Shopping for nail polish is usually an impulse purchase. Customers browsing OPI’s selection in an Ulta or Walgreens typically just select the color that best catches their eye, then carry on, according to OPI’s global director of digital products Violaine Muras.

To best translate that experience online, where every OPI shade from I Just Can’t Cope-acabana (a bold yellow) to Lincoln Park After Dark (a jet black) is available, the nail polish brand recently redesigned its website. The new is built around a personalized experience, with data algorithms tracking factors like purchase and browsing history, skin tone, color preference and recommended products based on similar customer and social media behavior. OPI has also now separated its online site into two customer journeys, depending on whether a shopper is a consumer (about 80 percent of traffic) or a professional nail technician (about 20 percent of traffic).

Finally, to make the checkout and online purchase process as simplified as possible, OPI has linked its online shopping cart directly to Amazon. When customers go to buy something from the OPI site, they’ll be sent to Amazon to check out.

“This time last year, we had an outdated website that was running on an 8-year-old design. As a color-centric brand, it didn’t show what we had to offer,” said Muras. “We needed to build a layer of customer data, so we used a combination of search behavior, social media insight and the Amazon integration to create a smarter set of algorithms.”

Screen Shot 2018-05-25 at 11.14.20 AM’s Amazon integration

OPI partnered with Qubit to build out two data sets based on the behavior of regular consumers and professionals. According to Muras, the two groups shop very differently: Professionals, on behalf of nail salons, buy in bulk and need to make sure they have the most popular OPI shades, like the sheer pink Bubble Bath and Big Apple Red, in stock at all times. Newness is important, too, so they’ll often purchase a full set of a limited-edition shade collection. Customers, however, are trickier to track: They buy based on individual mood, skin tone, color palette preferences, cult favorites and limited-edition availability.

To personalize the experience for that customer, OPI is building in functions that act on things like cart abandonment, search history and similar customer behavior within a shopper’s local region. (People in L.A. are more likely to go for a vibrant Princesses Rule! pink, while New Yorkers veer toward more classic shades like the burgundy I’m Not Really a Waitress.) This information influences the way OPI speaks to its customers in email marketing and social targeting: If a customer typically adds red shades to their cart, they’ll get an email notifying them that a new limited-edition red, Tell Me About It Stud, is part of its “Grease”-inspired collection, for instance.

“With a more advanced platform comes more progressed personalization,” said Nick Smyth, Qubit’s vp of sales in North America, who worked with OPI on the relaunch. “It all filters into programmatic product recommendations, social proof, mobile product discovery — things that push the customer to purchase.”

The catch, of course, is that all of the effort OPI has put into improving its online experience doesn’t translate into direct sales for the company. According to Muras, the Amazon integration was designed to offer the best customer experience. The Coty-owned company isn’t in a position where it can offer things like free shipping and returns, and it realized that customers aren’t likely to buy a single $12 bottle of nail polish online if they have to pay for it to get to them. The customer is already shopping on Amazon, particularly for beauty, so the goal is to push sales where the customer wants to make them.

The move gives up vital customer ownership, but it’s also led to better insight on what’s happening on Amazon. Amazon, of course, isn’t known to share data around what customers are shopping and searching for with its brands. By funneling customers to Amazon, OPI is building insight of its own, by connecting what’s being searched for and browsed with what’s eventually added to the Amazon cart. OPI and Qubit built a metric to represent Amazon conversion as true as possible, based off of add-to-cart and overall Amazon sales.

“If you check out with Amazon, that gives us valid data for what’s happening there, which we otherwise wouldn’t have,” said Muras. “We have more information on what’s happening on Amazon than ever before: [We know] what people are looking for and buying, and we can reconcile what’s happening on our site with what’s happening on Amazon. But people go to our site for something different. They want to experience the brand, and so personalization is huge for us.”