Alex Weinstein is Chief Digital Officer at Hungryroot, a personalized grocery delivery service. Before his current role at Hungryroot, Alex was a Senior VP of Growth at Grubhub and Director of Marketing Technologies at eBay. He began his career as a software engineer and holds a BS in computer science.
In our conversation, Alex talks about how Hungryroot has been able to drive industry-leading retention in a saturated field by focusing on differentiation, with AI as a key enabler. He shares his take on how “velocity of experimentation” serves as a predictor of where startups end up. Alex also discusses a key aspect of his role as a leader — creating an environment of intrinsic motivation.
Hungryroot is a personalized grocery delivery service. Last year, we sold $350 million worth of food. We’re profitable and growing 40 percent year-on-year.
Our differentiation is all about understanding our customers and the kind of food that they like to eat. We do this by asking them questions when they sign up, observing their behavior, and learning more about them. As they keep ordering from us, we get to know them better and better, which helps us save them time and money.
Many families tell us that they would otherwise spend many hours every week on achieving their health goals, but with Hungryroot, they only spend minutes and are able to eat in a healthier way.
We’ve been able to grow much faster than the rest of the grocery industry. We’ve achieved leading profitability and retention in a competitive space because of our focus on differentiation. There are many other ways to get food online, such as through Instacart, DoorDash, Kroger, Factor, and many others. Some of them are juggernauts that sell 100 times more food than we do.
Differentiation is absolutely key, and for us, it’s about saving customers time: meal planning, shopping, and cooking. We do this by understanding customers’ dietary’ preferences and health objectives. This allows us to completely reinvent the shopping experience — our customers start with a full shopping cart. We fill it with groceries and easy recipes that we know they will love.
AI is, of course, behind all of this, but AI is just a tool — it enables us to solve the customers’ problems. This is why you won’t see us talking much about AI on our home page or in our marketing. We instead focus on how we solve the customers’ problems.
Startups should continuously try to learn more about their customers. They shouldn’t be just executing based only on the knowledge they already have. Instead, they should be learning something new as often as possible through rigorous experimentation. By definition, a startup has not yet fully realized its potential yet.
We can contrast this with beloved legacy brands like McDonald’s that have been operating for decades. They’re now just trying to optimize their efficiency. They already know their customers and understand them deeply, so they really just need to execute their wishes.
This is why a startup’s velocity — particularly how quickly they learn important things about their customers — is a predictor of where they end up. And when I say learn important things, I don’t mean what color a button on the website should be. I’m referring to a true insight about users’ needs or possible solutions to those needs.
For example, something we learned at Hungryroot from our customer data and interviews is that many of our customers who love our products have families. At the time when we did this research, we didn’t carry many products for children. As a result, we’ve been experimenting with expanding our product offering to better serve families.
Fundamentally. Hungryroot understands enough about what customers want to be able to pre-fill their cart with food that they’ll love. If we haven’t done that, then we haven’t done our job. Our AI places items into customers’ carts and they have the ability to edit them before their box ships.
Today, two-thirds of what customers buy is selected by our algorithms. The rest is hand-picked by our customers as they customize their orders. We have been improving and still have more room to grow here, but I don’t think we’ll ever drive that down to zero. For example, if I was feeling under the weather this week, I might feel like having soup. AI can’t know that. However, having consumers think, “This sounds good. I know why they’re recommending this, but, I feel like something else today,” is a more realistic goal.
Existing customers will buy more and stick around for longer when you send them relevant communications. We’ve found that excellent ways to drive revenue per customer are by helping them figure out what they might try next, creating incentives for them to do so, and aiding them in buying or replenishing products.
One problem that I often see is marketing teams being held down by the technology they’re using. They get stuck sending what I lovingly call “batch and blast” emails — ad hoc campaigns. Instead, I would invite these folks to collaborate with engineering teams to invest in AI marketing technology that enables them to create event-driven, personalized communications. This will drive drastically better results than, say, blasting a July 4th email to every single customer.
On eBay, a meaningful chunk of items are sold in auctions — each with a bunch of bidders that didn’t win. These bidders have a clear purchase intent. If the platform just paid attention to that and told those buyers about other identical products that are listed for the same amount or less, those bidders would love it. It’s the most relevant communication ever! They, of course, would be highly likely to buy and stick around as a customer, which is the ultimate goal of these communications.
This approach is built on a deep understanding of what the customer was interested in, what they bought previously, and how much they were willing to pay. It requires a meaningful marketing and technology investment, which many companies do not make. I find that these are more than worth it.
