Jeffrey Vocell

Customers navigate a website in one of two ways: they browse or they search. Given the prominence of search and the impact it has on revenue, search should be one of the key areas in your testing plan.

There are a number of elements that can be tested within on-site search, such as layout of results, number of default items shown, size of product images, add to cart right from search results, product badging, and more. All of these elements can be represented by five critical areas that provide a foundation for on-site search optimization:

1. Default Options – Selection areas that allow users to re-arrange, sort, or change results are good primary areas to test. Testing these areas can result in some “quick wins” that prove value of testing on-site search to internal stakeholders.

2. Metadata – In the context of on-site search, metadata is aggregate information about the search results data. For example, have you seen a star-rating system that symbolizes all customer reviews, or an in-stock or out-of-stock badge on a site? Metadata shows information that can help a buyer make an immediate purchasing decision, without having to comb through vast amounts of product information. Try testing specific combinations of metadata to find out what converts and works best with your audience.

3. Filter Options – Many search results pages have a list of product refinement options available. For example, if you search for a digital camera, refinement options may allow you to narrow the search criteria for price, brand, lens focal length, storage card type, and more. One test idea is to alter the default state of refinement options that are shown, such as making them collapsed or expanded. In addition, depending on the number of refinement options, how they are sorted and in which order allows for additional testing opportunities.

4. Sharing – Search results that enable visitors to share items with their own networks on Facebook, Twitter, Pinterest, and other relevant networks should be optimized. Look at internal web analytics to determine which networks are being shared most, and test removing or adding particular networks to determine what the effect is.

5. Functionality – The functionality of your on-site search can produce some of the most meaningful testing results. Some examples of functionality of on-site search include corrections for search result misspellings, different search results algorithms, rearranging search results, and more. For example, many on-site search vendor’s algorithms have different features and functionality and one may work better for your audience than another.

However, as search and vendor functionality differ, it’s worth testing each functional element to find the optimal combination. From these five areas, you can come up with a list of hypotheses that can be categorized and ranked to determine which hypotheses will be tested. For example, here is what a list of hypotheses could look like:

1. Visitors are commonly misspelling words of products they are looking for, such as accidentally spelling calendar as “calendor” and our current search algorithm does not have the “did you mean” feature enabled that attempts to show visitors suggested search terms, such as “did you mean calendar.” If we had the “did you mean” feature turned on, more visitors would stay on the site longer, view the product page, and potentially add a product to their cart.

2. We only show 15 search results by default but traffic and heat mapping show that 87% of people at least scroll to the bottom of the first page of results; let’s increase the default number of results to 30 per page.

3. If we had an “Add to Cart” button next to search results more shoppers would convert by adding items directly to their cart, and the average order values of shoppers would increase.

4. When visitors hover their mouse over a product image, it should automatically open up a larger version of that image. This reduces the requirement that visitors must click-through to the product page to view additional images.

From your own on-site search testing ideas, rank the value of each test, and then begin to incorporate them into your testing strategy. Once you have achieved a few small wins with on-site search testing, take the next step to test additional elements of search.

Tags: A/B & Multivariate Testing Back-End Testing