Article

8min read

Client and Server-Side A/B Testing – The Best of Both Worlds

We’re enriching our conversion rate optimization platform with a server-side A/B testing solution. What is server-side A/B testing, you ask? It’s the subject of an announcement of ours that will make anybody who’s passionate about experimentation pretty excited…because it means they can now test any hypothesis on any device.

No matter if you want to test visual modifications suggested by your marketing team or advanced modifications tied to your back office that are essential in the decision-making process of your product team, we’ve got the right tool for you.

What’s the difference between A/B testing client-side, and A/B testing server-side?

Client-side A/B testing tools help you create variations of your pages by changing the content sent by your server to internet users in the web browser. So, all the magic happens at the level of the web browser (called ‘client’ in the IT world), thanks to JavaScript. Your server is never called, and never intervenes in this process: it still sends the same content to the internet user.

Server-side A/B testing tools, on the other hand, offload all of this work from the web browser. In this case, it’s your server that takes on the task of randomly sending the internet user a modified version.

4 reasons to A/B test, server-side

Running an A/B test server-side has many advantages.

1. Dedicated to the needs of your product team

Client-side A/B testing is often limited to surface-level modifications. These refer to visual aspects, like the page’s organization, adding or deleting of blocks of content or modifying text. If you’re interested in deeper-level modifications related to your back office – for example, reorganizing your purchase funnel, or the results of your search or product sorting algorithm – it’s a bit more complicated.

With server-side testing, you have a lot more options to work with, since you can modify all aspects of your site, whether front-end or back-end. With server-side testing, you have a lot more options to work with, since you can modify all aspects of your site, whether front-end or back-end.

All of this is possible because you remain in control of the content sent by your server to your website visitors. Your product team will be overjoyed since they’ll gain an enormous amount of flexibility. They can now test all kinds of features and benefit from a truly data-driven approach, to make better decisions. The price of this increased flexibility is the fact that server-side testing requires your IT team to get involved in order to implement modifications. We’ll get back to this later.

Your product team will be overjoyed to test all kinds of features

2. Better performance

Poor performance – loading time or the flickering effect – is often the first criticism made about client-side A/B testing solutions.

In the most extreme cases, some sites only add the JavaScript tag to the footer of the page to avoid any potential impact on their technical performance. This policy automatically means excluding using any client-side A/B testing tools, since a ‘footer’ tag is often synonymous with flickering effect.

When using a server-side A/B testing tool, you don’t have any JavaScript tag to insert on your pages, and you’re in control of any potential performance bottlenecks. You also remain responsible for your company’s security policy and the adherence to internal technical procedure.

3. Adapted to your business’s rules

In some cases, your A/B test might be limited to design-related modifications, but you have to deal with profession-specific constraints that make it difficult to interpret a classic A/B test.

For example, an e-commerce merchant might understandably wish to take into account canceled orders in their results, or else exclude highly unusual orders which skew their stats (the notion of outliers).

With a client-side A/B test, a conversion is counted as soon as it occurs on the web browser side when the purchase confirmation page loads or a transaction event type is triggered. With a server-side A/B test, you remain in complete control of what is taken into account, and you can, for example, exclude in real time certain conversions or register others after the fact, by batch. You can also optimize for more long-term goals like customer lifetime value (LTV).

4. New omnichannel opportunities

Server-side A/B testing is inseparably linked to omni-channel and multi-devices strategies.

With a client-side solution – which relies on JavaScript and cookies – your playing field is limited to devices that have a web browser, whether it’s on desktop, tablet or mobile. It’s therefore impossible to A/B test on native mobile apps (iOS or Android) or on connected objects, those that already exist and those still yet to come.

On the other hand, with a server-side solution, as soon as you can match up the identity of a consumer, whatever the device used, you can deploy A/B tests or omnichannel personalization campaigns as part of a unified client journey. Your playing field just got a lot bigger 🙂 and the opportunities are numerous. Think connected objects, TV apps, chatbots, beacons, digital stores…

Use cases for server-side A/B testing

Now, you’re probably wondering what you can concretely test with a server-side solution that you couldn’t test with a client-side tool?

Download our presentation: “10 Examples of Server-side Tests That You Can’t do With a Client-side Solution”

Included are tests for sign up forms, tests for order funnels, tests for research algorithms, feature tests…

How can you put in place a server-side A/B test?

To put a server-side A/B test in place, you’ll need to use our API. We’ve described below in general terms how it works. For more information, you can contact our support team, who can give you the complete technical documentation.

When an internet user lands on your site, the first step is to call our API to get a unique visitor ID from AB Tasty, which you then store (ex: cookie, session storage). If a visitor already has an ID from another visit, you’ll use this one instead.

On pages where a test needs to be triggered, you’ll then call our API passing in parameters the visitor ID mentioned above and the ID of the test in question. This test ID is accessible from our interface when you create the test.

As a response to your API request, AB Tasty sends the variation ID to be displayed. Your server then needs to build its response based on this variation ID. Lastly, you need to inform our data servers as soon as a conversion takes place, by calling the API with the visitor ID, and data relevant to the conversion, like its type (action tracking, transaction, custom event…) and/or its value.

