ETSY – “Keeping Commerce Human” through Machine Learning

How will Etsy use machine learning to retain customers in a sea of mass produced products?

A leader in the maker space, Etsy is known for bringing crafting into the 21st century. While the company stands by its people-oriented motto “Keeping commerce human,” Etsy has tied its future success to its ability to harness machine learning.

Process Improvement on

Established in 2004, Etsy is an ecommerce site that connects 2 million makers with over 37 million active consumers looking to purchase unique, artisan goods. In the third quarter of 2018, Etsy saw gross merchandise sales grow by 20 percent. Many experts, including Mad Money’s Jim Cramer, attribute Etsy’s recent growth to improvements in the website and the search experience [1].

Etsy has been able to expand its search and discovery capabilities by utilizing machine learning. With 50 million products on the Etsy website, the company’s key challenge is helping customers quickly and easily find items they are willing to purchase [2]. In 2016, Etsy purchased Blackbird Technologies, a firm that develops algorithms for natural language processing, image recognition and analytics. Blackbird allows Etsy to go beyond the traditional “purchase history” tool. By detecting patterns from massive data sets, Etsy can predict consumer spending based on data points like product preferences and purchasing habits [3]. For example, Etsy now offers a “Exploratory Search” feature that utilizes data to point shoppers towards products that better suit their tastes [4]. By offering nuanced, curated product recommendations, Etsy is able to improve the purchasing process, generate more sales, and offer a greater value proposition to buyers and sellers alike.

Mural at Etsy Headquarters. Photographer: Victor J. Blue/Bloomberg via Getty Images

Handmade Competition

The U.S. “creative products industry” is currently valued at $43 billion. To capitalize on this growing demand for artisan goods, Amazon recently launched “Amazon Handmade,” a platform to sell handcrafted products. To combat this competition, Etsy will focus in its machine learning capabilities to further personalize the shopping experience. For example, Etsy will use algorithms to better rank and categorize search. The front end of the platform will feature tools like “user-generated curated collections” as well as location-orientated data points that similarly help buyers find local sellers. CEO Josh Silverman says these features will give buyers “more reasons to come back [5].”

To expand their capabilities within the machine learning space in the next two to ten years, Etsy will harness the power of its Machine Learning Centers of Excellence around the world. Etsy has built new facilities in New York City and San Francisco and most recently in Toronto. These centers allow Etsy to work with thought leaders at major universities and recruit engineering talent in this field [6].

Future Challenges

Etsy was founded as a mission driven company with a commitment to “to reimagine commerce in ways that build a more fulfilling and lasting world.” CEO Josh Silverman was recently quoted saying, “We’re living in a sea of sameness. People are buying more and more of the same mass-produced goods from the same few logistics companies, and the world wants an antidote to that.” While Silverman may see Etsy as the natural “antidote,” many note the site’s recent influx of mass produced products. Sellers of handmade goods have noticed this trend as well and have seen their sales declining [7].

The challenge for the engineering team over the next couple of years will be to retain their original users who are continuing to search for artisan products amid a sea of mass produced items. Etsy should utilize language recognition to decipher customer comments left on product reviews. With this data, Etsy can determine which products best fit their brand of crafted items and prioritize them during the customer discovery and search process.

Who Holds the Data?

In a world where search engine optimization and inbound marketing techniques are becoming more common place, how will individual sellers on major retail sites, such as Esty and Amazon, be able to adapt to future trends when they cannot access their own consumers’ data?



[1] CNBC Television Mad Money, “Focus on Craftmanship,” [], accessed November 11, 2018.

[2] New York Times, “Inside the Revolution at Etsy,” [], accessed November 12, 2018.

[3] Harvard Business Review, “How Predictive AI Will Change Shopping,” [], accessed November 11, 2018.

[4] Tech Crunch, “Etsy Buys Blackbird AI to Bring More Machine Learning Into its Search Platform,” [], accessed November 10, 2018.

[5] Forbes, “No More of the Same,”

[], access November 11, 2018.

[6] Venture Beat, “Etsy Opens Machine Learning Center in Toronto,” [], accessed November 10, 2018.

[7] New York Times, “Inside the Revolution at Etsy,” [], accessed November 12, 2018.

(Word Count 755)


Amazon Studios’ experiment in crowdsourcing scripts: were they right to abandon open innovation in developing new content?


May the Salesforce be With You: A Rising Power in Machine Learning implementation

Student comments on ETSY – “Keeping Commerce Human” through Machine Learning

  1. I enjoyed the picture you painted of this company that is both “keeping commerce human” and forging ahead in its use of machine learning. I agree that Amazon Handmade will present a serious challenge to Etsy. The machine learning development steps that you mentioned Etsy is taking to fend off the competition make sense, though I wonder if they can seriously compete with Amazon in those capabilities (I would guess it’s a stretch). However, if they can continue to improve their algorithms and maintain brand strength (allowing consumers to channel their dismay at “sameness”), then I believe they will be well-positioned for the future.

    As Etsy looks to combat mass produced goods on the site like you mentioned, I think there are some lessons the company could take from Alibaba. While Alibaba embraced mass production, its anti-fraud efforts were similar in that they needed to sift through a mass of companies and products, and ensure that any outside of the desired type were removed.

  2. I am a big Etsy fan and loved reading this essay – it was excellent! I agree with your idea about using NLP to gain further customer data regarding product reviews. There is an interesting article about how this process can actually be applied: That said, one thing to consider is how much data can really be captured. For instance, I have purchased from Etsy many times, but I have never felt compelled to leave a review. Given this, I would encourage you to consider what changes would have to be made in order for this ML option to work. For instance, maybe before the customer can make their next purchase on Etsy, they will have to complete a review. Or perhaps Etsy could simply send emails after the product has been delivered prompting the customer to leave a review. Without some tactic to get customers to write reviews, I worry there may not be enough data to draw valuable insights.

Leave a comment