Of all purchases now occur via recommendations

Feature adoption rate increase

Average onboarding time decreased

You're part of a the team at PacSun that runs their online store. You just got past the holidays and finally have a moment to breathe. Between new products, promos, and catalog updates you barely made it though this year. To top it all off, your manager is talking about some Salesforce artificial intelligence feature that she wants on the site.

Great.

You indulge your manager on a slow afternoon, and discover this thing recommends products to your customers automatically. It could save your team a ton a time but, it's really hard to understand.

Now what?

From the research sessions I conducted, we identified two main problems with the current experience.

Hard to use

Adding product recommendations to your online store is highly manual and technical process. Customers feel discouraged from trying them, missing out on the 7-16% sales lift they provide.

Untrustworthy

Recommendations are a black box. (We don't show any metrics to our customers to backup our claims of improved sales numbers)

As a result, people expressed difficulty trusting an unknown algorithm to play such a large role on their site.

The following designs shipped publicly to everyone across the world.

- Over the next 6 months, our team saw objectively better usability scores. - Adoption rate increased from 18% > 34% - Average onboarding time decreased from 7 days > 1 day

OK, so I lied....

One year later, product recommendations actually accounted for: ‍ 11% (not 10%)....Of all online orders submitted on Salesforce's eCommerce platform.