What the future of AI-powered insights looks like for HelloFresh
With HelloFresh’s US Head of Marketing, Conor Feeney and Bounce’s Co-Founder & Co-CEO, Charlie Butler.
Insights leaders are facing an increasingly difficult decision when it comes to selecting partners. Deliver depth with legacy agencies, cut costs through DIY or chase speed with AI-native.
The pressure to do more with less is growing across industries right now. To make sure this drive for efficiency doesn’t compromise quality, it’s more important than ever to get the foundations of research right. That’s why HelloFresh’s US Head of Marketing, Conor Feeney is encouraging other insights and marketing leaders to experiment with and test AI use cases.
Conor took to the stage at IIEX North America, alongside our Co-Founder & Co-CEO, Charlie Butler, to share his experience adopting AI-powered insights. Here’s a recap.
‘Between a rock and a hard place’: Agency, DIY or AI-native
“Client-side researchers are faced with three pretty imperfect solutions right now,” shared Charlie.
“Some insights teams are relying on self-service tools, which are fast and cost-effective and were mandated into the organization 5 or 10 years ago. But, as insights budgets get smaller, research teams get smaller. Self-service tools are only as good as the expertise, the resource, and the time you have to put into them. So, there’s a growing gap there.”
“On the other hand, you have full-service agencies. These agencies are amazing, because they have some of the best researchers in the world, but their core competency is not in technology or AI. So, how can you rely on them to take you into the next generation of research? They would need to outsource your research to a technology partner, which is fragmented and doesn't feel like a recipe for success."
“Thirdly, you have AI-native companies. Fast and progressive, but they have no research context. No research expertise. They haven’t been designed by researchers. They’re just obsessed about the next frontier. About breaking things. I've met a lot of researchers over the last seven or eight years and we don't like to break things where we don’t need to."
“Democratising insights has to be done responsibly.” he adds. “I think one of the main reasons we haven’t truly achieved that goal is because software companies are obsessed with the quarter-inch drill, instead of the quarter-inch hole. With how amazing their technology is, not with the problem they’re actually trying to solve.”
.png)
‘Do more with less’: The importance of foundation
One of the biggest challenges insights teams face right now is the growing pressure to do more with less. To be faster and more efficient without compromising quality.
“We’ve never been able to get insights as fast and efficiently as we can now.” shared Conor. “But from my experience over the years, I’m seeing more and more manipulation and forced shaping of what the insight is or what it should be. So today, having the right foundation at the beginning to ensure you’re really thinking about the key consumer truth you want to go after is key. We need to be less internally focused and more externally focused on the consumer to take the right actions going forward.”
“I’ll give you two examples of how we’re using AI at HelloFresh. One recent example is our churn data. We were able to go back and look at our churn data over the last five years using AI to get a quick, comprehensive view of what’s driving churn in our business. This is something that would have taken months to do to piece it all together in the right way. Now, it’s just about optimising something that’s always going to be there and focusing on the key insights to build on that.”
“Secondly, something I’m more excited about. How do we leverage social sentiment data to move fast? Across our social channels, we’ve looked at what’s trending, have started pulling consumer comments and have built a live dashboard to see what’s working and how we can quickly pivot to focus on the right content formats and messages.”
.png)
‘Performance and progress’: The ideal insights solution
As more vendors enter the market with AI-led solutions, insights teams must be specific and steadfast about what they need and why. Conor shares his considerations when it comes to assessing the ideal insights solution.
“I view the perfect solution from three lenses,” Conor said. “One is the performance side. How do we spend more efficiently than we have in the past? How do we drive true incrementality for our business and cut out the noise in between? Secondly, there’s the consumer side, which is harder. What are the key truths that will enable us to cut through? Thirdly, it’s around the approach. How are companies approaching our challenges and partnering with each other?”
