Blog
-
Featured
-
How to Pick the Right Platform for Optimisation
-
10 min
Back to blog
How to Pick the Right Platform for Optimisation
Namitha Varma Rajesh
If you are a digital business, you must have website optimisation strategies in place. Experimenting with CX, UI, and content, and personalising content and recommendations for customers have become the norm, and if you are not doing so yet, it’s not too late to start.
There are several platforms in the market to help you optimise conversions for your websites, and some of them are even free!
A/B testing and personalisation tools come with several capability combinations. This gives you a variety of choice but also makes it difficult to decide which one is the best for your business. You need to know exactly what features each platform has, in order to align it with your business needs and make the right choice of platform.
Let’s start with listing the key features you can expect in a testing and/or personalisation platform.
Here are some important things to know about what these tools do, how they function, and what feature helps you achieve what result:
Most testing and personalisation platforms are good for website experimentation, and then there are some that are good for both website and mobile app experimentation.
However, if you are a large business with presence in online, retail, and other areas, then you should look for an omnichannel experimentation and personalisation platform.
What you can optimise with an omnichannel tool: Omnichannel experimentation tools allow you to test and personalise not just websites, mobile apps, email, and social media, but also SMS, digital ads, live chat, connected IoT devices, over-the-top media (such as Netflix, Amazon Prime, and smart TVs), physical stores, and even POS and customer support centres.
Client-side testing allows you to create page variations on the user’s browser instead of on your web server. Your server is sending the control version of a page being experimented on, but variations are displayed when the user opens the page on their browser.
This means you don’t have to do any coding or script changes at the server-end; the scripts (best to use JavaScript) are running on your users’ browsers directly.
Challenges of client-side testing tools: While this is an easier way to create and run tests (especially if you do not have sufficient developer support), the disadvantage of depending on client-side testing is that you can mostly only make cosmetic design changes to your variations. Omnichannel tests are, therefore, difficult to execute using a client-side tool.
Additionally, because the original page loads first and the variation is delivered once the testing tool’s script runs on the user’s browser, it is important that you choose an experimentation platform that is flicker-free and renders in the blink of an eye.
Server-side testing is when page variations are created on your web server directly and displayed to users, and no additional changes take place on the user’s browser. A variation of the page is fetched from the server based on how the visitors are segmented in the database.
Why use server-side testing tools: Server-side testing needs support from a team of developers, but it also allows you to do more sophisticated and deeper changes to your pages, including omnichannel experimentation.
Targeting refers to displaying specific content on pages or engagement messages through pop-ups and other overlays, based on user demographics or on-site behaviour. They’re also called nudge engagement.
For example, if you are a fashion website and a visitor arrives on your site from Aberdeen in the peak of winter, you could offer an incentive of a 5% discount on scarves or jackets to urge the person to make a purchase.
Segmentation is important for nudge engagement: For targeting, the experimentation and/or personalisation platform must have user data management and segmentation capabilities.
Demographic user data that can be used for segmentation and targeting include age, gender, location, language, device, browser, source of visit, income range, and profession.
Behavioural data used for segmentation and targeting include pages viewed, search queries, time spent on a page or on site, last visit date or time, clicks made on images, text, or CTAs, and exit intent.
You would want to use a platform that auto-updates these segments based on real-time or new data, so that your targeting stays relevant even if the users’ online behaviour and interests change.
Targeted personalisation is done by setting up rules and conditions for each user segment. You need to create multiple content/messaging options and then set up rules and conditions about which user segment should be shown what type of content, product, or messaging, and based on what behaviour or demographics.
While this is hard work, you don’t have to do it every other day. Ideally, you may have to review your rules only once in 3-4 months, except when your offers/incentives are time-bound.
With advances in machine learning (ML), targeted personalisation based on rules is fast becoming outdated. AI algorithms take personalisation to another level by constantly learning about each user through all available historical and real-time data.
How machine learning-based personalisation is different from targeted personalisation: Apart from on-site behaviour, ML algorithms understand a user’s intent and interests by looking at how they reached your website and what searches or views they perform on-site.
An AI-powered website doesn’t always personalise based on rules and user segments; it actually creates unique experiences for each visitor using everything it knows about their demographics and behaviour and by understanding their intent.
This means that if you are using a machine learning-based personalisation tool, you don’t need to set up rules and conditions; you just have to give the algorithm time to learn about your inventory (product or content) and the behaviour and intent of your users.
When your experimentation is also ML-powered, you’ll be able to show the most relevant pages to visitors based on their intent, instead of serving up random variations to random sets of people. Most AI-based testing platforms track user behaviour across multiple marketing channels, making it easier to recognise what variation will work best for them.
Benefits of machine learning-based experimentation tools: The support of machine learning allows you to conduct smaller but more targeted and relevant experiments. The biggest benefit of ML-based platforms is that your testing period becomes shorter and the results more successful. Since the tool dynamically allocates traffic to variations, conversion rates also tend to be higher.
If you do not trust the technology yet, don’t worry: most machine learning-based experimentation platforms give you the option of just viewing the algorithm-generated insights and suggestions instead of directly implementing them.
User research is the process of understanding your users better by actually viewing their on-site behaviour. It can include the following:
Usually, most experimentation and personalisation platforms do not come with user research functionalities. But getting a platform with user research built in can be useful if you do not want to invest in a separate tool.
Not necessarily. Here are some of the capability-combinations available in testing and personalisation platforms in the market currently:
Different platforms have different combinations of these capacities. For example, one software may only be able to help you with conducting A/B tests on landing pages and getting analytics on that, while another could do omnichannel client-side and server-side testing with AI-powered personalisation.
There’s never a one-size-fits-all answer to this. It depends on what your business needs are and how much you can afford to spend on the technology. As is obvious, the more the features in a platform, the costlier it gets.
Here are some tips on choosing the right experimentation and personalisation platform for your business:
.svg)
.svg)
.svg)
.webp)
.svg)