The Operators: Supply Chain as the Competitive Advantage, with Rachel Levy
Rachel Levy helped take the company that invented the mattress in a box from $34 million to a billion dollars, ran Victrola through COVID, and was most recently COO of Brooklinen. At every one of them, supply chain was the competitive advantage. On knowing your cost to serve before you take the order, why all growth is not good growth, and the ten percent of orders that can cost as much as the other ninety.
By Kelly Breakstone Roth, Co-Founder & CEO of Prysmic · October 2026 · 9 min read
Rachel Levy started in supply chain when the first e-commerce orders arrived by fax and were typed into the system by hand. She spent seventeen years at Sleep Innovations, the New Jersey company that invented the mattress in a box, as it grew from $34 million to a billion dollars, entered every new Costco country as the retailer's first vendor, and was acquired three times by private equity. She has done business with China since 2002. She was Chief Operating Officer of Victrola, which by 2021 was selling more than a million turntables a year across North America, Latin America, and Europe, and ran its supply chain through COVID. And most recently she was Chief Operating Officer of Brooklinen.
Levy is analytical to the bone. Every claim she made in our conversation came with a number attached, and every number came with the mechanism behind it. As a COO she read about twenty reports every morning, in thirty minutes, before the day began, and she would tell any operator to do the same: know your numbers before anyone asks for them.
She is also clear-eyed about the paradox she has spent her career inside. After marketing, supply chain is the biggest set of numbers on a consumer P&L: cost of goods, freight, fulfillment, warehousing, and the inventory sitting on the balance sheet. It is also, usually, the last function to get budget for the tools to manage them.
Her answer to that paradox is the argument of this conversation. Supply chain becomes the competitive advantage when it knows the cost to serve before the business commits, when it catches small deviations before they compound, and when it keeps the pulse of the business every single day. She has built it that way three times.
Supply chain is the competitive advantage
You've said that at all three companies you worked for, supply chain was the competitive advantage. Most consumer brands treat it as the cost of doing business. What did it look like when it was the advantage instead?
The first was Sleep Innovations, which became Innocor. We invented the mattress in a box, and I helped grow that brand from $34 million to a billion dollars. Along the way the company was acquired three times, by three different private equity firms, and each time a big part of what they were buying was the operation: a supply chain that could grow the business and generate EBITDA at the same time.
We built our own manufacturing and distribution in the US, we sourced our covers, we designed and developed our own products, and we had goods produced in China. So we were shipping from the US around the world and from China around the world. I called it my Around the World Project, because as Costco entered every new country, we were their first vendor entering every market with them. That only works if you already know what it costs you to be there. When Costco wants to go into a new country, you can't say yes and figure out the freight later.
And we grew fast. The year I joined we were $34 million. The next year, $100 million. The year after, $200 million. We sold through e-commerce and more than 130 retailers, and sales was committing to launch brand new products every six months: source the covers, get them produced, get them on the shelf across all doors. Operations delivered that, and every single year we were also taking cost out. That's what I mean by competitive advantage. Sales can make promises the competition can't, because the operation behind them can keep them, and the business makes money while it grows.
After Sleep Innovations you went to Victrola, consumer electronics sold on three continents with all of the production in China. What changed when the product changed?
Electronics taught me so much. Before, I thought it was just the plugs that changed when you travel. It's different chipsets, different wiring, different testing and regulatory regimes for North America, Latin America, and Europe. All of our sourcing and production was in China, and we designed and developed the products ourselves.
You joined in 2019, so you ran that supply chain through COVID, when container rates went through the roof and most importers paid whatever the market asked. You didn't. How?
I've been doing business with China since 2002, and what that gives you is the relationships and an understanding of how the whole chain actually works: who the factories are, who the forwarders are, where the capacity is, how the money flows. When rates went crazy, the people paying the exorbitant prices were the ones buying transactionally, going to whoever had space that week. We had partners who had made money with us for years, and we sat down with them and worked it out. We never paid anything close to what the market was asking. That comes from twenty years of showing up, being in the factories, and treating people fairly when it's their turn to need something.
And nothing is ever going to be perfect. In supply chain, something goes wrong every single day. The advantage is how you pivot, how fast you pivot, how you deal with it, and how you communicate.
In supply chain, something goes wrong every single day. The advantage is how you pivot, how fast you pivot, and how you communicate.
All growth is not good growth
You've said companies should build their cost-to-serve model at $25 to $35 million, long before most of them do. Why that early, and what happens to the ones that wait?
Because that's when the channels start multiplying, and it's still small enough to get your arms around. The ones that wait get to $100 million, $200 million, and they don't know. They're either not profitable or barely profitable, a million or two of EBITDA, which is tiny for a business that size. Then the pressure to deliver results arrives, from the board, from investors, and they've been operating for years with a mindset that says all growth is good growth. At some point, all growth is not good growth.
Growth in a channel that loses money on every order is not good growth. Growth in a country where the freight eats the margin is not good growth. Growth in a customer segment that only buys when you pay to fly the product to them is not good growth. And you can't see any of it if you've never built the model, because on the P&L it all shows up as revenue, and the cost lands in three other buckets that nobody connects back to the sale. We had that discipline at Sleep Innovations long before we were big, and it's why we could grow that fast without losing control of the economics.
