Target Spent $7 Billion in Canada to Learn Something the Third Question Would Have Told Them

Every week or two I run a research project to explore a business idea, and almost every time I learn it's a terrible, unprofitable one. The last one was digital courses for non-profit boards, because the director of a local history museum kept telling me how hard it was to get his well-meaning board to self-manage. Turns out the only thing board members will do less than pay for training is actually engage with training. That cost me $7.22 in API spend instead of months of building and testing.

I thought about that $7.22 while reading a new Harvard Business School working paper by Srikant Gokhale and Rajiv Lal, because it opens with Target's Canadian expansion collapsing in two years despite a $7 billion investment.

Same question. Nine orders of magnitude apart.

The failure was decided before the doors opened

Gokhale and Lal studied 20 global retailers drawn from a dataset of 250, and their conclusion is not the one the post-mortems reached. Every public account of these collapses blamed localisation: wrong store formats, wrong pricing, wrong product mix, wrong supply chain. The researchers argue the strategic error happened earlier, in a room, before any capital moved. They call it an absence of balanced adaptation – the discipline of identifying what must never change about your model, and adapting everything else to what the market structurally demands.

Home Depot is the clean case. It entered China in 2006 with the full US model, a model that, according to the paper, was generating over $70 billion in domestic revenue, and by 2012 all seven of its stores had closed. The interesting part is that Home Depot did adapt. It changed the store layout, the product mix, and the service model. What it could not change was the single thing its economics required: a consumer who wanted to do the work themselves. In China, homeowners hire contractors as a matter of course. No amount of localisation creates a behaviour the market never learned.

Compare Costco's first Shanghai store in August 2019, which closed early on opening day because, according to the study, 139,000 members had pre-registered and the car park gridlocked. Costco changed almost nothing – same membership fee, same bulk assortment, same warehouse format. That outcome was also settled years before, in Issaquah, when the company decided its model would never be modified for a market that couldn't take it intact.

Best Buy left the UK after nine years. Walmart left Germany after eight. The list is long enough that "they executed poorly" stops being a satisfying explanation.

The question nobody asks is whether the behaviour already exists

The paper reduces the pre-entry decision to three structural questions, and the third is the one that keeps getting skipped: does the required behaviour exist?

Not "can we teach it." Not "will it develop." Does it exist, now, in the people we're counting on. Home Depot committed to China without a credible answer, and according to the same research, Target committed $7 billion to Canada without one. Both were betting that conditions would emerge at the pace their format required, and both were wrong in the specific way that no operational excellence downstream can rescue.

There's a sentence in the study that should be uncomfortable for anyone who has sat in a roadmap review. The struggling retailers, the authors write, shared one diagnostic failure: none of them had a clear answer to why a consumer here would choose them over a local competitor who already knows this market.

That's not a retail question. That's the question.

After the money moves, you can't ask it honestly

Here's the part I find genuinely useful, and it's the reason this paper is about more than stores.

The researchers explain why these companies didn't course-correct once the evidence arrived, and the explanation isn't stupidity or denial. It's structural. Once capital is committed, the board has approved the investment, the leases are signed, and the supply chain is half-built, the line between what must hold and what must change "becomes almost impossible to draw honestly." Their phrasing on Home Depot is the sharpest version: people inside had noticed the DIY behaviour was absent, and the company stayed six years anyway, because the institutional cost of honesty under sunk-cost pressure exceeded the institutional cost of persisting.

Read that again with your own last big bet in mind.

The window in which the entry question can be asked cleanly is before the commitment, and it closes quietly. Nobody announces that it's shut. You just notice, somewhere in month eight, that raising it now would require someone senior to admit the question was never properly answered – and that the person who would have to admit it is in the room.

This is what I mean when I say a cheap yes leads to an expensive no. The yes was cheap because nobody made it pay for itself up front. The no arrives later, at full price, with a supply chain attached.

What to pre-commit, before you spend

The discipline that answers this is small and it is boring, which is probably why it doesn't happen. Before the money moves, write down what evidence would make you stop.

Not a risk register. Not a list of things to monitor. A specific, falsifiable statement – if by week six we have not found X, we are not doing this – agreed in advance by the people who would otherwise have to defend the decision later. The point isn't the document. The point is that you're purchasing the ability to change your mind at a moment when changing your mind is still cheap, and you're purchasing it from your future self, who will not want to sell.

Independent work on vertical expansion lands in the same place: most failures stem from insufficient pre-entry analysis, not bad execution. And the practice of trying to kill a market before you enter it exists precisely because the killing has to happen while it's still an argument rather than an asset.

One more number from the study, for calibration. Even among retailers that succeed internationally, few capture more than 10 percent of a foreign market – and the authors describe that ceiling as structural, not operational. That applies to the winners. It's worth knowing what the good outcome actually looks like before you model the great one.

This isn't about stores

If you run product at a post-PMF software company, swap "market" for "direction bet" and the whole thing holds.

You are deciding which segment to build for, or which adjacent problem to take on, or whether the enterprise tier is a real business or a story your board likes. The three questions are the same. What about our product must not change for this to work? What are we genuinely able to adapt? And does the behaviour we're counting on – the buying behaviour, the workflow change, the willingness to switch – already exist in these people today?

The third one is still the one that gets skipped, and software makes skipping it easier, because there's no lease to sign. The commitment accumulates in quieter ways: a hire, a partnership, a quarter of roadmap, a set of tickets written for a direction nobody formally chose. By the time it's expensive, it's also invisible.

I'm not going to pretend $7.22 of research would have saved the $7 billion that, according to Gokhale and Lal, Target put into Canada. Their problem was harder than mine and the number would have had more zeros. But the shape is the same, and the shape is what transfers: the cheapest moment to find out you're wrong is before anyone has to defend being right.

The question you can't afford to get wrong is worth answering while the answer still costs money instead of years.