How to reduce a pineapple's carbon footprint by >20%

Aligning the value chain around the best knowledge of the consumer

A thoughtful industry veteran commented on the last post saying that much of the supply chain disconnect stems from the fact that retailers and suppliers see themselves at opposite sides of the negotiating table. He argued that:

You need to prove to both sides that collaborating and coordinating results in less waste and more economic gain than maintaining the current cat-and-mouse dynamic.

“What’s the catalyst though?” he asked.

Here, through the lens of one last mystery about pineapples,I argue that there are opportunities for mutual gain hiding in plain sight, and that a better understanding of the consumer can unlock them.

I describe a suite of products I built that can help generate the necessary knowledge 10x faster in the produce section, which could allow the industry to test ideas and hone in on the best ones more effectively. Let’s go!

The Mystery

Would you buy these pineapples?

Down in Costa Rica we uncovered that if you cut the tops off of pineapples before shipping them, you slash their carbon footprint by nearly a third. Without the leaves, known as the “crown,” you can fit ~25% more pineapples in each cardboard shipping box, which drastically increases transportation efficiency, and leads to big benefits for all stakeholders. It is estimated that by switching to “crownless” fruit:

• Suppliers would use 15% less boxes & see a ~50% increase in gross profit margin. For every 10 million dollars in revenue, they would save about $500k. They would also be able to sell more of their harvest as fresh fruit, as pineapples with unsightly crowns get diverted to juice at a lower price.

• Retailers would get a ~25% increase in revenue per box of pineapples and an easy-to-understand sustainability story to brag about, not to mention space savings at distribution centers and easier packing for grocery delivery.

• Consumers, who tend to rip off the crown as soon as they are ready to eat the fruit anyway, might see lower prices for similar quality fruit, and could feel good knowing that instead of ending up in their trashcan, the crowns were being used to replant new pineapples, as animal feed, or to make vegan leather.  There is also a convenience factor, similar to how most people choose to buy carrots without the tops, (even though carrot tops are edible).

There are no major technical barriers either. Today, the small percentage of fruit meant for “fresh cut” — the plastic containers of pre-cut pineapple you see in grocery stores — is already shipped crownless.

But in most cases, the industry continues to pay through the nose to ship millions of pineapple crowns from Costa Rica – where in aggregate they are a useful material – across the ocean, where they become useless. 

Why has the industry failed to capitalize on this opportunity which seemingly benefits everyone involved? Why are the vast majority of pineapples still shipped with crowns? 

We asked the owner of the pineapple farm, who told us that “he would love to ship crownless fruit, but the retailers weren’t asking for it.” He was also worried that the retailer would eat into his improved margins if he did.

We then asked the retailer – an experienced pineapple buyer at a large European supermarket chain. He told us that it seemed like a bad idea. “It’s a visual thing,” he told us. “We want to be known for having the best fruit. Nobody will buy the crownless ones, and waste might even go up. We don’t think consumers will go for it.”  

Was that true though? We had already seen a lack of feedback lead to big disconnects across the supply chain. If the retailer was wrong, and consumers would in fact accept crownless fruit at the supermarket, then we were sitting on a gold mine. The dynamic was the following:

Little to no data → Retailer believes consumers don’t want crownless → Suppliers won’t ship crownless → Potential missed opportunity for industry and planet.

The investigation

I needed to find out how consumers would really respond. To do so, I would have to overcome some obstacles. First, we couldn’t just ask consumers whether or not they would buy crownless pineapples.  A massive body of research shows that what people say they will do is different than what they actually do, and that this is exacerbated when the purchase in question is impulsive and low value. To quote Teresa Torres, people might say that they want to eat healthier, but that doesn’t mean you should open up a salad restaurant next to a McDonalds.

We needed to move past stated behavior to measure actual buying behavior, which presented its own set of challenges. As product managers we’re used to the digital world, where more often than not our problem is how to make sense of a massive pile of data. The physical world of grocery is different; there often isn’t enough data, and nowhere is this more true than in the fresh produce section. This problem is acute enough that companies like Afresh have raised hundreds of millions of dollars to develop a new operating system for fresh, which adapt to the data gaps and errors inherent in this dynamic environment.

When we do get data, it is often messy and incredibly noisy, which can make measuring small differences with statistical rigor tough, slow, and expensive. A classic approach to answer our pineapple question would have been to send crownless pineapples to 100 stores, and normal pineapples to a separate set of 100 stores for 3 months, and measure the sales between groups. This sounds good in theory and can work well for consumer packaged goods (CPGs), for example something like a new flavor of Cheerios, where you can guarantee consistency across boxes.  

With fruit though, this approach is often intractable. The inherent variability in fruit quality, the lack of traceability across the supply chain, and the inability to discount stock-outs and other irregularities can make the data so noisy that any signal is undetectable.  On any given day, the fruit in one set of stores might be three days older, a slightly different size or color, from a different supplier, or displayed in boxes instead of bins at the front of the store, all of which can have a drastic impact on sales. 

It’s impossible to control for many of these variables, and to complicate matters further, you’re usually looking for a small effect size. A jump in sales of just 3% can represent millions of dollars in revenue, but finding this small signal in the noise of the retail environment can be a bit like searching for a needle in a haystack. 

For this reason, many such tests — which can cost $100k to execute and require extreme levels of coordination from the supplier and retailer — often lead to inconclusive results.

