World’s largest supplier of beauty devices turns to AI to enhance customer care

Profile picture for user gflood By Gary Flood September 30, 2020
By upping its game around fraud protection, CurrentBody sees order approval rates increase to over 98% and chargeback rates go down to under 0.5%

Image of currentbody beauty products
(Image sourced via CurrentBody website)

UK-headquartered ecommerce specialist CurrentBody claims to extensively research the global beauty device market to offer shoppers on its website the best at-home beauty devices that harness in-clinic technologies and which also deliver safe, effective results.

And it seems to be delivering against that promise. Ten years on, it has brand partnerships with market-leading names like Tripollar, NuFACE and Smoothskin. It is also a global strategic partner for L'Oreal's Clarisonic range, the best-selling skin cleansing solution using sonic technology, has a multilingual network of devices and has recently signed a joint venture in China with Thakral, one of the biggest distributors of this kind of consumer goods in Asia-Pac.

Its problem, according to its Head of Trading, Lyn Carbine: the fraud protection provider it had been using was unable to keep up-and as a result, good business was being turned away needlessly.

We were founded to offer consumers around the world the best health and beauty technology for use in their own homes, and we will support those products and our customers with a team that have exceptional product knowledge. But we're also growing at an exceptional rate; we're moving into new countries, and our then supplier's technology had a hard time differentiating between legitimate orders with novel consumer behaviour and fraudulent orders.

It also couldn't provide visibility into the declined orders, which meant we were constantly reaching out to get more information. An agent could spend up to 20 hours a week manually reviewing declined orders to try to capture the sales we should have been making.

When as much as 7% of orders were being rejected, Carbine and her team decided enough was enough, especially as she suspected that more than 70% of the rejected transactions were actually good orders.

‘Our risk management strategy was affecting our customers'

Apart from reducing this error rate, she notes, another imperative was to ensure unauthorised resellers could still be blocked from buying products in bulk from the site. And ultimately, too strict a checking policy was having a negative effect on her market:

We knew our risk management strategy was affecting our customers because we heard from them. There are few experiences more unsatisfying for a legitimate customer than to have an order delayed due to the need for manual review. Here's a shopper who wants your product and you're saying, ‘Wait.' Those cases were damaging all the work we've done, and do, to make sure we provide the best experience in commerce.

Intriguingly, a possible answer emerged: Artificial Intelligence (AI).

We wanted a solution sophisticated enough to recognise legitimate orders, even when they included anomalies. We also wanted visibility into why an order was declined. We needed a vendor that had worked with international beauty brands that sold electronic devices and beauty technology and which could give us automated chargeback management services.

Another mission deliverable for any changed fraud detection solution, she adds, is that customer service would not be in any way affected.

We had to be confident our new approach would help us continue to offer the kind of customer service we're known for. If we could make faster decisions on shipping orders, we could get them to our customers more quickly, and if we could do a better job of recognising legitimate orders for what they were, we'd stop disappointing our legitimate customers by declining their orders.

To fix all her problems, Carbine has turned to a specialist ecommerce fraud protection supplier, Signifyd, which claims to employ big data, machine learning and expert manual review to solve the false positive issue CurrentBody was starting to be frustrated by.

Reasons this supplier won her business: it was able to prove it understood the importance of automating chargeback recovery, had a way to head off the unauthorised reseller issue and the importance of great customer experience.

So how's it working out? Carbine says that things have completely turned round, with approval rate up by 5%, from 93% to over 98%, while what had one time been seen internally as a "worrisome" chargeback rate has also dropped, from 0.5% to .04% Even better, the 20 hour maximum agent time spent on on manual reviews of orders is now down to more like one. Plus, she adds, CurrentBody now has the confidence to continue to expand into new countries, because the new fraud vendor has agreed to reimburse it for any costs that resulted from fraudulent orders that its software mistakenly approved.

That's one of the reasons the company has decided to launch a new online story in a completely new market, Russia, she states, and it's also recently gone live in three new countries. The time saved chasing possible bad orders has also meant, she says, customer service is also being enhanced:

The team has more time to go deeper with our customers and help them find the best devices for their personal needs and can explore the possibilities for the best ways to use the CurrentBody devices they already have. We are establishing our team members as product experts and product owners, as go-to advisors for questions about treatment times, attachments, and of course the products themselves.

We want to empower these team members to meet customers where they are, whether that's social channels, live chat, e-mail-however a customer wants to gather information.