español  |   contact us  |   search

Experimenting with dynamic content

If you have an e-commerce site with many products, chances are that you use templates to dynamically build each product page using content from a database. For example, you may want to find out which of three image types for your products will generate the best sales: the original image showing the product on a mannequin, alternative 1 showing the product in a studio, and alternative 2 showing the product in an outdoor setting. The URLs for the product images are likely stored in the database.

This scenario is a fairly simple exercise in image variation, but it's complicated by the fact that the images are dynamically referenced. This scenario will illustrate how to experiment using dynamic content. Website Optimizer can support dynamic web pages or templates (such as ASP, JSP, PHP, CFM) and can work with dynamic data from a database, another web service, or code in your dynamic web page.

Before you begin: Design your experiment.

The work in this step is preparatory design work, done outside the Website Optimizer tool.

* Choose your test page
* Choose your conversion page
* Select the content and variations you want to experiment with
* Decide how many visitors you want to include in your experiment

Choose your test page

In this scenario, your test page will be the content management system (CMS) template used to build the product page. Although you'll work with only one file, all product pages using this template on your site will act as test pages.

Choose your conversion page

Because you're working with multiple test pages, it's a good idea -- at least initially -- to choose a single common conversion point for this scenario. A shopping cart page or a confirmation page is probably the best option.

Select the content and variations you want to experiment with

Your template contains both dynamic and non-dynamic content; you can experiment with either or both types. In this scenario, you'll be testing some dynamic content in the form of image URLs. You'll be experimenting with three different types of images, each one corresponding to a variation.

For this type of experiment, you'll need to do some preparation. First, develop a file-naming scheme to distinguish each type of image. For example, if the original image file is named P1234.jpg, the variation images might be named P1234_studio.jpg and P1234_active.jpg. With this decision made, make sure all the original and variation image files follow this scheme and are stored in the database. In this scenario, we'll assume all image files are stored in the same location. The location and naming scheme information will be used in Step 2.

Decide how many visitors you want to include in your experiment.

Traffic percentage depends entirely on the volume of traffic your test and conversion pages get and how many combinations you're testing. See the Website Optimizer Helpful Hints document for things to think about when deciding your traffic percentage.

Step 1: Identify experiment pages

Once you've completed the steps in the Before you begin section, you're ready to start entering information into the Website Optimizer tool. Go to the Website Optimizer page and create a new experiment. If this is the first experiment you've created, we suggest reading My First Experiment for help with getting started.

Enter a name for your experiment, and enter the template page and the conversion page URLs you identified earlier.

Click Continue.

Step 2: Add tags to experiment pages

Follow the instructions in Step 2 of the Website Optimizer tool to install tags on your test page (in this case, the template page). Install the control script and tracking script as instructed.

The page section script will need to be updated to handle the dynamic content. In this scenario, the goal is to construct an HTML image tag with the appropriate file name. As described in the Before you begin section, image files should follow a naming scheme that distinguishes the three types of images, such as P1234.jpg, P1234_studio.jpg and P1234_active.jpg.

In the example script below, the file name is broken into three parts: product_image (P1234, for example), image_type (such as _studio), and extension (.jpg).

The product_image name is retrieved from the database, the image_type is tied to the experiment variation and chosen by Website Optimizer, and the file extension is assumed to be constant. This example script then constructs the HTML image tag using the name and type data.







This technique can be adapted to a number of different scenarios. The key is to figure out a way to combine variations and dynamic content into a unique identity. We've added a couple of mini-scenarios for dynamic content at the end of this scenario.

Once tagging is complete, click Continue to move on to Step 3.

Step 3: Create Variations

Follow the instructions in Step 3 of the Website Optimizer tool to enter the content for your new variations. For this dynamic content scenario, the variation content is not displayable content, but rather a value that the test page script will combine with dynamic data to form the displayable content (in this case the HTML image tag). As described in Step 2, this experiment's variations specify the different image types to be displayed, in this scenario expressed as a studio image and an active image of the product.

To be consistent with the example script in Step 2, the two variations must be as follows:

For page section Product Image:

Variation 1:
Variation 2:

With these variations, Website Optimizer will randomly select an image type, which will cause the test page script to randomly display one of the three images for the product.

Click Continue to move to Step 4.

Step 4: Review and launch

Double check your experiment set up, and specify how much traffic to include in the experiment.

Click Start to launch the experiment.

View your reports

Although the reports won't offer any definitive conclusions for a while, you should check your reports after a few hours to make sure impressions and conversions are being recorded, and that the traffic recorded in the experiment is consistent with the traffic measured by your web analytics tools. Zero impressions and conversions for one or more variations may indicate that the experiment tags were not installed correctly.

In this scenario, because there is only one page section, data on the Page Section report and the Combination report will be identical.

Stop the experiment

You can stop the experiment at any time, regardless of whether or not Website Optimizer has identified a clear winner. The experiment will continue for as long as you want it to.

Check this article in our Dimnet Mexico site
Check this article in our Dimnet Venezuela site
Check this article in our Dimnet Argentina site
Check this article in our Dimnet United Kingdom site

Tell a friend

Print this page

Take it in PDF