I needed another approach. My budget was $10k and my timeframe was ASAP. What was the fastest and cheapest possible way of validating the assumption that consumers would accept crownless pineapples? I didn’t need conclusive results, I just needed a strong directional signal that would justify further investigation. 

Taking inspiration from the famous Wizard of Oz MVP popularized by Zappos, I designed a different type of A/B test. Instead of sending 100 stores crowned pineapples, and a separate 100 stores crownless pineapples, we would place both types of pineapples next to each other in the same store. In doing so, we would avoid the near impossible task of coordinating a new experimental supply chain with 200 stores, which itself could take months. Instead, our test store could order pineapples from their normal supplier, and we could just remove the crowns in person.  

Within two weeks we had organized the test with a fantastic independent supermarket, and were off to the races.  The produce manager was clear that he was betting against the crownless fruit, but was open to trying it. 

This setup drastically improved our signal to noise ratio. By focusing on a single store, we were able eliminate most uncontrolled variables, and by presenting each consumer with a choice – crowned or crownless – we were able to drastically increase the likelihood of seeing a strong preference, if there was one. This type of experiment, known as a “paired comparison” experiment, allowed us to measure, through direct behavior, consumer preference an order of magnitude faster than the traditional approach. 

Instead of needing to sell hundreds of thousands of pineapples to maybe get a useful signal, our data scientists helped validate that we could get statistically significant results after selling just a few hundred fruits, which allowed us to draw initial conclusions after just one or two days. 

The results

Here’s what we saw. Over the first few days, with no marketing, and with both groups at the same price, the crowned pineapples did have a slight edge, with about 55 crowned sold for every 45 crownless. The crownless fruit was doing better than the retailer expected, but not well enough.

But the story changed when we started playing with marketing and pricing. When we added a small sign highlighting the sustainability benefits, and dropped the crownless price by 10 cents per pineapple, crownless pulled ahead, with 60 crownless sold for every 40 crowned! Emboldened by this result, over the next week we carefully tested a handful of different messages and price points, and fed them into our statistical models. 

Our final results were surprising and encouraging. In the end, we saw that if we dropped the price of the crownless fruit by a marginal 2 cents — well within the savings afforded by removing the crowns in the first place — consumers bought just as many crownless pineapples as crowned ones!   

The retailer we had been testing with was impressed! He agreed that crownless pineapples were worth a try at a larger scale, and after seeing our findings, several sustainability-minded supermarket chains in Europe made plans to trial them as well. In under two weeks, by getting into the field and measuring behavior in a statistically robust way, we’d helped our retailer better understand his own consumers and proven that crownless pineapples could work.

We were seeing the beginnings of a new level of alignment among supply chain actors, built around the improved knowledge of the consumer that we had helped generate. 

The old dynamic:

Little to no data → Retailer believes consumers don’t want crownless → Suppliers won’t ship crownless → Potential missed opportunity for industry and planet.

Was starting to shift to: 

Compelling directional data → Retailer has seen consumers will accept crownless → Asks suppliers for them → Industry and planet both win. 

The scale up

I wish I could end this story by claiming victory, and telling you that ever since we ran this test, crownless pineapples took off like wildfire.

Instead, it was the demand for this sort of testing that exploded. Having seen how much more quickly we could validate key assumptions, leaders from across the business clamored for their questions to be answered.

This discovery work became my full time job. I found myself moonlighting as a produce clerk, building software from scratch to enable testing at dozens of stores at once, filling out a small team to support these efforts, and getting buy-in from senior leadership for a multi-million dollar scale-up (more on this next post). 

We weren’t retailers, and we didn’t own any of the fruit, so the experiments we ran were limited to our own business interests. But, as we ramped up testing and began to reap the rewards of shifting our process away from big, expensive, unvalidated bets and towards an agile discovery process, I couldn’t help but wonder how much this new approach to learning could help our supply chain partners align around what their shoppers were really looking for.

Final thoughts

Let’s return to a shopper standing in front of the shelf and choosing between a crowned or crownless pineapple. In retail marketing, this moment is called the “moment of truth,” and it dictates the rules for the rest of the supply chain. If our shopper chooses crownless, a pineapple farmer somewhere will get an order for crownless fruit. Otherwise, the crowns are staying on. 

I am struck by how much of a black box the moment of truth continues to be in the produce section.  Millions of shoppers make choices every day, and the nuances of most of them go un-captured. Beyond crownless pineapples, what other opportunities are hiding in plain sight that could benefit business and the planet? Would consumers accept off-size avocados? Blemished peaches? Crooked carrots? How many of our beliefs about what consumers want and don’t want are rooted in data, and how many in dogma and gut feeling? Would better tools open the door to answering these questions?

We barely scratched the surface here. There’s an opportunity for technology companies that engage with the moment of truth to not only help facilitate operations within the retailer, but also become a source of some of the data that the supply chain lacks today. A company like Afresh, for example, could become not just a new operating system for fresh, but also the new Nielsen of fresh, enabling retailers and other supply chain actors to run experiments, measure consumer preference at scale, and make more informed decisions.

By finding ways to step inside the black box and develop a nuanced, data-driven view of the consumer, we can nudge retailers, suppliers, and the rest of the supply chain into alignment around a shared goal.  When we – as food system technology providers – are the owners of this knowledge, it puts us in a position of power.