You have a number. It arrived from a board, a sponsor, a lender, or a forecast that stopped working, and it is larger than anything you can find by being slightly more careful with travel. Every function has sent you a list. The lists add up to about a third of what you need, and the parts that would actually get you there are the parts nobody wants to write down. Now somebody has to decide what goes first.
These are field notes on that decision. Not on what to cut, which depends entirely on your business, but on the order, which mostly does not. In my experience the order is worth more than the target. Two companies with the same number and the same cost base end up in completely different places depending on what they did in week one, and the difference compounds, because the early moves determine how much time and credibility you have left for the hard ones.
The organizing principle is not size. It is this: cut in order of how easily you could undo it if you were wrong, and how quickly it turns into cash. The cheapest mistakes go first.
Why does the order of cuts matter more than the size?
Because the early cuts buy the time for the late ones. Fast, reversible savings with no capability loss create cash and credibility inside a quarter, which is what lets you do the slow structural work properly instead of hacking at it under a deadline. Reverse the order and you spend your best options in month one.
There is a second reason, and it is the one that shows up in the numbers a year later. Cuts differ enormously in how likely they are to come back. Stopping a subscription is permanent if somebody owns the review. Cutting a maintenance program is temporary, because the work does not stop being necessary, it just stops being scheduled, and it returns as a capital event with a worse price attached.
Anything you cut that is currently doing necessary work will find its way back into the cost base through a route you are not watching. Overtime. Contractors. Expedited freight. A vendor invoice coded somewhere unfamiliar. If you sequence badly, your reported savings and your actual savings diverge, and the gap does not become visible until somebody asks why the run rate has not moved.
Cut in order of how easily you could undo it if you were wrong. The cheapest mistakes belong at the front, where you can still afford them.
What should you cut in the first two weeks?
Everything you are paying for that is not currently buying you anything. Duplicate vendors, unused licenses and subscriptions, maintenance contracts on equipment that has left, rented space holding obsolete stock, rates nobody has revisited since signing, and standing approvals for spend that was authorized under different conditions. Free money, zero risk.
The work here is clerical rather than strategic, which is why it gets skipped in favor of something that feels more serious. Pull every recurring payment in the last twelve months, sort by value, and assign each one a named owner who has to answer a single question in writing: what does this buy us today. Not what it was for. What it buys today.
Two practical notes. First, do this centrally, not by asking each function to self-report, because a function reviewing its own recurring spend is being asked to document its own inattention and will produce a shorter list than the data would. Second, put a decision date on it. Two weeks, not a quarter. The exercise has a half life, and past about twenty days it turns into a standing committee.
Alongside it, freeze new commitments rather than cutting existing ones. A freeze on new contracts, new vendors, new hires, and new capital costs nothing, damages nothing, and is fully reversible in an afternoon. It also stops the cost base growing while you are trying to shrink it, which is a real phenomenon and an embarrassing one to explain later.
What comes next, and why does it take a quarter?
Cost that exists for a reason you can remove. Indirect spend consolidated across sites, freight lanes and modes rebid, internal service levels reset to what is actually needed, and the complexity that drives cost across every function at once. These need analysis before action, which is why they are second rather than first.
The largest pool in most mid-market manufacturers is here, and it is the least visible, because it has no budget line and no owner. The cost of running four hundred active part numbers against demand that would justify a hundred and fifty is charged to purchasing, scheduling, warehousing, engineering, and quality in fractions that never get added up. It does not appear on any function’s cost reduction list because no function controls it.
Getting at it requires a sort the business does not normally do: actual profitability by product and customer with full cost attribution, not standard margin. That takes a few weeks of finance work and it is the highest return few weeks available. Until it exists, you are choosing cuts without knowing which parts of the business are paying for the others.
This is also the tier where demand management works. A large share of internal cost exists because somebody is requesting something at a volume nobody has questioned. Report packs, custom quotes, sample runs, engineering changes, expedites. Cutting the team that produces these leaves the demand intact and the work undone. Reducing the demand removes the cost permanently.
When do you touch capacity?
After the complexity work, not before. Plants, lines, shifts, and footprint should be sized against the mix you are going to have, not the mix you have while the rationalization is still running. Consolidating first and rationalizing second is how businesses end up with a network built for a company that no longer exists.
The temptation to invert this is strong, for a reason worth naming. Capacity decisions are the ones that produce announceable events. A closed line is a number you can present. Nine hundred retired part numbers is not, even when it is worth more. If the program is being run to produce announcements rather than run rate, capacity will move to the front and you will pay for that ordering for a decade.
There is one exception that is genuinely urgent: capacity that is costing you money right now with no volume behind it at all. An idle site, a mothballed line still carrying fixed cost, a lease on space you do not occupy. That is not capacity sizing, it is carelessness cost wearing a capital costume, and it belongs in the first two weeks.
What goes last, and what never goes?
