A Profitero study showed Target’s online prices were 25% more expensive than Wal-Mart’s, which were just slightly more expensive than prices on Amazon.
A/B testing confirms it can expand into new product categories with little risk.
Online marketplace uSell.com, which allows merchants to bid for consumers’ used smartphones and electronics, wanted to add trade-ins for other types of used products but worried that doing so would hurt conversions in its core category, says chief product officer and co-founder Christian Croft. After running two experiments with A/B testing vendor Optimizely, however, the e-marketplace learned that expanding its offerings would only add incremental revenue, he says.
In introducing new product categories, uSell initially calculated that it could handle a decrease of up to 5% of its conversion rate on electronics, which it measures as a uSell customer placing an order to sell to a merchant, he says. Through Optimizely’s web-based portal, his team set up an A/B test to compare uSell’s previous home page with one featuring five new categories—kids’ clothing, women’s clothing, textbooks, video games and gift cards—along with electronics, he says.
For two weeks in late May, the e-marketplace sent half its desktop traffic to the first version of the home page and the other half to the second, comprising about 140,000 site visits altogether. In that test, electronics conversions took a hit, dropping by 3.6%, Croft says. Although that was in uSell’s acceptable range, it still made his team slightly nervous because it was closer to the 5% drop threshold than to breaking even with the conversion rate, he says. Any more of a decrease on electronics sales, and adding the new product categories would not increase net revenue, he says.
Seeking more information, uSell repeated the experiment for another two weeks. This time, Optimizely showed no significant change in electronics conversions with the new categories present on the site, Croft says. Because the test was exactly the same as the last one, his team was reassured that the site changes would, at worst, not be significantly detrimental. On June 15 uSell switched over to the new home page for good. Two weeks later, it now processes 20% more orders in total on the site than before, he says.
Croft says it’s important to validate the effect of any changes uSell considers making to its web site before finalizing them. To that end, it conducts four to six A/B tests each month, he says. Though they’re usually for more minor changes than updating the whole home page—for instance changing the text in just one spot—they still ensure that the business is always moving forward, he says. “You could fall into the trap of making changes based on your gut instinct all the time, but you’d never be able to look back and prove that you made progress at each step.”
The technology’s biggest advantage is its simple toolset and the time it saves uSell on engineering and development, Croft says. He can set up an experiment in a few hours on Optimizely’s web site, then start running it as soon as he adds a snippet of code to the uSell web site. Previously, uSell ran A/B tests with the now-defunct Google Website Optimizer, he says. That tool required at least two weeks of advance planning so developers could set up an experiment, two weeks building it and two more weeks to run a test. Because of those limitations, uSell conducted far fewer tests each month with the Google tool, he says. Optimizely also simplifies sharing results, he says, because it provides a shortlink, or a truncated version of what would otherwise be a long, unwieldy URL address, that an experimenter can give to anyone on his team to allow them to watch the results as they come in.
USell pays Optimizely roughly $359 per month to run tests for up to 200,000 site visits, plus $5 per each additional 1,000 visits, Croft says. That’s the third, or Gold, tier of the vendor’s pricing model, which starts at $17 per month to test up to 2,000 site visits, plus $9 per each additional 1,000 visits.