Historical Stock Chart
2 Months : From Feb 2020 to Apr 2020
By Sara Castellanos
Online crafts marketplace Etsy Inc. has doubled the number of experiments on its website, including those involving data analysis and machine learning, as moving to Alphabet Inc.'s Google Cloud has given the company access to more computing power.
Last year, the Brooklyn, N.Y., company ran several hundred experiments every quarter, more than double the levels in 2018, said Chief Technology Officer Mike Fisher.
Such experiments determine how best to sort product reviews or how to alert potential customers when a particular item might be snapped up by someone else, with the goal of nudging people to buy the handmade and vintage wares available on its website.
The experiments eat up a lot of computing power, which Etsy now gets through Google Cloud instead of relying on its own infrastructure. Before the shift to the cloud, creating and running data science experiments could take several days or weeks and now they can be done in a few hours.
"Now, our engineers can run the data, get it processed and be ready to train their models in a day or two. The speed of innovation is so much faster now," Mr. Fisher said.
The recently completed move to Google Cloud, announced in 2017, was partly motivated by the company's interest in ramping up experiments involving machine learning, the science of getting computers to act intelligently without being programmed.
The shift allowed Etsy to not renew its contracts with two data centers and cut its number of servers to a few hundred from a total of 3,000, Mr. Fisher said.
Etsy's website carries more than 65 million items from more than 2.5 million sellers. Before the move to the cloud, Etsy had purchased the servers and computing power it needed to run its website about once a year, making it difficult to plan for tests engineers would want to run. It wasn't cost effective to buy extra computing power that would likely remain idle in hopes of running future tests, Mr. Fisher said.
Other businesses across industries have also shifted to the cloud. Previously, companies that managed their own IT infrastructure stored data across various systems and units, which made information difficult to analyze, said Chirag Dekate, senior research director at research firm Gartner Inc.
Another benefit of the cloud is that engineers who were in charge of maintaining back-end servers and hardware infrastructure are now freed up for other tasks, Mr. Dekate said. They can help develop new apps, IT systems or artificial-intelligence projects that directly benefit customers and therefore add more value for the business, he added.
At Etsy, about 75 of its 500 engineers have been reallocated from infrastructure management to coding and software tasks that have a direct impact on customer-facing initiatives such as the website, Mr. Fisher said.
"Our customers don't care if we're the best in the world at supporting hardware. They care about us having the best marketplace, with the functionality and features they want," Mr. Fisher said.
Because of the cloud, Etsy can now run various search experiments with more data to inform the algorithms. For example, software engineers have recently tested different variations of nudges on the website to see which ones might be most effective in persuading a customer to buy a product, Mr. Fisher said.
A nudge might be a box near an item in a customer's shopping cart showing how many people are interested in that product, how many also have that item in their cart, or how many of the items are left.
Etsy is also testing whether customers are more likely to buy a product if the reviews they see first are negative ones. In some cases, buyers could interpret negative reviews as "honest feedback," Mr. Fisher said.
Write to Sara Castellanos at email@example.com
(END) Dow Jones Newswires
February 19, 2020 15:53 ET (20:53 GMT)
Copyright (c) 2020 Dow Jones & Company, Inc.