Consultancy & Design of Experiments | Cultural Design

Consultancy and Managed Service

Test and conversion page

The first step in optimising the client’s websites is for a decision to be made upon the test page. In general this tends to be the page which will potentially provide the greatest conversion lift. Our experience has shown that this is likely to be the page linking to the start of the conversion process (for example a landing page with a call to action in the form of a button with “Buy now” linking to a checkout). The client then specifies the conversion page. This page should provide a positive indication that a user has converted (for example the “Thank you” page at the end of the buying process).

Platform installation

Once the test and conversion pages are chosen the testing platform is installed. This requires:

The insertion of a single script tag into the HTML of each of the test and conversion pages to load our software.
The addition of an HTML page for communication between the client site and our software.

This task can be performed by the client, with guidance from Oban Multilingual if necessary or alternatively arranged by us if access to the site is given.

Research

Oban Multilingual maintains a proprietary knowledgebase of cultural issues, interface design, human computer interaction and established best practices. This allows us to evaluate chosen pages within the context of the website itself and local/global internet trends and suggest possible optimisations for both the initial and any follow-up experiments. We then provide a short list of those combinations which we hypothesise as potentially producing the greatest positive conversion lift. This allows decisions to be made on which elements to prioritise and the changes to make to these elements.

Experiment setup

We use several factors to aid us in deciding upon the optimal number of combinations to test from the short list and the time to allot to the experiment. These include monthly estimated traffic, current conversion rates and unusual external factors, identified in discussions with the client. Once a provisional length of time for the experiment is determined, start and end dates can be set. Dates should be considered flexible in the case that a significantly higher or lower traffic volume than expected is obtained. To make the experiment as representative as possible the date range chosen should not involve exceptional circumstances (for example end of a tax year). If, however, budgetary and/or time constraints mean it is not possible to change a proposed date range, the experiment can still proceed. Due to the identification the possible external factors affecting the experiment, the estimation their effect and the running of the original variation in parallel as the control, we can cater for both predicted and unpredicted factors. When the experiment is running we test optimised combinations in parallel with the original page. This gives the results a frame of reference as conversion rates without any optimisation will be shown.

Experiment running and report

After the testing platform is installed and a final decision has been made on the experimental factors (elements) and levels (alternatives to those elements) to test create the combinations. Each of these will be tested prior to the experiment going live. On the start date the experiment goes live. At this point each combination, including the original will be assigned an equal percentage of all visitors, with visitor and conversion data being collected. Oban monitors the experiment over its running period, and following the end date produces a report on the results. This report includes conversion rates for combinations and elements, our ongoing research, an explanation of the trends and the results and suggestions for follow-up experiments.

Please email sales@globalmaxer.com or call on +44 (0) 1273 704 434 for further details and pricing for managed or consultancy services.