As a stock image website, Bigstock aims to provide an intuitive and fluid experience for finding appropriate images of all types. Every potential customer’s experience typically begins with a search. The easier it is to find the right image, the better Bigstock’s bottom line and customer happiness.
Boris Masis, a product manager at Bigstock, uses testing to guide Bigstock's product rollout strategy. Running controlled experiments with Optimizely helps to determine whether product improvements and new features are making a measurable impact on the users' experience.
When it came time to revamp Bigstock's website search functionality, Boris and his team asked: How can Bigstock improve the search experience for users to rapidly find the image they are looking for?
The success of an image search on Bigstock depends on the search term and the search algorithm. Since the method a visitor uses to arrive at the perfect image is subjective, Bigstock examined the search algorithm, a static code that controls the results a user sees. They thought trying a different search algorithm might improve overall user experience and lead to more downloaded images.
The original search algorithm – an exact auto suggest – populated results based exactly off the word entered in the search field. For example, when entering “c-a-l-a,” the top result would be “calamari.” Most often, people searching “c-a-l-a” were actually searching for images of a calendar and just misspelling the word. Showing users top results for calamari instead of calendars might deter people from finding the perfect image, Bigstock thought.
One of their developers created a “fuzzy autosuggest” algorithm that predicted the intended search term for commonly misspelled words. Now, “c-a-l-a” would populate “calendar” as the first search result, since it’s such a common search term. Bigstock needed to determine if their site visitors would be more likely to select results with the fuzzy or the exact search algorithm.
Optimizely provided the platform for Bigstock to set up an A/B test that bucketed all their live site visitors onto either the fuzzy or exact search page. They were then able to set up a tracking goal with Optimizely to measure how many times a visitor in each bucket selected a search result. Whichever algorithm received more result selections would be the winning variation, taking the guesswork completely out of the equation. Integrating Optimizely with Bigstock’s Google Analytics account provided even deeper insight about how each algorithm impacted conversions further down the down-load funnel.
On Optimizely’s results page, Bigstock could clearly see that visitors selected results from the fuzzy auto suggest 9.6% more often.
With Google Analytics, they were also able to see significant improvements throughout, including a 6.52% increase in the number of images added to the cart, and a 3.2% increase in downloaded images. In under a week, Bigstock used Optimizely to solve a business challenge that enhanced user experience and increased revenue.
Testing ensures that our work is adding value and quantifes that value. Of course, we also do testing when we are unsure of which path to take.— Boris Masis, Product Manager, Bigstock
Each change, improvement or new feature roll-out on your site presents an opportunity to test and measure.
Test the big pieces first and do exploratory work before optimization work. For example, try out entirely new algorithms or page layouts first, and then do smaller tests like tweaking buttons.
Try Optimizely free for 30 days
You can get the very best of Optimizely without spending a dime.
Try it out for 30 days, on us.
Something went wrong
If this error persists, please let us know at our support page.
Our engineering team has been notified.
Please tell us about yourself and your company (all fields required):
Create a developer account
Get a free account with full access to Optimizely's APIs and SDKs.