Don’t hesitate to use our expertise to analyze and optimize your test results thanks to our dynamic traffic allocation algorithms, which tackle the so-called ‘multi-armed bandit’ issue.

As you’ve seen, putting in place a server-side A/B test absolutely requires involvement from your tech team and a change in your work routine.

While client-side A/B testing is often managed and centralized by your marketing team, server-side A/B testing is decentralized at the product team or project level. While client-side A/B testing is often managed and centralized by your marketing team, server-side A/B testing is decentralized at the product team or project level.

Should you stop using client-side A/B tests?

The answer is no. Client and server-side A/B testing aren’t contradictory, they’re complementary. The highest performing businesses use both in tandem according to their needs and the teams involved.

  • Client-side A/B testing is easy to start using, and ideal for marketing teams that want to stay autonomous and not involve their head of IT. The keyword here is AGILITY. You can quickly test a lot of ideas.
  • Server-side A/B testing is more oriented towards product teams, whose needs involve more business rules and which are tightly linked to product features. The keyword here is FLEXIBILITY.

By offering you the best of both worlds, AB Tasty become an indispensable partner for all of your testing and data-driven, decision-making needs.

Don’t hesitate to get in touch to discuss your testing projects – even the craziest ones!

Subscribe to
our Newsletter

bloc Newsletter EN

We will process and store your personal data to respond to send you communications as described in our  Privacy Policy.

Article

10min read

Customization and Personalization: Two Sides of the Same (Millennial) Coin

Whether you believe Millennials are narcissistic, ‘special snowflake’ generation or an economically squeezed, misunderstood cohort, most agree they deserve a marketing strategy all their own.  Digitally savvy, uncompromisingly mobile, participants in the sharing and gig economies, owners of ‘networked selves’…Millennials’ formative experiences are distinctive enough to warrant a closer look when trying to understand their consumer habits.

This is where we come to our double-sided coin: customization and personalization.

Strategies that have come into their own since the advent of the internet, customization, and personalization are perhaps two of the best marketing ideas that Millennials actually respond to. Here’s why:

First: Let’s Define Customization and Personalization

To the untrained ear, customization and personalization sound more or less the same. In fact, they refer to two distinct processes. (Also, if we’re getting technical, we really should refer to our first term as ‘mass customization’.) First identified in 1987 in the book, Future Perfect by Stan Davis, ‘mass customization’ has come to mean, “the process of delivering wide-market goods and services that are modified to satisfy a specific customer need.” Distinct from mass production – mass customization’s homogenous, less evolved precursor – mass customization allows consumers to assemble their own, unique product according to their particular tastes and needs.  Think of picking out colors, patterns, and designs to craft your very own Nike shoe or Burberry coat. It’s still branded Nike or Burberry – but it’s also distinctly you. Mass customization is a practice driven by the consumer that allows them to become ‘co-producers’ with the brand of their own uniquely personal product and is a process that must be reproduced at scale by the company.

Burberry Bespoke
The high-end brand Burberry launched their mass customization option, Burberry Bespoke, in 2011.

Source: The Wall Street Journal

Personalization, on the other hand, is a marketing strategy driven by the company, based on consumer data, that aims to create personalized, or ‘one-to-one’ communication with a single consumer or consumer segment. We might date the idea back to the 1993 book, The One to One Future, by Don Peppers and Martha Rogers. When you receive an email that uses your first name in the subject line or see product recommendations algorithmically chosen ‘just for you’, or if a website’s homepage changes to reflect your past purchase history, location or gender…you’ve experienced personalization. The company in question is trying to create a hyper-relevant message designed specifically for you, in the hopes that it will cut through marketing noise to grab your attention and bring you through the sales funnel.

Sephora personalization
Our client Sephora saw a 16% increase in transactions after personalizing their website according to whether visitors were loyalty card members or not

Customization Deep Dive

There is a premium on individuality, especially with millennials and gen Z. The customization thing is like the ‘you but better’ movement, but almost the reverse. It’s like ‘this thing but better, because it’s about me.’

–  Jackie Chiquoine, Racked 2016

You can certainly see the similarities between the two strategies. Both aim to make consumers feel uniquely understood and marketed to as individuals. Both put a high value on self-expression and personal tastes and experiences. Both create the illusion that consumers are being given bespoke treatment and have a one-on-one relationship with the brand. And both, in a way, involve the consumer as a co-producer of value – for customization, it means relying on their predilections to craft the product, and for personalization, it’s sharing their data (unbeknownst to them or not) to create personalized marketing messages.

You can see how those who dramatize Millennials’ navel-gazing tendencies certainly favor these two strategies for targeting them. But customization and personalization also speak to Millennials’ ability to see right through traditional marketing, their distracted viewing habits (customization and personalization more easily grab attention), and their penchant for social networks.  Burberry, for example, savvily allowed anyone to share their customized Burberry creation on social networks, even if they couldn’t actually afford to buy the product.

