In 2004, a 35-year-old strategist walked into LEGO's boardroom in Billund and delivered a number that left everyone stunned. Around 94% of the company's products were making no money. Yes, 94%.
These were sets losing cash on every unit sold across almost the entire catalog, while the company shipped over a billion dollars' worth of product a year.
The strangest part is that nobody inside LEGO had known. Debt had climbed past $800 million, sales had dropped 26% in a single year, and a brand that had just been voted Toy of the Century sat weeks of bad decisions away from handing itself to its creditors.
The story usually gets told as a lesson about focus. The more useful version is a lesson about instruments.
LEGO scaled a business it could not measure
Through the late 1990s, LEGO tracked profit by country. Sales in Germany, sales in the US, margins by region were all reported cleanly. What the company did not track was profit by product.
Designers shipped sets with electric micro-motors and fiber optic lights that cost more to manufacture than they retailed for, so every sale burned cash. Mark Stafford, a LEGO fan who joined as a designer in those years, later described a business with no idea what most of its bricks cost or what individual sets earned.
So leadership watched top-line revenue grow roughly 17% between 1999 and 2002 and concluded the strategy was working. The revenue was real. The margin underneath it was guesswork.
If you run a startup with more than one SKU or pricing tier, blended numbers will hide the same problem from you. Aggregate metrics are where dying product lines go to stay invisible.
The licensing hit was the anesthetic
LEGO signed Lucasfilm in 1999 for its first licensed line. LEGO Star Wars beat internal sales forecasts by 500%. Harry Potter followed. BIONICLE landed. On paper, the innovation push was working.
Those wins created a second problem that took four years to surface. Licensed sets sold in movie years. Star Wars films arrived in 1999 and 2002. Harry Potter in 2001 and 2002. Then came 2003 and the first half of 2004, with no film from either franchise.
Demand collapsed while the cost base held firm because manufacturing at that scale is a fixed-cost business: above a certain volume you print money; below it you burn cash fast. LEGO fell below the line and started losing roughly a million dollars a day.
Two blockbuster licenses had been covering for a portfolio where almost nothing else worked. Worth sitting with if one channel or partnership carries your revenue chart.
The parts audit nobody wanted to run
Jørgen Vig Knudstorp became LEGO's first non-family CEO in 72 years and sent supply chain lead Bali Padda to count the bricks.
The audit found roughly 12,900 active pieces in production, and about 90% of them appeared in exactly one set. Some were near-duplicates, created by designers who had no way of knowing a similar part already existed. Each piece carried mold tooling costs, warehouse slots, supplier contracts, and handling overhead.
Knudstorp cut the library to about 6,500 elements. Designers argued the smaller palette would kill creativity, and five of seven senior manufacturing executives were let go in September 2005 for refusing to implement it.
The constraint became the product strategy. A designer working on a fire truck now built it from parts that appeared in dozens of other sets, which meant high-volume molds, better margins, and a system of play that actually connected across the range.
Sequencing was the actual masterclass
CFO Jesper Øvesen built product-level profitability tracking and set a rule every manager could repeat: any new product needed to target at least 13.5% return on sales to get approved. A cool idea that had no clear path to that number was put on hold.
Then came the phasing. LEGO named three stages and refused to run them in parallel: manage for cash through 2004 and 2005, manage for value from 2006 to 2008, and manage for growth from 2009.
This order is why LEGOLAND went to Merlin Entertainments in 2005 for around €375 million, with LEGO keeping a 30% stake. Great brand asset, terrible returns for a toy manufacturer, so LEGO sold it during the cash phase.
The follow-up is worth remembering. In February 2026, LEGO bought 29 Discovery Centres back from Merlin for roughly £215 million. Same asset class, different balance sheet.
Where the discipline landed
LEGO closed 2025 with revenue of DKK 83.5 billion, around $13 billion, up 12% year-over-year. Net profit hit DKK 16.7 billion, up 21%, on 868 new products, the most the company has ever released in a single year. Consumer sales grew 16% while the business took share at more than twice the rate of the wider toy market.
The company that nearly died from too many products now ships more products than ever, with visibility into what each one earns.
What you can take from this
A great comeback, isn't it? If you are a founder, here’s what you can learn from LEGO:
- Build product-level P&L before you need it. If you sell more than one thing, know the contribution margin for each. Pull your top ten SKUs, services, or plans this week and calculate what each actually earns after delivery cost.
- Watch your revenue concentration, not just your revenue. Map what share of last quarter came from your single largest source. Anything above 40% is a dependency worth actively diluting.
- Treat constraints as a margin lever. Fewer components, fewer plans, fewer integrations. Every unique element you add carries a cost that lives on your books long after the launch excitement fades.
- Give approval a number. LEGO used a 13.5% return on sales. Pick your own gate, write it down, and apply it to the next feature or product someone pitches you.
- Sequence your recovery. Stabilize cash, then fix margin, then chase growth. Founders who attempt all three at once usually run out of runway during step one.
Forward this to a founder who is scaling faster than their numbers.