At some point, all growth is not good growth. And you can't see it if you've never built the model, because on the P&L it all shows up as revenue.
Why is cost to serve so hard to figure out? On paper it sounds like arithmetic.
Because the cost of serving a customer doesn't live in one place. The product cost is in one system. The freight is on the forwarder's invoice. Fulfillment is on the 3PL's bill. Returns show up in customer experience. Chargebacks come back on the retailer's remittance, and marketplace fees are netted out of the payout before you ever see them. Every one of those is managed by a different person in a different system, and sometimes the systems tell conflicting stories. Nobody's job is to connect them back to the order.
And there's an openness problem too. People don't always want you coming into their sandbox. So assembling the truth about what one order actually costs you takes real know-how, it takes a team that owns the question, and it changes every month as the channels and the rates move. That's why most companies never do it, and it's why the ones that do have an advantage nobody can see from the outside.
So what is the model? Walk us through what goes into it, and what it changes.
Start with inventory, because most companies have an inventory problem. They don't understand their inventory position or the cost of carrying it. Then it's everything it takes to get a unit to a customer and keep it there: the fulfillment cost, the shipping by zone and by mode, the returns, and the cost of the channel itself. Every channel has its own economics. Take Amazon. You have FBA, you have Seller Fulfilled Prime, you have multiple ways to do business with them, each with a different fee structure and a different inventory commitment, and you should model out every one of them before you choose. There's no tool today that lets you do that. You have to have the know-how.
Once you have the model, it changes the questions you can ask. Can I do this at all? Am I charging the right price? Am I set up correctly before the opportunity arrives, instead of after, when you look up and you're not making money? I was just talking to a company doing between one and two million dollars a month on TikTok, and they're barely making money, because they never built a model of the cost to serve that channel. The demand is there. The pricing and the setup were decided before anyone knew what a TikTok order actually costs to fulfill and to take back, so the growth arrived and the margin didn't.
That's the point I keep coming back to. Everyone is focused on solving the current problem: fix the inventory location, fix the carrier mix, fix the freight rate. All of that matters, and I've done all of it. But the biggest opportunity is the opportunity before it's a problem. If you know your cost to serve before you take the opportunity, you take the right opportunities, and you price them right.
Most companies run that analysis once, when someone on the board asks. You ran it as a monthly discipline, with a team behind it. How did it work in practice?
At Sleep Innovations we did it at the product level. I built a whole product profitability team, and their job was to keep the model true. When Costco came along, Costco has its freight factor, and I have to include these crazy numbers for shipping to Alaska and Hawaii, so that went into the model, and we knew not just the margin but how much EBITDA the opportunity would generate for the company. What's the size of it, what's the impact on the business. Every month we looked at the economics by channel, in our monthly package, for my own internal team. Every month, because the dynamics change. Amazon was interesting: at some point you have the choice of fewer locations or more, and the right answer moves.
Canada is the example I always give. We were selling into Canada, and it looked like growth. I built the model to show what each package actually cost us to get across the border, with the freight, the duties, the brokerage, and the returns, and every single package was losing money. So we literally turned it off. We stopped selling into Canada. Then we had our 3PL stand up a warehouse in Canada, so the product was already in the country and the cost per package came down, and we turned it back on and actually made money. Same customers, same demand. The difference was that the second time, we knew what it cost before we took the order.
We stopped selling into Canada. Then we stood up a warehouse there, turned it back on, and made money. The difference was that the second time, we knew what it cost before we took the order.
Ten percent can cost what the other ninety costs
Everything so far is structural: channels, countries, product lines. Most operators triage the same way, biggest channels first, biggest SKUs first, and they wave away the odd order. Where does that instinct go wrong?
It goes wrong because small in volume and small in cost are two different things. I also oversaw customer experience at Brooklinen, which includes fraud, returns, the whole pre- and post-purchase experience, and at one point I noticed a pattern of very large orders going to Hawaii, at certain times of year, over and over. My fraud instincts went off first, because it was an abnormal pattern. When we dug in, they were property managers furnishing vacation rentals, buying whole houses of bedding at once. So that's a real opportunity, and we moved them into our business-to-business channel, where the pricing and the service fit what they were.
But it also had a very high cost to serve, because Hawaii is air only. Large orders, large freight, and you're flying it. A handful of orders that looked great on the revenue line could cost as much to ship as hundreds of normal orders going by ground on the mainland. So then the question becomes, what are the other options? Can we consolidate, can we get freight over by ocean and hold it there, can we price it differently for that segment? Things like that are small in prioritization, and sometimes they're big in cost. It's the 90/10 rule the opposite way: ten percent of what you do can cost what the other ninety percent costs.
The same thing is true with retailers. If you're shipping to brick and mortar and something in your compliance is off, a label, a routing rule, a ship window, and you don't catch it on the first shipment, you keep shipping the same way and the chargebacks stack up, shipment after shipment. Amazon works the same way. One penalty is a rounding error. The same penalty repeated across every shipment for a quarter is a line on the P&L.
