Note: I’ve been sitting on a draft of this post since my days at Banyan. They say that if you’re not embarrassed by the first version of your product, you shipped too late. Consider this shipped.
I’ve uttered “You can’t cut your way to growth” under my breath a good handful of times across my career. It didn’t occur to me until recently that I should do some actual research into the axiom. Oops.
Turns out it’s true more often than not, but not for the reason I assumed. Cuts don’t fail because of the compressive effects of the cuts themselves. They fail because most cost cutting is done separately from strategy rather than in service of it. The question to ask: are you simply shrinking the company, or are you refocusing it?
The pattern that produces the second outcome is real, repeatable, and (unfortunately) super obvious: cut to create focus, cut to fund reinvestment, and cut to remove drag. Don’t “cut to make the numbers look better.” Easier said than done. But why?
Why cuts get separated from strategy
Cuts get separated from strategy because the two have different owners, different units, and different deadlines.
The cut belongs to the CFO and the board. It’s measured in dollars and heads, and it has a due date: by the end of quarter, before the financial covenant test, before earnings. The reinvestment belongs to product and go-to-market. It is measured in growth metrics, over several quarters, and it has no end date. Put a dated commitment and an undated one in the same plan and you can guess which one gets prioritized. Reinvestment becomes “phase two,” and phase two is where good intentions go to die.
There’s a second asymmetry. A cut is legible. A board can audit it line by line and confirm it happened. A growth thesis is a bet, and nobody can validate a bet until it pays off (or doesn’t).
You can see the result in how the target gets allocated: Finance sets a number, the business spreads it pro-rata, and every function eats the same ten percent. A flat percentage across the org is a tell that no hard decisions got made. It’s the cost-cutting equivalent of peanut butter, and it’s the default precisely because deciding is expensive (details matter!) and spreading is free.
None of this gets fixed with a better spreadsheet. It gets fixed by making three hard decisions, in order, before the spreadsheet opens.
Three decisions
Decision 1: What will you protect?
Before you touch a budget line, write the growth thesis down in one OKR-ish sentence. Something like:
“We will grow by winning [Segment X] with [Product Offering Y], measured by [Metrics M/N/O].”
If you can’t write it, you can’t protect it. If you don’t explicitly protect the growth engine before the cuts begin, it will get hit anyway. The growth engine is usually the biggest line item, and the biggest line item is where a percentage target is easiest to hit.
This results in that all-too-familiar, awkward conversation between a manager and their finance partner:
“How am I supposed to hit the OKRs this plan depends on if we’re cutting the people who hit them?”
“… ¯\_(ツ)_/¯”
So make a Protected List: the few capabilities that sustain your advantage. Core product work if speed is the moat. Sales enablement if distribution is the constraint. Onboarding if time-to-value is the bottleneck. The list should be short: if it isn’t, you haven’t made a real decision. A long Protected List either pushes the whole cut onto whatever is left, or, more likely, gets quietly reopened the moment the first number misses.
Skip this, and cuts land where they’re easiest, not where they’re smartest.
Decision 2: What complexity will you remove?
This is the step most miss. Cut people without cutting complexity and you end up keeping the mess while losing the capacity to fix it. Congrats: the team is both smaller and slower.
The reason is the same legibility problem. A headcount reduction has a number and a date and shows up in next quarter’s model. Killing a product results in a customer migration plan, a fight with the executive who owns it, and a revenue line that goes to zero before the cost does. So leaders take the legible cut. Same roadmap, fewer people, more burnout, slower delivery, less trust. That’s not cost reduction, it’s work compression.
The highest-leverage cuts almost always start somewhere else, and the list is longer than most leaders think:
- Governance: councils, review boards, and steering committees that exist because nobody owns the decision, plus the pre-meetings and post-meetings that orbit them.
- Goals: every OKR, KPI, and dashboard is a workstream, not a number. Three hundred OKRs is three hundred jobs.
- Portfolio: products nobody uses, features under one percent usage that still need testing and support, grandfathered pricing plans that keep old billing logic alive.
- Customers: bespoke contracts, discount exceptions that each need their own approval path, accounts that cost more to serve than they pay.
- Systems: dead code, feature flags nobody cleaned up, the second CRM from the acquisition, and the most expensive kind of all, the half-finished migration where you pay for the old system, the new one, and the team keeping them in sync.
- Org: layers, coordinator roles that route information between teams that shouldn’t need routing, and roles built around a person rather than the work.
- Zombies: everything marked “in progress” for eighteen months that survived the last three reorgs because nobody was sure who could kill it.
These are complexity taxes: overhead you pay not for what you’re building but for the accumulated weight of decisions that you haven’t reversed yet.
Cut the complexity first. Then resize the org to the simpler reality. Zero-based thinking is the formal version of this: every bucket of activity has to earn its way back in each planning cycle. It’s a useful forcing function and, as Kraft Heinz will demonstrate below, a lousy substitute for a strategy.
Drucker‘s version fits on an index card: if we weren’t already doing this, would we start it today? Ask it of every forum, metric, product, contract, and system on the list above. Anything that fails is a tax, whatever the budget line calls it.
