The Price Was Wrong Before the Product Shipped

Design the product. Find the manufacturer. Build the prototype. Source it. Ship it. Then figure out what to charge.

This is how most physical products get priced. In a survey of 1,615 companies across 40 countries, Simon-Kucher & Partners found that roughly 80% do not discuss pricing until the product is nearly finished.¹ The price is the last question asked before launch. The engineering, the sourcing, the tooling, the first production run. All of it happens before anyone asks what the market will pay.

Seventy-two percent of those products fail to meet their revenue or profit targets.¹

The standard explanation is that these products were priced incorrectly. Too high for the market. Too low to sustain margin. Wrong for the channel. The explanation implies the pricing was a decision, and the decision was wrong.

After 25 years inside retail systems, I've come to a different conclusion. In most of the failures I've watched, there was no pricing decision. The price was a mathematical consequence of two numbers that were already fixed. The founder just hadn't calculated the gap between them yet.

The equation solved backward

In 1963, Toyota developed a practice called target costing. The Japanese term is genka kikaku. The method inverted the sequence that most companies follow when bringing a product to market.²

The standard sequence works like this:

The product comes first. The cost follows from the product. The price follows from the cost. This is cost-plus pricing. More than 80% of companies rely on it.³

Toyota's method ran in the opposite direction:

The price comes first. The cost is constrained by the price. The product is designed within the constraint. Robin Cooper and Regine Slagmulder, who studied the system across Japanese manufacturers, documented that as much as 90 to 95 percent of a product's costs are determined during the design phase.² Once design decisions are made, the cost structure is locked. Cost management must happen during design, not after production begins.

Toshiro Hiromoto described the philosophy in Harvard Business Review in 1988: "They design and build products that will meet the price required for market success, whether or not that price is supported by current manufacturing practices."⁴

What this looks like in practice

Porsche used this method to develop the Cayenne. They started by identifying which features customers would pay for and which they would not. Engineers proposed a six-speed racing transmission. Customer testing showed buyers would not pay a premium for it. It was removed. The Cayenne was designed around the price the market would bear. Within a decade, it accounted for half of Porsche's total profit.⁵

Fiat Chrysler took the opposite approach with a compact car in 2009. Executives publicly described "kicking out the finance guys" from development. The product was engineered first. The price was added last. Market disaster. Temporary layoffs.⁵

Both companies made cars. One started from the market and worked backward to the product. The other started from the product and discovered the market too late. The equation ran in different directions. The outcomes were not comparable.

The window closes

The reason the backward equation persists is not ignorance. It is timing.

A founder has an idea. They design the product. They choose materials, features, form factor. Each choice narrows the range of possible manufacturing costs, but none of them lock it permanently. The founder is still in the theoretical zone. Everything can still change.

Then the tooling order goes out.

A production injection mold for a typical consumer product averages around $12,000. Complex multi-cavity molds for high-volume production run $30,000 to $100,000 or more.⁶ Once the mold is cut, the design is physically encoded in steel. A change to the product means a new mold, new testing, new timeline, new capital. The decision to proceed is partially or completely irreversible, to use the language of Dixit and Pindyck's foundational work on investment under uncertainty.⁷ The optionality is gone.

Then comes the minimum order quantity. Five hundred to five thousand units for standard components. More for custom parts. The MOQ commits capital. Wire transfers go out, often 30 to 50 percent up front, for products that have never been sold.

Then come certifications. ETL, UL, FCC, CE. Ten thousand to fifty thousand dollars and eight to sixteen weeks. Mandatory for retail. Budgeted against a price that hasn't been validated.

In 2024, Humane launched its AI Pin at $699 plus a mandatory $24 monthly subscription. The device required custom hardware: a laser projector, camera array, proprietary processor. Fewer than 10,000 units sold. Returns outpaced new sales. At that volume, the per-unit manufacturing cost was enormous, and there was no path to the production scale that would have reduced it. The company was sold to HP for $116 million, roughly 14% of its peak valuation. HP did not acquire the product. All AI Pins were remotely deactivated on February 28, 2025.¹⁴ The tooling was committed. The COGS was locked. The market never arrived.

By the time a founder sits in the buyer meeting with a retailer, the cost of making the product is no longer a variable. It is a fixed number. The tooling is paid for. The first production run is manufactured or ordered. The certifications are scheduled. The founder has spent six to eighteen months and tens to hundreds of thousands of dollars.

The pricing "decision" now reduces to a single question: is the gap between what I spent and what the market will pay wide enough to survive?

The number nobody calculates

For every product at every retailer, a single number answers this question before the money is spent. It is the maximum allowable manufacturing cost. The most a founder can spend making the product and still survive the channel's economics.

The calculation is straightforward:

The inputs are discoverable. Retailer margins follow category patterns that are not published but are well understood by anyone who has operated in the channel. Costco takes approximately 14% on HOME products. Home Depot takes approximately 37%. Canadian Tire's combined structure takes approximately 38%.⁸ Deductions vary by retailer, typically 3% to 8%. Cost-to-serve can be estimated from industry ranges and existing operational data. These numbers are structural facts of the channel. They exist before the founder arrives.