It's the 90/10 rule the opposite way: ten percent of what you do can cost what the other ninety percent costs.
Sometimes you're the one causing it
Freight forwarders, customs brokers, carriers, warehouses, suppliers: everyone in the chain sends you an invoice, each in its own format with its own charges, and checking them line by line is a job in itself. You've done it for twenty-five years. What have you learned?
The first lesson is that everything gets audited, because that's how you find the cost to take out, and that it is hard, because every provider bills differently and the money hides in the small lines. At Sleep Innovations, mattresses are big. You get 240 on a truck. So whether we were paying the freight or the retailer was, freight was one of the biggest numbers in the business, and we built the audit around it, outbound and inbound. When I started, the invoices were carbon copies, and you still had to watch every charge: the accessorials, the demurrage, the detention, the fuel. None of that has gone away. Twenty-five years later it all still exists, and there is a lot of room to get misbilled.
The lesson that surprises people is that sometimes you're the one causing it. So I'd go to the suppliers, to the warehouse, to the providers and ask: are we doing something that we need to fix? Because a lot of the time it's innocent. What was the recommended package size, and are we actually using it, or are we paying dimensional weight on air? Is someone on the dock doing something that triggers a charge? We never found huge anomalies, but when a deviation showed up, we wanted to know what was driving it. Was it one operator? Is it systematic?
And the lesson that matters most is that it's never the one-time occurrence that hurts you. If you ignore it, it's a big problem in three or four months. A small deviation you didn't chase turns into a very large deviation, and then you're sitting in front of the board and somebody asks why freight expense went up, and you don't have an answer. So yes, go after the cost takeout first. Then the real work is catching it in real time, so it never gets to compound.
It's never the one-time occurrence that hurts you. A small deviation you didn't chase turns into a very large deviation, and then somebody on the board asks why freight went up.
The last line to get a budget
Back to the paradox: the biggest numbers on the P&L after marketing, and the last function to get budget for tools. You managed the back-end technology spend yourself. From inside the job, what actually stands between an operations team and the technology it needs?
The barrier isn't fear of the technology. It's prioritization. Ops and supply chain are not the priority, sales and marketing are the focus and the push, and there isn't endless spend. So everything comes down to the trade-off, and I had to build an ROI business case for every tool we brought in. If I implement this, what does it deliver, what does it save, and how do we measure it? If your system costs $5,000, do I get my $5,000 back in one month, or three?
Then it's the team. You have people who are junior, people who are very advanced, and some who want to keep using their tools their way, and you have to convince them why this is going to help them versus doing it the way they've always done it. And anything that needs engineering resources is a different conversation entirely, because ops rarely gets prioritized for those.
So it isn't really about wanting a tool. It's about who owns it, and where it sits in the pecking order of importance to the business. Sales and marketing always have their tools. The question I've asked at every company is how supply chain gets its own.
It isn't really about wanting a tool. It's about who owns it, and where it sits in the pecking order of importance to the business.
Nothing told me what was moving
You said on a panel at Manifest: "You don't want people to go get that information. You want the technology to bring it to them, so you can spend time on solutioning." Before any of this was automated, how did you get the information, and what did it cost you to get it?
Business intelligence, first, always. I built it out at three companies. Getting all the data connected and accurate, fixing the inventory syncs, writing the right rules on allocation so the inventory you show is the inventory you have. Reporting is number one, because you cannot run a business on the reports that come out of the box.
And then every morning I looked at about twenty reports in thirty minutes. That's how you get the pulse of the business. It's one number off, somewhere, and you learn to see it. What I didn't have was a tool that told me this is going up, this is going down. I was the tool. I was the one doing the comparison against yesterday and last week and the same period last year, in my head, across twenty reports. What it cost was the first half hour of every day, and the fact that the person doing the pattern recognition for the company was its COO.
I was the tool. I was the one doing the comparison against yesterday and last week and last year, in my head, across twenty reports.
Why was that so hard to see with the tools you had? Where does the deductive reasoning break down?
Variation. If you're in cosmetics or apparel, your shipping profile is very consistent. It's always a bag, it's always this size, and when a cost moves you can usually point to why. All three of my companies sold large, bulky items, and the profile changes with every order. Someone buys a comforter today, and tomorrow they buy towels and a robe, and those two orders have completely different dimensional weight, different packaging, different carrier economics. So when the freight number moves, you have to look at the product mix, where the goods were going, which states, which mode of transport, which carrier, and there are too many variables for a person to isolate what is actually driving the cost. In a contained environment it's easier. In big, bulky goods it's very hard to get to the trend, and by the time you can see it in a monthly number, it has been running for weeks.
That's the piece I would want a machine to own. You don't want your best people going to get the information. You want the information to arrive with the deviation already flagged and the driver already isolated, so the human time goes to the decision.
In summation, supply chain leadership hinges on three pillars: empower your people, build resilient relationships, and stay relentlessly curious about your “why.”
