Strategy · June 2026

Growth, then profit, then out.

Choosing the right yardstick at the right time, and the discipline for judging a strategic project the P&L can’t yet see.

“We have a strategic project that’s underperforming financially. Do we keep backing it?”

A question I get from leaders more than almost any other.

Most strategic projects that matter go through a stretch where the numbers look bad. The mistake is treating that stretch as the verdict.

A financial number answers one question: is this making money now. It rarely answers the one that matters, which is whether this is the right thing to be building and whether it will matter once it is built. Conflate the two and you will kill a good bet and protect a bad one in the same meeting.

I want to walk through three companies that faced exactly this, and got very different grades for how they handled it. Then I want to get to the harder question underneath, the one boards rarely ask out loud.

1. Disney: name the date you switch yardsticks

New ventures inside large companies get judged on the parent’s metrics and the parent’s timeframe, and both are usually wrong. A business built to compound over five years will look like a failure at eighteen months if you grade it like a mature line. So before you judge the project, judge the yardstick.

Disney did this on purpose. At its 2020 investor day it chose subscriber growth over near-term profit, raised the Disney+ target to 230 to 260 million subscribers, and said out loud that it would absorb losses to win scale. That was not indiscipline. It was a chosen yardstick. Direct-to-consumer losses peaked around four billion dollars in fiscal 2022.

The discipline showed up in what happened next. When Bob Iger returned, he reset the metric to profit, on a named date, and committed the business to its first profit by the end of fiscal 2024. Disney’s streaming business turned its first operating profit a quarter early: positive forty-seven million dollars, against a loss of five hundred and eighty-seven million the year before. Iger later admitted they had gone in “way ahead of possible returns.”

A non-financial yardstick is legitimate only if you name the date you will switch to the financial one, and then you actually switch.

What the P&L cannot see

Three kinds of value never show up in the financials. There is option value, the right but not the obligation to a much larger market if the thing works. There is learning value, what the project teaches you that you could not have bought any other way. And there is strategic value, what it protects or positions you for. A project can be cash-negative today and still be the cheapest insurance you will ever buy against a threat to the core.

This is also where most “strategic” arguments quietly fail. “Strategic” cannot be the word you reach for when the numbers are bad and you do not want to stop. Which brings me to the next two companies.

2. Amazon: cut to the size of what is true

Amazon sold the Echo near cost on a clear thesis: people would shop by voice, and those purchases would pay back the subsidized device. The thesis did not survive contact with behavior. Reporting found only about two percent of Alexa owners ever bought anything by voice, and most never did it twice. The rest used Alexa for timers, music, and the weather. As one insider reportedly put it, the worry was that they had hired thousands of people and “built a smart timer.”

The device business reportedly lost on the order of twenty-five billion dollars across 2017 to 2021. When the mechanism the bet depended on did not appear in the data, Amazon did not redefine success to keep it alive. It absorbed the learning, kept the parts that worked, and cut the organization built on the failed thesis.

Cut to the size of what is actually true, rather than refinance the original thesis.

3. Meta: “strategic” is not a kill criterion

The opposite failure is far more expensive. Meta renamed itself around a thesis, that the metaverse would be the next computing platform, and has run its Reality Labs division to more than eighty billion dollars in cumulative operating losses since 2020, including a four-billion-dollar loss in a single recent quarter on barely four hundred million dollars of revenue.

There was no public, falsifiable metric and no stated point at which the company would stop. The “it is a decade out” framing functioned as a permanent shield against the P&L. Real cuts arrived years late, and largely once AI gave leadership another place to put the money.

Without a named metric and a pre-committed gate, conviction is indistinguishable from sunk-cost momentum.

Read the exhibit and the lesson is not “spend less.” All three made a real bet. The difference is entirely in the mechanism. Disney named the date it would switch yardsticks and switched. Amazon cut when the behavior data killed the thesis. Meta never set a number it could fail against, so it could not tell conviction from inertia, and the bill kept compounding.

The foil: do not confuse speed for discipline

The answer is not to kill faster. Quibi killed faster than almost anyone, 1.75 billion dollars raised and shut down in six months, and learned nothing on the way out.

So the discipline lives between two failures. Meta held on too long with no trigger. Quibi let go too fast with no learning. What sits in the middle is unglamorous: a metric named in advance, and a gate you committed to before your ego was in the room.

The four questions that decide a single bet

So when a strategic project is underperforming, do not ask whether it is making money. Ask:

• Are we measuring it on the right metric and the right timeframe, or grading a five-year bet like a mature line?
• What value, option, learning, or strategic, is not showing up in the P&L?
• What would have to be true for this to win, in numbers, by when? If you cannot answer that, you do not have a strategic bet. You have a hope.
• Did we set the kill criteria in advance, before anyone’s reputation was attached to the answer?

Answer those honestly and the decision usually makes itself, in both directions. The weak number was never the decision. It was the prompt to finally make one.

The harder question: not this bet, but all of them

Here is where the real boardroom tension lives, and it is rarely the one on the agenda. The agenda says “keep or kill this project.” The actual question is one level up.

Because you can’t do nothing. A company with no strategic bets is not safe, it is the classic innovation trap, and it gets disrupted while the core looks healthy. And you can’t run everything. A company with a dozen bets and no discipline is stretched so thin that none of them get the focus, talent, or air cover to actually land. Both ends fail. One slowly, one expensively.

The question is almost never “do we keep this one project.” It is “what are our strategic bets, and how do we run them as a portfolio.”

Once you hold it that way, a few things get clearer. There is a right number of bets, and it is not zero and not infinite. It scales with the size of the company, what it can fund, staff, and pay attention to at once. Treat them as a portfolio, not a list of pet projects: triage the bigger bets by size and by stage, fund the ones earning the next stage, starve or kill the ones that are not, and keep a few cheap early options running precisely because you cannot yet tell which one matters.

And the part that actually separates the companies that compound from the ones that thrash is not vision. It is the mechanics. The named metric. The pre-committed gate. The honest stage review. The discipline to switch yardsticks on time, like Disney, and to cut to what is true, like Amazon, and to never let “strategic” become the word that ends the conversation, like Meta. The process is the strategy. Everything else is just conviction with a budget.

A position firmly stated and loosely held. Tell me where I have it wrong.