Segmenting your A/B testing by marketing channel

Blog / 05 Sep 2019

By the end of 2019 - for the first time ever - digital ad spend will represent more than 50% of total global marketing spend. In 2018, the UK was considerably ahead of this trend, with 63.8% of total ad spend attributed to digital, and expected to rise further to 66.4% this year. With eMarketer forecasting a 17.6% year-on-year growth (to $333.25M) in worldwide digital marketing spend, the need to ensure each of your marketing channels is delivering the best possible ROI has never been higher.

Within the Conversion Rate Optimisation (CRO) space, most brands conduct A/B Testing without fully taking into account which marketing channel or source their customers have come from. Customers are typically bucketed into various user segments based on their purchase history, onsite behaviour, geographic and demographic data. However, users within the same audience segment can often demonstrate varying behavioural attributes when navigating through the purchase funnel, across countless online/offline touchpoints.


Let’s take Paid Search as an example.

If a user arrives on your website via Paid Search, you already know what they searched for and which ad they clicked on; however, users who click on the same ad, but searched for different terms/items, will often experience the same User Journey. For instance, if a customer has searched for "Luxury Men's white shirt" – not only do you know the item they are looking for, you also know they are looking at the higher end of the market.

A/B testing the landing page a user is taken to is quite common, but you can go a step further and explore how to change the experience for the customer based on their search criteria.

A potential testing idea could involve pre-sorting these shirts by Highest Price first, and on the Product Listing Page (PLP) displaying all the available Men's White Shirts. This can develop into Personalisation if the user has visited the site previously, within the cookie period; e.g. by storing size data within the cookie, you could pre-select the shirt size which the user filtered by on their previous visit. Reducing the number of clicks and filters it takes a user to find their item can only have a positive impact on Conversion Rate, especially on mobile.

Research conducted by Insider suggests that on mobile, the average sale/conversion requires 5 clicks, 3 searches and 5 filters. So by showing a customer the items they're looking for, sorted by their desired price point and filtered by their size, makes this purchase journey more tailored to that specific user. 


Making the Affiliation between A/B Testing and Voucher-Cashback Partners.

By applying this testing method to the Affiliate Channel, you can optimise the largest click and revenue drivers; namely voucher and cashback websites. After all, you can already assume that users coming from these two Affiliate types are both online-savvy and price-sensitive.

Voucher and discount websites should have a Conversion Rate of at least 20-25% on mature Affiliate programs – so any of these Affiliates who have a Conversion Rate lower than that, represents an opportunity for incremental revenue. For cashback sites, expect this figure to be upwards of 40%.

A test idea for these two Affiliate types could be to re-enforce the discount or cashback offer listed on the Affiliates' website. For instance, if the deal was "Save £15 when you spend over £100" – you could use a "loading bar" at the top of the page which gradually fills up as you add items to your basket, until the user hits the spend threshold to activate the discount. For cashback sites, you could test a cashback calculator onsite, which automatically calculates the amount of cashback the user will earn if they purchase everything currently in their basket. This type of gamification can be incredibly effective in increasing the number of units per sale and, in turn, Average Order Value.


Serve Less content, But More dynamically.

"Content is King" – we’ve all heard it before, but can we be smarter in how we serve it?

Content, and specifically Dynamic Content, is another channel where source-based A/B testing can improve Engagement, Click-Through-Rates and Leads/Sales. If you know the article or blog post a user has come from, you can use this insight to serve them relevant and Dynamic Content and make the User Journey more seamless and less detached across the two sites.

User Journey analysis shows that visits to Content sites usually happen in the "Discovery Phase" of the purchase funnel – including product review sites, influencers, news/magazine sites and blogs. Such content is often informative and persuasive; perfect to push the user towards the bottom of the funnel.

Some of the more content-heavy merchants, such as Insurance brands or high-end technology retailers, will have an eclectic and extensive array of content across their website, making navigation more muddled. A solution? Reduce the amount of content onsite and instead store the less frequently visited content pages elsewhere, to then be served dynamically.

For example, if a user looking to buy insurance is reading up on Excess and the impacts it has on a claim and future premiums, the existing content about Excess could be tweaked accordingly – which could be as simple as changing the title of an article, calling out the keywords or changing the order of the content on that page.

As the capabilities of A/B Testing and Personalisation Platforms continue to evolve, the way you test and analyse a User Journey should also follow suit. One of the major challenges of Channel/Source Specific Testing can be a lack of traffic volume. If you have insufficient traffic, it will take a while before a test reaches significance. For example, the 5th highest Paid Search term, or 4th largest Voucher site probably won't have the volume to justify running an A/B Test on.


Want to keep learning?

Join our upcoming webinars on Experience Optimisation (17th September) and Personalisation Strategies (9th October). They're free to attend and will provide you with many actionable insights.