Run the calculation for a product with a $100 shelf price across four channels. Same product. Same category. Same consumer price. Different economics:

A product that costs $55 to make is viable at Costco and dead at Canadian Tire. Not because of execution, marketing, or product quality. Because the arithmetic of the channel cannot support it. A product that costs $75 to make is dead everywhere except Costco, and even there the margin is thin.

The number does not move. The product must be engineered to fit inside it.

Most founders never calculate this number. They calculate their COGS first, then approach a retailer, then discover whether the gap is wide enough. By then, the tooling is paid for. The MOQ is committed. The answer to the question is already determined. The buyer meeting is not the moment the price gets set. It is the moment the founder discovers what was already true.

The double lock

The manufacturing side locks the cost. The retail side locks the price. The founder faces two fixed variables simultaneously.

In software, pricing is iterative. A SaaS company can adjust a pricing page, measure conversion, and iterate weekly. The cost of being wrong is the time to run the experiment. The cost of correction is close to zero.

Physical retail does not work this way.

In 2022, Diego Aparicio of IESE Business School and Duncan Simester of MIT Sloan published a study in Marketing Science using comprehensive US retail scanner data that examined what happens when a new product's initial price is wrong.⁹ They found that retailers face systematic price frictions: when the surrounding category hasn't had recent price movements, when physical labeling creates cost barriers to changes, when 99-cent price endings create demand-curve kinks that make adjustment costly.

Their finding: when a new product has weak initial sales, the retailer's response is not to experiment with a different price. It is to discontinue the product.

The initial price becomes the only price the product will ever have. If it is wrong, the product is deleted. There is no A/B test. There is no second attempt. One chance at one price.

This creates what I think of as the double lock. The founder cannot change the manufacturing cost. It is sunk in tooling, MOQs, and production runs. The retailer will not change the shelf price. It is held in place by category friction, labeling costs, and the economics of the shelf. If the gap between these two locked numbers does not support the channel's full cost structure, no adjustment can save the product. The economics were determined before the buyer meeting happened.

A study of 83,719 new consumer product SKUs across 31 categories found that 25% are no longer purchased within one year. That rises to approximately 40% within two years.¹⁰ Euromonitor's analysis of 54 FMCG categories across 32 countries found that nearly one-third of brands launched in 2022 had disappeared from the market by 2023.¹¹

These products do not die slowly. They die within one or two reorder cycles. The double lock leaves no room for correction. The window between launch and delisting is the time it takes the retailer to confirm that the initial velocity isn't there. It usually takes less than a year.

Allbirds: the channel changed, the cost didn't

Allbirds went public in November 2021 at $15 per share. A company valued at roughly $2 billion, built on sustainable footwear. Merino wool. Eucalyptus fiber. A proprietary foam called SweetFoam. The materials cost more per unit than conventional footwear. At DTC margins of 53%, the economics were sound.¹⁵

Then they expanded into retail stores and wholesale. Each new channel added a cost layer the COGS could not absorb. In a single year, gross margin fell from 53% to 44% as the retail footprint grew 60%. Revenue peaked and then declined. By late 2024, the stock had fallen below $1 per share. In January 2026, Allbirds closed all remaining full-price US stores. The product was the same shoe. The channel killed the economics.

Allbirds could not reduce their COGS without abandoning the sustainable materials that defined the brand. The wholesale and retail channels could not pay DTC-level margins. The gap was discovered after the stores were signed and the inventory was committed. The double lock in reverse.

When the retailer locks the door

In June 2025, Loblaw, Canada's largest grocer, delisted all Folgers coffee products. Parent company J.M. Smucker proposed a price increase driven by record-high commodity coffee costs and US tariff exposure on green coffee imports. Loblaw's category director called the proposed increases "significant and unjustified." The products were removed entirely.¹⁶

Loblaw did not negotiate a different price. Loblaw did not test a different shelf position. The retailer discontinued. The Aparicio and Simester finding, published in an academic journal using scanner data, played out in the largest grocery chain in Canada on a 175-year-old brand. The mechanism does not discriminate by brand age or company size.

The cascade

If the gap between COGS and the channel's maximum exists but is narrow, the cascade determines whether it survives.

I've written about this cost chain in the beliefs that kill products. Between the factory and the bank account, at least eleven categories of cost eat into the margin the founder thought they had. Freight. Warehousing. Insurance. Returns. Compliance. Customer service. For retail channels, the retailer's margin and deduction schedule sit on top of everything else. For brands using distributors, a third erosion layer takes another 20 to 35%.

In 1992, McKinsey studied this cascade and gave it a name: the price waterfall. Their analysis of a lighting manufacturer found that invoice prices ran 32.8% below list price. Off-invoice costs, the costs that don't appear on the invoice at all, added another 16.3 percentage points of erosion. The manufacturer's pocket price, the actual cash collected per unit, was roughly half the standard list price.¹²

In the same study, McKinsey found that a 1% improvement in realized price produced an 11.1% increase in operating profit. Three to four times more powerful than equivalent improvements in volume, variable cost, or fixed cost.¹² The leverage works both ways. A 1% error in the wrong direction destroys the same proportion of margin.

