture of testing and learning from the results, whether they are positive or negative. A test that doesn’t deliver improvements is not seen as a failure that impacts someone’s career – instead, it is simply providing useful data to drive an optimised customer experience. All e-commerce companies therefore need to embrace this “test and learn” culture if they are to improve their online performance and out-compete their rivals.
Act on test results
Sometimes, even when a test shows that a change can make a positive difference, it isn’t pushed through for internal reasons. For example, senior leadership might be reluctant to try something new, or the branding team finds that a suggested change goes against their guidelines. Overcoming this issue requires education and potentially starting small, such as testing incremental improvements in specific areas, to prove that the process delivers results.
Take a granular approach to testing
A test might show that a new offer or promotion doesn’t work for a broad cross-section of customers. However, if you look more deeply, and your test is comprehensive enough, you might spot a positive impact on a key segment, such as first-time visitors, frequent buyers, or those with an affinity to a specific brand, colour, style, and more. By taking a granular approach, you can improve the experience and conversions for these key segments.
Look beyond short-term revenues
The nature of A/B testing requires traffic to be directed to different variants of a page or element, some of which won’t perform so well. Senior managers could worry about how this impacts revenues. However, looking longer-term, improvements should increase overall revenues. Besides, these days it’s possible to use AI machine learning to minimise any immediate revenue risk by automatically switching traffic to the winning variant as soon as statistical significance is shown.
Become data-driven
Many online retailers still take decisions, such as what products or product bundles to recommend to buyers, based on gut feel or how they’ve always done things – or even what has worked in physical stores. In the fast-moving world of Ecommerce, what has been effective in the past doesn’t necessarily work now. That means becoming data-driven and acting on test results based on actual customer behaviour and needs.
Allocate sufficient resources for testing
In many companies, nobody really owns testing, which means it never receives the budget and tools it requires. This has a knock-on effect on resources, making it a complex, manual process that marketing or e-commerce teams cannot carry out themselves, requiring support from overstretched IT departments. Allocating responsibility for testing and investing in automated tools will help make the process simpler and more manageable, with a positive impact on the number of tests that can be run.
In competitive markets, retailers need to constantly enhance the Ecommerce customer experience – experimentation provides the data and insights to deliver these improvements, improving engagement and delivering greater revenues.
