Not that long ago, online retailers would have had a website designed and built, then launched and prayed. But these days they have a new tool to rely on that takes away the guesswork: Experimentation shortens the time it takes to go to market and speeds the improvement process.
That, says Paul Tannock, head of ANZ and emerging markets with Niteco, can double visitor signups, increase margins and – best of all – boost sales because experimentation, as a process, can deliver real-time intelligence on how features of the site are performing and how to improve them even further.
In the past, he says, a company might have been super creative in designing a site but missed the mark in sales performance. “Their customers aren’t engaging with the brand as much – but without the experimentation process, you might not know for six, nine or 12 months why it didn’t work, because you’re just guessing.
“That’s the key part: we are not only launching and praying now, we’re experimenting and iterating.”
Tien Thuy Le, head of digital marketing services with Niteco, adds that many people in the retail industry think of experimentation as being about testing the effects of changing a font, a colour or the text here and there, or an element of a page. “But now experimentation has got so advanced the tools allow you to do multiple things. We can do personalised experimentations for a segment so we can define if people come to a site based on a certain search query. Or how they behave if they visit using a certain device, or from a certain location. We can design tests for all of those segments one by one which help people when it comes to analytics and personalisation.”
Data has limited value if you cannot interpret it
Retailers often collect a lot of data but lack the skills to interpret it and thus are missing opportunities to improve their sites or their processes, says Le.
“They collect all this data, and then they see that their checkout rate is very low, but they don’t know what to do about it because the data tells you where the problems potentially are – but not what the potential solution can be. That’s where experimentation comes in – it bridges the gaps and brings out a few hypotheses, a few potential solutions, and then you let the data prove which solution works.”
The benefits to a retailer of experimentation are easily quantifiable. One client of Niteco – a large Australian retailer – achieved a 95 per cent increase in signups by experimenting with offering different incentives at different points in the customer journey before settling on the mix that proved the most successful. The solution sees different offers at different points according to the time of the day or where the client is located – a level of personalisation that boosts visitor engagement and sales.
“As the complexities and functionality of commerce platforms and customer expectations have grown, that level of personalisation has grown exponentially,” explains Tannock.
Niteco is a Swedish-headquartered multinational technology company with a long history of building e-commerce sites and providing end-to-end digital marketing services to corporations including Electrolux. Now it has evolved into one of the leading CRO agencies globally.
Experimentation is usually divided into pre-purchase and post-purchase experiences. In the pre-purchase phase, Niteco’s team would look first at the shopping flow: how they look at products from the homepage or a product listing page all the way through to clicking on a product and adding to their cart. They will also look at where consumers are engaged on the site – areas such as blogs or advisory sections, for example – because that shows a browser is interested in what the retailer is offering and the content being shared.
For the checkout flow stage, they would look at the purchase rate and then the ‘sign-up’ process – depending on whether a retailer allows customers to check out without signing up, usually, a key conversion opportunity that allows the retailer to collect information for upselling or future customised marketing.
“Post-purchase would be around how engaged the users are – whether they come back to check out stuff on the site, whether they buy more accessories or other products after they have purchased the first time, whether they can easily find support information like user manuals or contact information, that sort of thing,” explains Le.
Forget the focus group
In the past, large retailers might have planned their e-commerce sites by testing concepts in focus groups or in workshops with hundreds of consumers, a process that takes many months and external consultants. However, the more efficient approach of experimentation starts with advising the company to build a minimum viable product that meets all the key requirements. Then you rapidly follow that up with analytics and experimentation.
“So, as opposed to guessing how customers are going to interact, and the products they are going to like and the content they engage with, we use experimentation platforms to learn how customers behave and what they want in a much more accurate and agile way.”
There are multiple ways to display a promotion. You can add urgency, it’s selling fast. You can add a percentage discount or you can display the actual discounted value. Experimentation will work out which methods work best in different markets: consumers in Australia, for example, might respond better to a discount percentage, while the Japanese might prefer the absolute amount. That’s valuable intelligence for a company like Electrolux with sites in 70 markets.
Results can even vary during the day. “In the morning versus evening, you’ve got mobile versus desktop, you’ve got rapid purchasing as opposed to browsing. People might be more likely to sign up at some times rather than others so you might want to vary your click to sign up now button,” says Tannock. For example, someone who sees some shoes advertised on the way to work might not have the time to complete the registration at the office – they just want to buy the shoes and sign out, so you can encourage them to sign up at another time of the day.
The importance of partnerships
“When it comes to experimentation, people often think they need to find a partner to help because they don’t know how to use the tool,” says Le. “The real value of having an experienced partner lies in their ability to formulate a well-defined experimentation strategy that uses everything those tools have to offer.”
A lot of companies are using Google Optimize to run testing but Google is sunsetting that product by the end of September and now they are trying to find another tool that fits their needs.
As an agency, Niteco uses multiple tools. Which one it recommends for a client will come down to factors like pricing, capabilities, and how many tests the client wants per month (some companies might want one or two tests a month, while others might want simultaneous tests of multiple features across multiple markets).
Tannock says the company has had clients who say: We don’t need to experiment because we know the only reason our customers come to us is because we’re cheaper than our competitors. That’s a big bold statement. But if everything is always on sale at 20 per cent less than everybody else, that’s a huge opportunity lost if you don’t test to see if an exclusive line of sheets might sell at a higher price. That’s 20 per cent in margin you are losing.
“So, that’s the biggest challenge: people say they don’t need to experiment because they already know. But they might not know.”