Skip this, and you’re running the same roadmap with fewer people and more pressure. Edward Yourdon had a phrase for this: death march.
Decision 3: What will you fund with the savings?
This is where “cut to grow” either becomes real or stays a talking point: savings that go towards margin improvement and nothing else are not a growth strategy.
The reinvestment doesn’t have to be big, but it has to be specific, and it has to be written into the same plan as the cut: a number, a date, and an owner, booked before the savings hit the P&L. Once they land in the margin they become the baseline, and getting them back is a new fight. That’s the “phase two” from above, where cost cutting usually loses steam: the thing that was supposed to turn the company around was never funded, only promised. Whatever it is (a new product, a new vertical, a new geo, etc), the specific idea matters less than the commitment: a named investment, protected, with someone accountable for the outcome.
Then change how decisions get made, or the complexity will grow back. Limit active priorities. Add kill criteria. Put proof-of-life milestones on any big bet, so you’re funding a sequence of small wins instead of a Death Star. Remove the approval layers that exist to spread risk rather than improve decisions.
Skip this, and you’ve executed an austerity program. Any growth that follows will be an accident.
A self-check: cutting to grow, or cutting to cope?
You’re probably cutting to grow if you can answer yes to these:
- Are we stopping meaningful work, not just cutting headcount?
- Did we explicitly protect investment in the growth engine?
- Can leadership name the top three tradeoffs we made?
- Did we simplify the business so execution got easier?
- Did decision-making get faster, not just louder?
If the answer is mostly no, you’re cutting to cope.
What the famous turnarounds actually protected
Every turnaround you’ve heard of is a story about what got protected, not what got cut.
- LEGO, 2004: Knudstorp sold most of the theme parks, killed the non-core lines, and protected the brick. Revenue went from 6.3 billion DKK to 7.8 billion in two years, and a 1.2 billion operating loss became a 1.5 billion profit.
- Apple, 1997: Jobs cut the product line to a two-by-two grid and protected the design of what was left. Net sales fell from $7.1 billion to $5.9 billion while he did it, then climbed to $8.0 billion by 2000.
- Starbucks, 2008: Schultz closed 600 US stores and, a few months earlier, had closed all 7,100 of them for an afternoon to retrain baristas on espresso. The cut was the footprint. The protected asset was the thing in the cup. Fiscal 2010 revenue was up 9.5% to $10.7 billion on 7% comps.
- Meta, 2023: the Year of Efficiency took out roughly 10,000 roles and 5,000 open reqs, removed layers of management, killed projects that weren’t performing, and protected AI and infrastructure capex. Revenue grew 22% in 2024 and 22% again in 2025. Say what you will about the metaverse detour; the reset was a textbook example of Decision 3.
The failures are stories about what didn’t get protected.
- Kraft Heinz ran 3G’s zero-based budgeting across everything, brands included, and in February 2019 disclosed a $15.4 billion write-down alongside a dividend cut and an SEC subpoena; in 2025 it announced it would split itself in two. Savings, no growth engine.
- Sears spent roughly $5 billion on buybacks between 2005 and 2008, more than double what it put into its stores, and cut the one thing customers touched until they stopped touching it.
- Bed Bath & Beyond cut until suppliers stopped trusting it to pay, at which point the shelves emptied and bankruptcy followed. Cut operating credibility and growth isn’t slow. It’s structurally impossible.
The failures didn’t cut too little. They cut without a thesis.
But what about X?
The obvious objection is X. In 2022 Elon Musk paid $44 billion for Twitter, and then fired roughly 80% of staff on a schedule that did not seem to involve a plan. No Protected List, no complexity audit, one request that engineers print out their code so it could be reviewed on paper (which I guess is a kind of audit?). Everyone assumed the site would fall over. It didn’t. X still exists, with a quarter of the employees and a bit more than a third of the revenue, and revenue per employee is meaningfully higher, and so a certain sort of person will say: see, the “three decisions” thing is something consultants made up so they could bill for the meetings.
I guess?
Revenue per employee is the number that goes up when you fire people. That is what it is for. X’s revenue is down by more than half from Twitter’s peak and has been flat for three years; its equity was folded into xAI at $33 billion, which is a smaller number than $44 billion. One way to cut your way to growth, it turns out, is to cut a company, merge it into a company that is growing, and count the growth. This works if you own both companies. I’m not so sure if you only own the one.
That X stayed functional says more about the cost structure of the business before than the method behind the cut. The one thing worth stealing is the speed. Twitter decided in a week, and really went “full send” on it.
The next round is coming
I’m sure the next few years will see a good number of cuts in the name of “AI efficiency.” Still, the test doesn’t change: what did you protect, what complexity did you remove, and what did you fund with the savings? If the honest answer is “we did the same work with fewer people,” that’s work compression with a better press release.
Turns out, you can cut your way to growth. Most companies don’t, because they never actually decided to.
If you’ve watched it go differently, I’d like to hear about it.
Discover more from Wil Writes
Subscribe to get the latest posts sent to your email.