A/B Testing Gets Results with 1Above

16th Oct, 2014
by Rory Braatvedt

Cta Join Us

Sign Up

Get Union Blog
updates straight
to your inbox

Sign Up

×

When it comes to deciding upon the best option for a client, sometimes it is best to remember the wise words of the ‘El Paso’ taco girl… “Why not both?”

One of the first projects we took on for our flight-tonic client, 1Above, involved a Google AdWords campaign to test the waters of Search Engine Marketing. This was previously an unexplored space for the brand, being both new to the market and selling an entirely new product category. The goal was to gain awareness around the brand and the features/benefits of its product, ultimately leading to purchase. In order to more comprehensively understand the user’s journey in regard to path to purchase, we decided to build dedicated landing pages for the campaign – custom pages with minor changes on key components to measure changes in click/conversion rates.

Sometimes it is best to remember the wise words of the ‘El Paso’ taco girl… “Why not both?”

In fact, we built 5 landing pages: a control page and four other pages, each targeting a specific flight/travel related ailment that related back to what the user searched for in Google. We then had variations of those landing pages which customised the content, layout, and product mix to measure the affects against the user’s journey.  Further to that, the landing pages were differentiated by other variables such as product price being visible/invisible in order to determine the impact on conversion rate.  

1_above_image

We then used our variation tracking platform to determine which landing page and subsequent variation resulted in the best conversion rate and ultimately, the best result for our client. We also investigated click-maps (example pictured above) to understand where the user was clicking on each variation of the page. The beauty of this approach was there was a clear winner, notably outperforming other variations and reinforcing the importance of A/B testing to begin with.

We were then able to prune the campaign back to the best performing option, and let the results speak for themselves. We weren’t carried away by a horde of outstretched arms as per the ‘El Paso’ taco girl, but it was still a successful implementation of A/B testing.