It’s that time of year again—the holiday season. The task of present picking has begun.
Whether you’re considering a 3D print of an F1 circuit for Dad, a crocheted replica of a pet for your partner, or a vintage leather jacket for your sister, with over 7.5 million sellers, Etsy likely has it.
Founded in 2005, US-based Etsy has evolved from a platform for local artisan market stalls into a global e-commerce site.
By providing a digital space for handmade goods, vintage items, and unique creations, Etsy calls itself a “global marketplace for unique and creative goods.”
The platform’s increasing scope, which sees sellers competing to sell their over 120 million different wares, has the potential to overwhelm consumers with choice.
As of Q3 2024, Etsy reports over 95 million active buyers, attracting 400 million monthly visitors, with nearly $2 billion in revenue generated during the first three quarters of the year.
“When I joined 10 years ago, Etsy hadn’t yet hit a billion sales,” says Rachana Kumar, Etsy’s CTO, speaking at Web Summit in Lisbon. “Yet, last year, Etsy sellers saw over $12 billion in sales.”
Crafting a technology strategy
Kumar explains that over the past decade, Etsy made significant changes to its technology strategy to support its rapid growth.
Initially, the company built most of its technology in-house, developing tools and platforms to meet its specific needs.
“For example, we created our own experimentation platform to test new features,” she says.
“But as the tech landscape evolved, we began to see value in external tools and services that could accelerate our progress. Now, we increasingly buy technology rather than build it ourselves.”
She says that this shift has allowed Etsy to focus more on innovation and user experience while leveraging external expertise for areas like cloud computing, artificial intelligence, and machine learning.
To support this, Etsy has grown its engineering team from 100 to nearly 1,000 engineers across product and tech divisions, ensuring that the platform can handle both its growth and its users’ demands.
Then, in 2019, Etsy transitioned fully to cloud-based infrastructure. This move came just in time for the lockdowns over the COVID-19 pandemic, which saw more online shoppers and more sellers uploading products they were sourcing, crafting, or 3D-printing, just to name a few.
“If we hadn’t migrated to the cloud, scaling to meet the demand during COVID would have been much more difficult,” Kumar acknowledges. The shift provided greater scalability and laid the groundwork for holiday season peaks.
Now, as the holiday shopping season ramps up, Etsy is adjusting its systems to handle the spike in traffic that comes with Black Friday and the weeks leading up to Christmas.
Preparing for peak demand
Planning for the holiday surge begins months in advance.
“We start preparing for the holiday season in the summer,” Kumar says. Etsy works closely with its cloud provider, Google Cloud, to evaluate its capacity needs based on data from the previous year.
“We use those projections to ensure that our systems will scale effectively during peak traffic periods,” she adds. To ensure the platform can handle high demand, Etsy runs scale simulations — simulated traffic spikes that allow the engineering team to test how the site performs under heavy load.
Etsy also strongly emphasises real-time monitoring during high-traffic events like Black Friday and Cyber Monday, making adjustments on the fly to maintain a smooth user experience.
“We closely monitor traffic and performance metrics during key events and quickly adapt to any unexpected spikes in demand,” Kumar explains.
The growing role of AI and ML
In recent years, Etsy has incorporated more AI and machine learning (ML) into its platform to improve user experience.
Kumar says that in 2016, the firm acquired the machine learning platform Blackbird Technologies to enhance its search functionality.
“Over the last couple of years, AI has become integral to many parts of our platform,” says Kumar.
“We started by using traditional machine learning models to improve search. The basic problem is when someone comes to Etsy and searches for an item, we need to match their search query with relevant products and then rank them based on what we think they’d prefer.”
“When someone searches for an item, like a gold necklace, it’s relatively straightforward to understand their intent.”
“But with recommendations, it’s more complex. We have to predict what users might be interested in next, based on their browsing history, preferences, and past interactions.”
Etsy’s AI-powered recommendation system helps users discover products they might not have explicitly searched for but could still find interesting.
“People come to Etsy looking for unique, one-of-a-kind items, so our recommendation system needs to consider broader interests,” Kumar says.
“We aim to suggest items that align with users’ preferences, even if they haven’t searched for them directly.”
In addition to improving recommendations, Etsy has introduced a “Gift Mode” feature powered by machine learning to assist users in finding personalised gift ideas.
For example, if a user shows interest in wine, Etsy might suggest wine-related products suitable as gifts for others.
Beyond personalised shopping, Etsy uses AI for other operational efficiencies, such as optimising shipping routes and predicting delivery times.
The platform also employs machine learning to improve trust and safety by identifying and flagging items that violate Etsy’s policies. “AI is deeply integrated into many parts of our platform, not just for search and recommendations, but also for improving operational processes,” Kumar adds.
While Etsy builds many of its machine learning models in-house, it also partners with external companies, especially for large language models, which Kumar says are costly to develop internally.
“We’ve created an internal platform that allows our engineers to integrate external LLMs easily into their work without dealing with the complexities of the underlying technology.”
Innovation through a lean team
Despite its significant expansion, Etsy maintains a relatively small team — just over 2,000 employees — compared to other e-commerce giants.
Kumar explains that the firm encourages innovation in several ways and emboldens employees, including product managers, engineers, analysts, researchers, and designers, to work together to solve specific problems.
“Our engineers take ownership of the problems they work on,” Kumar explains. “It’s a different environment from larger companies, where it can sometimes feel like you’re just a small part of a much bigger machine.”
“We also hold hackathons a couple of times a year, where engineers are encouraged to experiment with new ideas.
“Some of Etsy’s most popular features, such as Etsy Lens, came out of these hackathons,” Kumar adds.
Rachana Kumar, CTO, Etsy
In addition to hackathons, Etsy allows engineers to dedicate 10% of their time to exploring new technologies and developing prototypes.
This time is often used to experiment with emerging tech, like LLMs, to improve search and recommendations. “This commitment to innovation helps ensure that Etsy remains agile and can quickly respond to changes in technology and user expectations,” Kumar says.
Looking to the future
Looking ahead, Etsy plans to continue focusing on using technology to improve the seller and buyer experience while staying true to its core mission of “keeping commerce human.”
“We’ll keep pushing the boundaries of how we use technology, but always in ways that enhance the human experience of shopping and selling on Etsy,” Kumar says.
AI and machine learning will play an even greater role in the future, especially in helping Etsy’s diverse group of sellers — many of whom are independent creators working from home — streamline their operations.
“We’re focused on using technology to help sellers save time, connect with buyers more easily, and expand their reach. This allows them to focus more on their creative work and less on the complexities of running a business.”