Capability goes last: the people and programs that make you good at something specific. By the time you reach them the target is usually covered. If it is not, cut capability only where someone can name the product line or segment it was protecting and that line or segment no longer exists.
The never list is shorter than people expect but it is firm. Preventive maintenance, because deferring it converts a controllable operating expense into an uncontrollable capital event on a timeline you do not choose. The technical or application function that gets your product specified rather than quoted, because once you are being quoted you are competing on price against a lower cost structure. And anything that is currently containing a problem: inspection, rework, expediting, the person reconciling two systems by hand. Cut the cause and the cost leaves on its own. Cut the cost and the cause finds a more expensive route, usually through the customer.
A useful test at this end of the sequence: ask what breaks, when, and who notices. If the answer is that nothing breaks, take it. If the answer is that something breaks in eighteen months in a line item you do not control, you have found a cost transfer rather than a cost reduction.
When does this sequence break down?
In five situations, and they are common enough that you should check for them before starting. A genuine liquidity event, an exit inside twelve months, a single dominant customer at risk, an integration clock you do not control, and a notice period structure that inverts the calendar. In each case, something other than reversibility sets the order.
The liquidity case is the clearest. If the constraint is cash this month rather than cost this year, you take cash wherever it is fastest and you accept that you are damaging the structure. That is triage, not cost work, and pretending otherwise kills companies. If you are here, stop reading about sequence and go build a thirteen week cash forecast.
The exit case inverts things differently. A business being sold inside a year needs savings that will appear in the reported numbers a buyer will diligence, which means visible, documented, P&L-landing cuts. The structural complexity work is worth more to the business and will not land inside the window. Uncomfortable, but that is the trade, and the honest version of this advice acknowledges it rather than pretending the long game always wins.
The concentrated customer case overrides everything: if a single account represents enough revenue that losing it changes the company, nothing that touches that account’s service gets cut at any point in the sequence, regardless of how attractive it looks on the list.
The integration case is about somebody else’s clock. Transition services agreements after a carve-out expire on dates you did not set, and stranded cost becomes real on those dates whether or not you were ready. The TSA calendar sets the order, full stop.
The notice period case is the most often missed. Where contractual or statutory notice runs to months, the decision has to be taken long before the saving appears, which means it has to be made on analysis you would normally do in the second tier. That does not change the principle. It changes when the analysis has to happen.
If a single account is large enough that losing it changes the company, nothing touching that account’s service gets cut, at any point in the sequence.
How do you know the sequence is working?
Watch three things: whether the savings are showing up in the run rate rather than in a report, whether cost is appearing in unwatched categories such as overtime, contractors, and expedited freight, and whether anyone is asking for a second round. A second round inside a year means the first one removed spending without removing any reason for it.
One more thing worth watching, because it is the earliest reliable signal and almost nobody tracks it: the rate at which exceptions are being requested. Every cost program creates an exception process, formally or informally, and the volume of requests coming through it tells you more about whether the cuts will hold than the savings report does. A handful of well-argued exceptions is a healthy program. A steady stream of them means you cut something the business genuinely needs and the organization is routing around you rather than arguing with you.
Route those requests through one person and read them weekly. They are the cheapest diagnostic available on whether your sequence was right, and they arrive months before the run rate tells you the same thing. In a business I ran myself, the exception log was the first place the mistakes showed up, every time, and it was the only report I read that nobody had prepared for me.
Track the released cost against the categories it came from, not against the target. A program that hits its number entirely through the first tier has not finished, it has front-loaded, and the run rate will tell you that in about two quarters.
The full sequence, in order:
- Days 1 to 14. Freeze new commitments. Cancel everything that buys nothing today. Exit idle capacity that carries fixed cost.
- Weeks 3 to 8. Build actual profitability by product and customer with full cost attribution. Rebid indirect spend and freight centrally.
- Weeks 6 to 16. Reduce internal demand: report packs, custom quotes, expedites, engineering changes. Remove the request, not the team.
- Quarter 2 to 3. Rationalize the range and the customer tail. Gate new part numbers so it cannot regrow.
- Quarter 3 to 4. Size capacity and footprint against the post-rationalization mix.
- Last. Capability, selectively, and only where the thing it protected is gone.
None of this is quick, and the objection is always that the number is due this year. Sometimes that is true. More often the deadline is a reporting deadline rather than a solvency one, and it is being used to justify a pace that guarantees the savings will not hold. Eighty percent delivered permanently beats a hundred percent that gives half of it back, and the second conversation with your board is much easier when you have the first kind.
The full case for why cost structure change outlasts cost cutting sits in the companion piece on how to cut costs without gutting the business, which covers the four categories of cost this sequence is moving through.
Monday morning: pull the last twelve months of recurring payments, sort by value, and put a named owner and a written answer against the top fifty.
If your leadership team is about to run this and you want the sequence argued in front of them rather than circulated as a document, that is a talk I give: see the speaking page.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as Global VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
See also: Why Across-the-Board Cuts Always Fail