Back in the day at eBay, we did $85 billion in sales every year and were able to add billions of dollars through these event-driven, ultra-personalized campaigns.
Besides my day job, I also do advisory work helping startups that reached product-market fit accelerate their growth. I worked with a half dozen teams over the past couple of years. I often find that it’s difficult to get product teams and marketing teams pulling in the same direction.
The root cause, crudely, is that product people are often convinced that they’re intellectually superior, and marketers often believe that product people are uninterested in business outcomes. There’s an aura of contempt on both sides.
So, a lot of what I do is effectively couples counseling — building trust by encouraging both sides to see the strengths of the other and the wisdom of the other side’s approaches. Then, I work to slowly take steps that both sides see as mutually beneficial, but, most importantly, are serving the customer. Small successes create a foundation to align a bit more and maybe work on a more audacious project together. That’s one recipe I found for getting the teams working more closely together.
The end result is worth it. The product actually does what the marketing message promises, customer acquisition cost will be lower, retention will improve, and the organization will be able to execute on audacious bets.
These conflicts are typically about what gets worked on — feature work versus technical debt and infrastructure. If a company has reached the stage where they’re worried about scalability and tech debt, it often is a good sign. It means that they’ve found product-market fit and are attempting to scale.
In this situation, my advice is to avoid the extremes. One extreme is engineering saying, “We can’t ship anything in the next three months because we need to rebuild a component from scratch. This extreme is less common — the market pressure to deliver customer value doesn’t allow for this.
The other extreme is never paying down technical debt. Symptoms of this extreme include constant interruptions to the engineering team and duct tape everywhere. Productivity plummets, which is subsequently followed by the CEO saying, “Why are our engineers so slow?”
Instead of these extremes, I advise regularly evaluating tech debt and infrastructure needs and handling these as a part of the regular roadmap work. In my opinion, it’s best to consistently dedicate 5–20 percent of the team’s time to make sure the future isn’t mortgaged. This, of course, requires a strong technical voice to be the counterweight for the business asks — and alignment of the whole team to operate against long-term goals.
Autonomy, Mastery, and Purpose is a framework created by Daniel Pink. You can read about it in his book, Drive. He did an amazing TED talk on this topic as well. I subscribe to this framework because I had incredible mentors early on in my career. They showed me how much more responsible for the outcome I felt when I had autonomy, the joy of getting better, and the pride of being part of an organization where you’re contributing to a bigger purpose. I felt the positive effect of these values firsthand, which inspired me to instill them in the teams that I lead today.
A prerequisite for this type of value system is, of course, hiring people who want to be a part of this kind of environment. If you hire folks who want to be trusted, want to get better, and want to be part of something that makes people’s lives better, you’ll create an environment of intrinsic motivation. I see my role as a leader to foster that.
One example that I’m proud of is the Hungryroot data science team. They are taking huge leaps forward with personalization algorithms. They’re taking bold bets with generative AI and recommender systems. If they want to lead the industry and be on the bleeding edge of research, they know that they have the autonomy to do so. They have the permission to take risks and make bold bets that may not work out. I’m providing them with the context and a framework to judge progress. If we make better recommendations, our customers will stick around.
I believe the concept of velocity of learning applies to people as well, not just organizations. People who take pride in being better today than they were yesterday learn faster, stay on top of best practices, and aren’t discouraged when previous patterns don’t work. This trait is especially important in a field like AI because it changes so rapidly.
When I interview candidates, I love to ask them about a time when they were wrong. Being wrong doesn’t necessarily mean that they failed, but that they were able to change their mind based on new knowledge and therefore widen the scope of potential outcomes. This is a great sign of a growth mindset.
LogRocket identifies friction points in the user experience so you can make informed decisions about product and design changes that must happen to hit your goals.
With LogRocket, you can understand the scope of the issues affecting your product and prioritize the changes that need to be made. LogRocket simplifies workflows by allowing Engineering, Product, UX, and Design teams to work from the same data as you, eliminating any confusion about what needs to be done.
Get your teams on the same page — try LogRocket today.
Want to get sent new PM Leadership Spotlights when they come out?
To help demystify stakeholder management, you can use tools that introduce a structured approach for your product team.
Role-definition frameworks like RACI, RAPID, and RASIC take a structured approach to assigning roles and clarifying accountability.
Jann Curtis talks about understanding your audience’s purpose and what they hope to get from the conversation.
Scaled agile is an approach that allows you to extend agile principles across multiple teams, projects, or business units.
Learn more →
