HT Talks Tech: How Noodles & Co. Streamlined Product Recommendations

The brand's former recommendation engine required time-consuming manual updates which meant recommendations for menu item add-ons were left unchanged for long periods of time.
Anna Wolfe
Senior Editor, Restaurants
wolfe
Noodes  digital ordering recommendation engine

Noodles & Co. was spending hours manually updating and configuring product recommendations for digital customers. However, these recommendations were not rooted in data or user behavior, leaving potential revenue on the table.​​ Noodles & Co. was eager to embrace  a technology solution that uses machine learning to provide smart solutions and save time. HT caught up with Tom Finley, Director of Digital, Paid Media & 3rd Party, Noodles & Co. for the story behind the tech transformation in this vital area.

HT: Tell us how long have you been with Noodles and what roles have you worked in?

Tom Finley, Director of Digital, Paid Media & 3rd Party, Noodles & Co.: 

I have been at Noodles since January of 2022, where initially I came on as the Director of Digital, which encompassed ownership of our web and app experiences. Additionally, Paid Media and Third Party Delivery responsibilities were shifted under me in July of 2023. Before that I’ve held roles in digital strategy, product management, eCommerce and digital advertising at Vail Resorts, Southwest Airlines and VML.

HT: As Director of Digital at Noodles, what’s under your purview?

Finley: My team is responsible for the guest experience and conversion of our owned digital channels including web and app.  We’ve been iteratively updating the web and app experiences for the past two years with the help of our partners at Bounteous. As you know, no digital experience is ever done so we are continuously rolling out new features and enhancements on a monthly agile cadence. Additionally, my team is responsible for driving traffic and revenue via paid media efforts and our third party delivery partner strategy.

HT:  Tell us about Noodle's former recommendation engine that required manual updates and the problems associated with it.

Finley:  The existing recommendation engine when I joined required us to manually update our product recommendations. We found ourselves constrained both by the structure of our menu and the time it took to maintain the configuration. It was so cumbersome that we didn’t have bandwidth to continually keep up with it. We fell into a trap of mostly considering what we would think of as logical add-ons from certain categories, like suggesting a dessert or drink if you have an entree in your cart, using our gut feeling, rather than taking a data-driven approach. We also weren’t easily able to experiment with other thoughts of how guests build an order and pair items from different menu categories to drive KPIs like growing cart size and PPA. The manual and time-consuming nature of the updates and configuration meant we kept the recommendations similar for long periods of time, which made the suggestions staler, and updating it was something that fell low in priority as we re-configured the menu for new items.

HT: Why make a switch?

Finley: Over the past two years we’ve made significant strides in cleansing our digital analytics data and applying more rigor to data-driven decision making. As we’ve gained insight into our top to bottom conversion funnel, we’ve identified where our leaks in the bucket are.  We found opportunities to both drive conversion but also delight our guests by making the experience easier and anticipating their needs at each step. Bounteous helped us identify the product recommender solution as a way to flex their team’s muscles in AI and help us drive revenue at the same time.

HT: What were you looking to do?

Finley:  We wanted to ensure that our conversion remained strong while growing revenue and recommendation engagement/attachment.  We also wanted to make the management of our recommendations easier and take that manual effort off the team. Let the data do the decision-making.

HT: What are you are able to do now?  Would you please give us an example?

Finley: One interesting development since we launched the NomNom Recommendation Engine from Bounteous is that we’ve started to see more of our menu being suggested to guests beyond our standard ‘add-on’ categories. For guests, this is often a way to introduce or bring more visibility to other entrees they may not be familiar with, but that we see other guests with similar behaviors are ordering. Our menu is full of variety across several different cuisines, so this has allowed us to better showcase all that we offer. We’ve seen that this not only increases engagement with the ‘You May Also Like’ section in the cart, but also has led to overall conversion rate increase ... Primarily, our machine learning models are trained based on online ordering data from our partner Olo. We’ve also experimented with examining individual purchase preferences and order history. In the future, we hope to also incorporate other data points – things like daypart, weather, and so on.

HT: What infrastructure was needed to be able to support these updates?

Finley: It (The NomNom Recommendation Engine) is fairly lightweight – we leverage a lot of the infrastructure we’d already be using to power mobile and online ordering for brands that leverage NomNom, which in addition to powering cross-sell recommendations is a much broader experience platform for dining and convenience store brands. So, we already have things in place like the NomNom API, and we use a modified version of that to feed the intelligent cross-sell recommendations into the Noodles ordering site.

HT: In what other ways is Noodles using AI?

Finley: I guess it depends on your definition of AI as nearly all the technology we use in Marketing incorporates automation in a variety of ways. We are using several deep-learning models through our data and Insights team to analyze business metrics and testing predictive models for sales, purchase behavior and media audience targeting. We’ve integrated our new CDP into our go-to-market process and are using the insights generated from it to inform our rewards strategy and CRM communications. Our Guest Services team is also using it to better respond to feedback and streamline the process for refunds.

HT What’s on the horizon technology-wise, marketing wise, that you can tell us about?

Finley: Personalization will be more ubiquitous across our guest communications in 2024. As I mentioned, with our CDP integrated into our Marketing process, we have an abundance of guest data to get smarter with who we target, what offers engage them and what channel to reach them with. Ultimately this will feed into a more personalized digital experience on our website and especially our app where we see a vast majority of authenticated Rewards users. We’ve also recently completed the installation of digital menu boards in all our corporate restaurants and have been testing and learning how to utilize the agility they afford us to better present our menu to guests and promote limited time and special offers. Additionally, we have a guest service chatbot that will be launching soon to automate remediation of guest issues.

X
This ad will auto-close in 10 seconds