Posted by Kristine Jacobson

AB testing imageEvery marketer understands - or every marketer should understand - how important websites are for business. A brand's website is the pillar of its identity online, and - for many prospects - a first introduction to the business and its products.


That being said, not every website serves the same purpose. Some exist to sell products; some exist to build a personal brand; others exist to host an application. Whatever a website does, the key to its success boils down to one major factor: web design.

Web design determines how likely visitors are to stay on a site, make a purchase, click a CTA (call-to-action) or browse to a certain page – and these are all goals a business may want to accomplish. But knowing the best way to achieve them can seem like a shot in the dark.

Wouldn’t it be nice if there was a way to take the guesswork out of web design and get a great result every time? Fortunately, there is, and it’s called A/B testing.

What’s A/B Testing?

A/B testing – or split testing – is the process of comparing two different versions of a webpage or web element and choosing the one that works best for a desired outcome.

When repeated many times, A/B testing can turbocharge a website’s performance by gradually breeding the best possible version for a brand’s goals.

With the right analytics, the potential of A/B testing is practically limitless: in theory, it could even be used to produce the ugliest website possible. The method doesn’t care what you use it for – the point is to use it correctly.

How to Conduct A/B Tests

In one example of A/B testing, an online shop raised its CTA click rate by 49% simply by adding text to the button. Another company increased its leads by 256% by creating a mobile optimized website. While these are good outcomes, replicating them does take patience and precision.

Many tools and suites exist to facilitate A/B testing, and we recommend you use them. However, good A/B testing can’t be fully automated: it requires specific goals, good data and thoughtful analysis.

Setting Goals

Every successful split test begins with a specific but general goal. It’s easy to get sidetracked during the testing process, so it should be written down before any steps are taken.

Common examples include:

  • Drawing more visitors to a page or site feature
  • Reducing bounce rate
  • Increasing conversion rate
  • Raising sales

Goals flow from the objectives that define your business model – whether that’s selling a certain product or getting more users for an application, there will be certain steps along the way.

Identifying Relevant KPIs

After a goal has been defined, it’s necessary to identify key performance indicators (KPIs) that are relevant to accomplishing that goal. In some cases, the right KPI to track is obvious – if a business wants to retain more visitors, it should watch bounce rate and time on page.

In other cases, the right KPIs are more obscure: sales numbers may be related to time on page, return traffic, pages per session and many other metrics. Narrowing down the right one involves a combination of intuition, research and experimentation.

In some cases, it is possible to gather qualitative data directly from customers through polls (“What made you abandon the shopping cart?”). Other times, case studies from other businesses can provide good clues.

But most of the time, watching your own analytics for correlations between metrics and performance is the best way to find relevant KPIs for accomplishing goals.

Gathering Accurate Data

Without a good analytics service, measuring KPI performance over time is impossible. And while there are a variety of third-party services for monitoring user behavior on a website, by far the most common is Google Analytics.

Google Analytics provides tracking code called a “pixel,” and an advanced dashboard for measuring, comparing and exporting vast amounts of information about your visitors. Best of all, it’s free.

In theory, it would be possible to A/B test a website by making changes and manually tracking KPIs. But Google – along with other third-party trackers – offers A/B functionality to make this task much easier.

Performing the Test

Once you’ve established your goals, KPIs and data sources, it’s time to create and compare two web pages. With A/B testing tools like Google Optimize, Optimizely and Kissmetrics, it’s possible to do this conveniently without disrupting normal website operations.

Split/multivariate testing tools often provide limitation features, allowing users to exclusively test a single audience segment. This allows for more refined experimentation and vectors for analysis.

While every tool differs in its workflow, we’ll use Google Optimize as a paradigmatic example of the steps involved:

  1. Create a test hypothesis – in this step, the user records a hypothesis about cause and effect (“the bounce rate is high because the font is too small”)
  2. Create test variants – in this step, the user creates one or more alternative versions of the page to display to users (larger fonts)
  3. Set objectives – here, the user tells Optimize exactly what they hope to accomplish through the test variant so the service can measure the results automatically (in this case, a lower bounce rate)
  4. Launch and monitor results – a successful A/B test requires a reasonably amount of data to be reliable. Website traffic ebbs and flows, leading to false positives when not enough time has passed. 1-2 weeks is generally considered adequate.

Keep Refining

If there is any such thing as a “living document,” it’s your website. Even if you land on the perfect design, times change, and so do people. When it comes to the digital environment, the lifeblood of success lies in continual testing, modification and refinement.

A/B testing provides a scientific way to optimize your site’s performance without resorting to pure guesswork. As we have just shown, it’s a method that is now easily available to anyone who wants it. If you haven’t tried yet, there’s no time like the present to start.

Conveyance Marketing Group is a team of bright, innovative and talented veteran marketers dedicated to big ideas, fresh insights and measurable results. We pride ourselves on taking challenging marketing issues and turning them into opportunities for our clients, on pointing brands in the right direction, and on getting our customers noticed both online and off. From branding to websites to digital marketing, and public relations, we handle all your marketing communication needs! Web Design and Development | Brand Strategy | Inbound Marketing | Social Media | SEO | PR

Kristine Jacobson

Kristine Jacobson

Kristine Jacobson has more than 25 years of marketing and communications experience with notable corporate leaders as well as emerging market contenders. She offers expert marketing strategy with a touch of creative flair. Her extensive knowledge of strategic marketing, marketing plan execution, and branding illuminate the big picture without losing sight of the details.