“The legacy approach is sticking with one tool and being overly restrictive when it comes to experimenting with AI. Honestly, those guys are going to lose. We need to move fast. We need to experiment quickly. We need to embrace what’s happening right now, otherwise we’re going to fall behind. I think it’s a great time for disruptors, and it’s going to be really challenging for legacy entities, depending on how they approach it.”
Charlie adds: “From my point of view as someone who’s building a business, speed, cost efficiency and expertise is an incredibly hard blend to find in a supplier. One of the reasons for this is something Conor touched on earlier. Internal over external. Not leading with the customer problem. When you ask a vendor what they do, they might say they’re a full-service agency. But that’s their business model, not what they do. They might say they’re a tool. But that’s something they’ve built. Again, it’s not what they do. What do they actually help people with? This is where many of the vendor challenges come from.”
.png)
‘Leave legacy behind’: Commit to being a champion
“People are still central to everything we do, but I can already see that AI-enabled talent is much better. On the flip side of that, AI-enabled laziness is still lazy. I’m sure everyone in this room has been subject to the proliferation of internal tools and the bureaucracy that comes with that,” said Conor.
“Businesses shouldn’t be afraid to put what they had before in the bin and really rethink it to get things right from the beginning. They shouldn’t be afraid to stop using legacy approaches. This is critical in terms of how we do things moving forward. That’s why we’re working with Bounce right now, and I hope it’s something we can continue to build on in the future.
.png)
‘Craft and context’: Automate the parts you don’t like
“To build on Conor’s point, be the champion of change,” said Charlie. “It’s not as scary as you think. Ask for help from your research partners on how you can be the person to drive change in your organization and identify the areas that need change. Don’t blindly mandate automation and AI in the areas of your job you actually love. When we speak to our customers, they love the start and the end of the research process. But they don’t want to do survey design, and ultimately, systems can probably do it better. So, we identified areas of the process they hated and focused on them first.”
“Now the change doesn’t have to happen through you, but someone is going to do it. It has to happen in the research team if it’s going to be done well, because only researchers know how to do it responsibly. Or else, someone outside of the research team is going to try to design a system, and they’re going to lose decades of ways of working. That’s one of the reasons research is so valuable to organizations at the moment, and this is something you can capitalise on.”
.png)
What is Bounce's Insights Operating System?
The Insights Operating System exists for insights leaders like Conor, who want to leverage AI to be faster and more efficient without compromising on quality.
It’s for insights leaders who use full-service agencies for large, in-depth studies, but also need a flexible solution to help them stay agile throughout the rest of the year.
In short, the Insights Operating System has two parts. First is the Retrieval Engine. Studies show that over a third of all the research done at an enterprise level is duplicative - a huge waste of budget and effort. The Retrieval Engine unifies all of your past research and enables you to ask questions of it in plain text to see if the answer you need already exists. No need to dig through slides, scan datasheets or rely on your memory.
.png)
“Knowledge management, like many DIY tools, failed, because it was like a Google Search for your information. It would simply tell you which of your reports mentioned ‘drinkers in Mexico’. But it didn’t give you the actual answer.” said Charlie.
“Only recently has it been possible to ingest structured and unstructured data and auto-tag it. We can integrate into your central data infrastructure and understand the context [behind your question], so instead of telling you where the answer might be, we can give you the answer.”
For the questions your past research can’t answer, the Insights Operating System runs a gap analysis and builds a custom study using the Research Engine, which is run and delivered at speed by your dedicated Bounce consultants.
.png)
“One of the critical components for an operating system is that it can be independent. For a lot of our clients, we’ll tell them the gaps in their knowledge and they’ll go off and run the research with a different partner.”
“We’re not saying we can do absolutely everything. When speed is important and the need is quant, we will design a survey ready to go live for you in a matter of hours. But if you need deep qual, you can take the brief and take it straight to your qual partner. You may not need to run any research at all, you may need to buy Nielsen data. The point here is independence. The clarity of what you know and what you don’t know.”