What the cascade looks like on a $100 product

Allbirds: the cascade at company scale

Allbirds is the cleanest public proof case of the cascade killing a company that had demand, brand loyalty, and a product people genuinely wanted to buy.

At DTC, Allbirds' cost structure worked. The proprietary materials (merino wool, eucalyptus fiber, SweetFoam) cost more per unit than conventional footwear. But selling direct at full margin, no retailer taking a cut, the gross margin was 53%. That 53% absorbed the material premium, the fulfillment cost, and the customer acquisition spend. The equation balanced.

Then the company expanded into retail stores and wholesale partnerships. Each channel added a new layer to the cascade. Store leases. Retail staff. Inventory duplication across locations. Wholesale partner margins of 30 to 50%. The COGS did not change. The materials were the same. The shoes were the same. But the cascade got deeper.

Nine percentage points of gross margin disappeared in a single year, the year the retail footprint grew 60%. Revenue peaked at $298 million and then fell for two consecutive years. By late 2024, the stock had declined 97% from its IPO price. In January 2026, Allbirds announced the closure of all remaining full-price US stores and a pivot to distributors.¹⁵

BMO Capital Markets confirmed the pattern in a 2024 study: DTC brands that expand into wholesale consistently face margin compression because the cost structure was designed for a channel that retains all the revenue.¹⁵ The moment revenue gets shared with a retailer, the cascade deepens. If the COGS was engineered for full-margin DTC, it often cannot survive the additional layers that retail demands.

The product never failed. The shoes were good. Customers wanted them. What failed was the gap between what the shoes cost to make and what the retail channel left after the cascade. The forward equation would have shown this before the first lease was signed. The Max Allowable COGS for wholesale, given Allbirds' material costs, did not support the expansion. Nobody ran that calculation until the stores were open and the losses were compounding quarterly in SEC filings.

What we built

I've been on the wrong side of this equation. We built a product, committed over $250,000 in capital, and discovered the economics too late. The failure was messier than any single cause. Wrong product for the channel, insufficient marketing, a market too small to sustain the volume we needed. But the pattern it revealed, the one I kept seeing across hundreds of other products afterward, was always clean: the equation was backward, and nobody ran the forward calculation before the money was spent.

The response was the same as the response to the cost chain failures I described in the beliefs piece. Not a framework or a set of principles. A calculation. A system that forces the question before the capital is committed.

The Margin Stack Calculator exists because I watched the backward equation produce the same failure in product after product. Its Section 6 is titled "What Would Fix This?" and its first output is the maximum allowable COGS: the most a founder can spend manufacturing their product and still generate profit at a specific retailer, after the full cascade. The formula is the target costing equation applied to retail:

Toyota arrived at this formula in 1963. The retail version produces the same structural insight: if the founder knows this number before committing to tooling, before signing the first PO, before wiring the deposit, the product can be designed to fit inside the channel's economics. The equation runs forward. The cost follows from the price, not the other way around.

Every question in the calculator was born from a specific omission. Every cost line was added because a founder encountered it for the first time after the purchase order was signed. The tool is not proprietary intelligence. It is the verification step that should exist in every product development process but almost never does.

McKinsey found that 80 to 90 percent of poorly chosen prices are too low, not too high.¹³ Founders leave money on the table because they anchor to COGS rather than to what the market would bear. More than 80% of companies price based on costs or competitive benchmarks rather than on the value the product delivers.³ The equation runs backward because the process runs backward. The failure is not intellectual. It is temporal, embedded in the order of operations.

The pattern

Every system I've written about operates on a belief nobody checks.

In Formula 1, the belief was about what the product is. It looked like a racing series. The economics said it was a franchise model disguised as a sport.

In retail, the belief was about whether the economics work. Founders held a margin number that had nothing to do with what ended up in their bank account.

In the Premier League, the belief was about what determines the outcome. It looked like a meritocracy. The data said it was a payroll receipt with a ten-month delay.

In pricing, the belief is about when the economics get decided. Founders believe they decide the price when they set the number. The price was decided when they committed the COGS. Everything after that is discovery.

I've watched this sequence play out across more than a thousand product programs. A founder designs a product they believe in. They find a manufacturer. They commit to tooling. They place the first order. They approach a buyer. The buyer tells them the margin structure. The founder runs the math for the first time. The gap is too narrow. The product is already made. The money is already spent. There is no path backward, and the retailer offers no path forward.

The verification step takes an hour. The Max Allowable COGS calculation requires a shelf price estimate, a retailer's margin structure, and a cost-to-serve estimate. None of these inputs are proprietary. All of them are discoverable before the first dollar of tooling is committed. Toyota knew this in 1963. Cooper and Slagmulder documented it in 1997. Every buyer at every major retailer knows the math before the meeting starts.

The founder is the only person in the room who hasn't run the calculation. Not because they couldn't. Because the process they followed told them the price comes last.

The price was wrong before the product shipped. Not because the wrong number was chosen. Because the right question was never asked.