Search engines on large e-commerce sites often spit up a lot of results, which is great if you want to browse, but otherwise annoying. Twiggle thinks it has the solution. Internet conglomerate Naspers agrees, because it has led the Tel Aviv-based startup’s $12.5 million Series A. Yahoo Japan, State of Mind Ventures (a returning investor), and Sir Ronald Cohen also participated in the round.
Twiggle, which has raised $14.7 million so far, will launch on several e-commerce sites in August. It hasn’t revealed who its clients are yet, but they will include other Naspers portfolio companies. Some of the group’s most notable e-commerce investments around the world include Flipkart in India, Argentina’s Avenida, and Filipino marketplace OLX. Yahoo Japan also runs some of the largest Japanese e-commerce businesses.
Twiggle was co-founded in 2013 by chief executive officer Amir Konigsberg, who was one of Google’s first employees in Israel before becoming managing director of price comparison site MySupermarket.com, and chief technology officer Adi Avidor, formerly a lead software engineer at Google Israel. Its search engine combines natural language processing, data science, and artificial intelligence.
Better search results of course, means that e-commerce customers are more likely to make a purchase instead of getting frustrated and turning to a competitor. Current search engines on many e-commerce platforms, including Amazon, require shoppers to filter their queries by criteria. For complicated purchases like electronics or home appliances, however, a lot of criteria sometimes returns too many irrelevant results instead of successfully narrowing down results.
Twiggle’s founders want to improve on current e-commerce search engines by letting people enter what they want in normal sentences, even if those sentences are unwieldy run-ons. For example, instead of typing “double door white refrigerator” and then filtering results, they can write “I want a two-door fridge that doesn’t make too much noise which is also white, which is new, has great reviews, and is a good value for the money,” Konigsberg says. Twiggle then analyzes that sentence for keywords and decides what kind of technical criteria it needs to find in product descriptions (for example, the types of motors that ensure a fridge isn’t too noisy).
For shoppers who don’t know what they want, Twiggle also provides context about different types of products so they know what to look for. That means a search for a smartphone will return specs about important features, like internal memory and camera megapixels, and then rank it against competing products. Konigsberg says Twiggle’s results will be based on information gathered from the entire web, not whatever e-commerce site it is running on.
“We want to treat search as a way to give people answers to questions and give them timely information. We don’t want to send them to a product page and then have to find the information themselves,” says Konigsberg. “If they ask for a quiet laptop, then we want to rank results by quietness, instead of creating more confusion. We look at the Internet at large and create a very large knowledge graph with use cases of products and the relations between them.”
Google already takes a similar approach for its Quick Answer Boxes, or cards with data and graphs that pop up when you enter a search term like “how many people are in the U.S.” Twiggle’s goal is to do the same thing for e-commerce searches. The startup will use its Series A capital to hire more engineers and integrate its search engine into e-commerce sites. After that, it will build an ecosystem that will include mobile and voice-to-text search products.