What Kellogg learned from Bear Naked's AI-powered e-commerce portal

The Bear Naked Custom Made Granola website is as polished as you'd expect of a premium brand from the Kellogg Company stable. But in one important respect it's still very much a work-in-progress — Kellogg's is constantly testing and reviewing what works well and what could work better, from the performance of individual custom ingredients to every aspect of the user experience.

For Chris Tutor, VP of Marketing at the Kellogg Company, the launch of the site was a departure from normal practice at Kellogg's, using an external design agency, Avionos, to build the site in less than four months:

We wanted to launch the e-commerce portal quickly. We worked with Avionos to take the idea from paper to end market in six months — the technology work took three to four months.

It launched as a minimum viable product. The way we approached the project was very startup-oriented.

Clearly, this is using the definition of minimum viable product popularized by British UX designer Paul Boag:

A minimal viable product doesn't mean a half finished. It means doing a small number of things to a high standard. pic.twitter.com/2rTfQEm1a1

— Paul Boag (@boagworld) May 5, 2015

Building Watson into Salesforce

The site was notable at its launch in April last year in being the first consumer brand to offer recipes inspired by IBM Chef Watson, a service from IBM's cognitive computing division. As a post on the IBM Watson blog explains:

Chef Watson, for example, read up on the chemical composition of hundreds of different ingredients and analyzed some 10,000 recipes from Bon Appétit. By combining that data and detecting certain patterns, Chef Watson has learned to suggest up to four different ingredients that blend together seamlessly.

The input from Watson was essential as the concept of the site was to cater to consumers wanting to combine new ingredients and textures in their granola — Bear Naked offers 50+ custom ingredients ranging from traditional nuts and dried fruit to more unexpected flavors and textures such as bourbon, lavender, coffee brittle and dried olives. As consumers select their ingredients, Chef Watson provides guidance on their picks, says Tutor:

The thought was that would help consumers work out which would work well together ...

We can identify if someone likes jalapeños, we can use that to suggest other ingredients.

In building the site, the Avionos team integrated the Chef Watson API into an e-commerce portal built on the Salesforce-native CloudCraze digital commerce platform, which significantly shortened the development timescales. Kellogg's IT department was actively involved in managing the project, says Tutor. A cross-functional team now manages the continuing evolution of the site, comprising Avionos designers and creative and media experts from Kellogg's.

Learning from experience

Since launch, learnings have been fed back into refining every aspect of the site, as Tutor explains:

We've learned across all dimensions, from the website experience to our marketing targeting.

We've added and removed ingredients based on performance — we've added bourbon and cut some coffee ingredients.

On the website, we've done standard A/B testing on big and small pieces of the UX. If people abandoned before checkout, we looked at what happened and made changes to improve the checkout experience.

On marketing, we're able to test various creative expositions and targeting to figure out what's working for us, so we can double down and market to the right folk with the right message.

The project has had a positive reception from consumers and they've been sharing experiences and recipes on social media, says Tutor.

Consumer reaction to this is pretty fervent. They want to share concoctions they came up with. This offering is ripe for a community. You can see it on social media.

In combining IBM Watson with a Salesforce infrastructure, the project is an early example of tie-ups that are set to become a lot more common after this week's partnership announcement between the two computing giants. Avionos President Scott Webb comments:

With Einstein and Watson joining forces, marketers can accelerate AI-driven experiences well beyond what traditional rules-based segmentation can accomplish. We saw some of this in our work with the Kellogg’s Bear Naked brand.

Combining the customer insights of Einstein with the cognitive computing power of Watson will allow companies to provide a level of personalization that has typically been difficult to achieve.

My take

This is a textbook example of how cloud platforms and services help to accelerate ideas to market — and an interesting demonstration of the use of digital technology to deliver customized products at scale. AI plays a crucial but very specific role here in informing the selection of ingredient choices — a pragmatic use of the technology.

I'm almost willing Kellogg's to take this to the next level by adding a community element to the site, so that this eclectic band of granola fans can more easily exchange ideas and perhaps participate in suggesting and choosing new ingredients. It would be an obvious next step — provided the business case stacks up.