Burberry share my design
Even if the consumer can’t afford their £1,295 creation, note the ‘share my design’ buttons on the bottom right: they can still express themselves through their customized creation – and promote the brand – on social networks.

Source: Luxury Society

Why Does Customization Work?

If we want to dig a bit deeper as to why customization techniques work – for Millenials or anyone else – Customer Experience Psychologist Liraz Margalit has a few ideas. She explained to TechCrunch that two general principles are at stake: “I Built It, Therefore I Own It”, and “I Own It, Therefore It Is Superb”.

“I Built It, Therefore I Own It” refers to the idea that participating in the elaboration of an item – even virtually, as on a website – leads to a sense of ownership of that item. In Margalit’s words, “The opportunity to take part in a process and influence the end result promotes emotional attachment that leads to psychological ownership, the feeling that something is “mine” even without legal ownership.” She then referred to work by Dan Ariely and his so-called ‘Ikea Effect’, which shows that people put a higher value on objects they have actually worked to produce.

Similarly, “I Own It, Therefore It Is Superb” refers to the Endowment Effect or the finding that, “consumers value an object more once they have taken ownership of it.” Margalit talks about a study at Cornell University in which students favored objects they had ‘owned’, even for a short time, over those they had never had a claim to.

Basically, when you customize an item, you pour a little part of yourself into it, you feel that it already belongs to you, and you value it more – and are therefore motivated to finalize your purchase, and even pay a premium.

These psychological underpinnings, in addition to Millenials’ particular need to be wrenched from constant multitasking and their penchant for self-expression, makes customization a good choice for this age bracket. Throw in the idea that co-creating a product with a brand approximates what Pine and Gilmore call ‘an experience’ (à la, The Experience Economy), and you’ve got yourself a winning hand.

Joseph Pine BBC
Joseph Pine explains how Mass Customization led to the Experience Economy – and what’s next in line in terms of virtual experiences.

Personalization Deep Dive

Millennials want a customer-centric experience in which they feel wanted and valued. Whether it is in-store or through social media channels, showing interest in these shoppers creates loyalty. In order to do this, retailers need to closely examine what they’re currently doing with customer data, and ensure this information is being utilized to deliver a more personalized in-store experience.

– Tom McGee, Forbes 2017

Ultimately, though a product strategy on the part of the company, mass customization is driven by the consumer. Especially appealing to Millennials, it’s a way of validating their own sense of self through ‘build-a-bear’ style product production.

With personalization, the onus is much more on the company to bring value and deliver meaning to the consumer audience they’re targeting.

Because of their ‘digital native’ status, and possibly due to some kind of ‘me me me’ generational fluke, Millenials are more receptive to personalization techniques. Salesforce noted in their State of the Connected Consumer report that, “Sixty-three percent of Millennial consumers and 58% of GenX consumers are willing to share data with companies in exchange for personalized offers and discounts.”

Salesforce personalization
Source: Salesforce

This is, of course, a boon for marketers, since data is the backbone of any personalization strategy. It opens up many avenues for creating relevant – and therefore high converting – experiences for consumers. Personalization strategies can go from relatively simple campaigns, like segmenting emails by industry and inserting someone’s first name into the text, to more sophisticated onsite personalization campaigns to complex multi-device operations. Finding the kinds of scenarios and messaging that will resonate with Millennials will probably take some brainstorming sessions, inspirational reads and inevitably some trial and error – and will, of course, depend on your industry and product. Here’s a quick look at what some marking executives are using for personalization efforts:

eMarketer Personalization
Source: eMarketer

Aside from the platitude of, ‘the more relevant, the better’, is there any psychological data to explain why personalization strategies lead to higher conversion rates?

The quick answer is ‘yes’ – namely a study conducted at the University of Texas that was picked up a few years ago by HubSpot. As they explain, this research admitted the efficacy of digital personalization techniques and boiled its success down to the human desire for control and information saturation.  

People naturally crave to be in control of their surroundings (and their own fate, free will…) and personalization techniques create a cherry-picked environment that feeds into that need. Bargain hunting for low-priced kitchenware from your favorite outlet? If ads for the very items you’re looking for suddenly appear all over the internet, it somehow creates a feeling of empowerment, as if in some impossible way, your wishes and needs automatically manifest.

As for information overload, this is the classic argument that contemporary consumers are exposed to a dizzyingly high number of marketing messages, far too many to ever register consciously, let alone remember or act on. Personalization strategies both (hopefully) limit the number of marketing messages a consumer is exposed to in the first place, as well as stimulate the brain to recognize these stand-out strategies, as opposed to letting them sit in the background of our consciousness.  This is the idea of selective attention or the fact that your brain will automatically pick up on potentially important stimuli – i.e. those most relevant to you.

The Takeaway

Though not the same, (mass) customization and personalization strategies revolve around people’s deep-seated desires for self-expression and recognition – as well as their limited attention spans.  Particularly relevant for the Millenial age bracket, successful customization strategies will make the consumer a co-producer of value alongside the brand, and successful personalization techniques will identify valuable data and use them to create and deploy ultra-relevant marketing messages.