You asked how to run a product portfolio review. The question underneath it is narrower and harder: who is allowed to kill a product line, on what evidence, and by what date. A review that cannot answer those three things is a reporting meeting with a strategic name on the invitation.
That is why most portfolio reviews produce slides rather than decisions. The agenda is built around presenting performance, and presenting performance is something every line owner can do indefinitely without anything changing. What follows is a review built around the opposite premise: every line leaves the meeting with a verdict, and every verdict moves a resource.
The decision content itself, what fund, harvest, and exit mean and how to tell which applies, is in the guide to product portfolio strategy for manufacturers. This article is about the meeting that produces those verdicts and the machinery around it.
What is a product portfolio review actually for?
To produce a verdict on every product line and then move engineering hours, capital, and constrained capacity to match. The review exists to reallocate scarce resources on purpose rather than by default, and any review that ends without a reallocation has not happened yet.
There are three outputs, and only three. A verdict per line and segment. A named owner and date for every verdict that needs work to execute. And at least one resource that moves this quarter as a direct result. Everything else on the agenda, the performance recaps, the market commentary, the competitor updates, is input to those outputs or it is decoration.
Notice what this rules out. A review whose purpose is to inform senior management is a briefing, and briefings are useful but they do not need a grid or decision rights. A review whose purpose is to hold line owners accountable for their numbers is a performance meeting, and it belongs on a monthly cadence with operating measures. Neither one reallocates anything, and running either under the portfolio banner is how businesses conclude that portfolio review does not work for them.
The reallocation requirement is the one that does the real work, because it is the only part of the review that costs something. Verdicts are free. A verdict that moves two engineers off one line and onto another has a price attached, which is exactly why it forces the organization to mean it.
Test any review you already run against that list. If last year’s review produced a deck and no reallocation, the honest conclusion is that the portfolio was left exactly as it was, and that the decision to leave it alone was made without anyone being accountable for it.
I dig into governing bets versus cash machines in my podcast episode Alphabet Wasn’t About Diverse Businesses.
Isn’t this what the annual budget already does?
No. The budget allocates money against next year’s plan, and that plan is built from this year’s shape. It asks how much each line will spend, not whether each line should exist at all, and those are different questions with different answers arrived at by different people.
Watch what a budget process actually does with a weak product line. It gets a slightly smaller number than last year, its projects get deferred rather than cancelled, and it stays in the portfolio consuming engineering attention and constrained capacity that never appear in its budget line at all. Nobody decided to keep it. The process simply had no step where killing it was on the table.
There is a timing problem underneath this as well. Budgets are usually built bottom up, starting with each line’s own plan, which means the first complete view of the portfolio arrives after every line has already written its case. By then the conversation is about trimming numbers rather than choosing directions, and nobody wants to reopen work that took six weeks to assemble.
The portfolio review sits upstream of the budget and tells it which lines are allowed to grow. Run them in that order, with the review far enough ahead that its verdicts can actually shape the numbers, and the budget stops being the place where portfolio decisions are quietly made by people who were not asked to make them.
One more thing the review is not for: rescuing a line by talking about it. Attention is not a resource the portfolio allocates. A line discussed at length every year and funded in none of them has been harvested by default, and saying that out loud is more honest than another year of concerned discussion.
Who should be in the room, and who decides?
Keep the room small: the person who owns the decision, the line owners who supply evidence, and whoever controls the engineering queue and the constrained capacity. One person decides. Line owners correct the inputs and answer questions, and they do not vote on their own lines.
You will object that the people closest to each line know the most about it, and that is exactly right, which is why they belong in the room. What they cannot do is judge their own line against the others, for the same reason nobody grades their own exam. Their incentives, their history with the product, and their team’s future are all attached to one answer.
There is a version of this that looks like delegation and is not. Some businesses give the decision to the person who runs operations, on the grounds that they are neutral between product lines. They are neutral between lines and not neutral between outcomes, since almost every exit simplifies their week and almost every funding verdict complicates it. Neutrality has to mean an interest in the whole portfolio’s return, not an absence of interest in the lines.
The single most common failure in portfolio review design is consensus. A committee that requires agreement will keep every line, because every line has one defender and killing anything requires everybody. Consensus feels like rigor and functions as a veto.
Size the room accordingly. Beyond about eight people, the meeting becomes a presentation regardless of the agenda, because the cost of speaking rises and the people with the least attachment to any given line stop contributing. If your portfolio genuinely involves more stakeholders than that, split it by product family and run two reviews with the same decision owner rather than one large one.
Bring the capacity owners because verdicts that do not touch the schedule do not survive contact with the quarter. If the person who allocates engineering hours is not in the room, a funding verdict is a sentence in a document, and the hours will go where they always went.
What has to be on the table before anyone talks?
Five inputs per product line and segment: contribution and its three-year direction, the advantage a customer can name for buying from you, the constrained capacity the line consumes, the engineering hours it consumed last year, and the obligations attached to it. Circulated in advance, not presented in the meeting.
The advantage is the input that gets faked most often, and it is the one that decides the most. Written properly, it quotes customers in a segment on why they buy from you. Written lazily, it is a sentence from your own marketing material. The test is simple: if it needs your explanation to make sense, it is not an advantage a customer can name.
Notice how few of these five are financial. Two of them, the advantage and the obligations, cannot be pulled from any system, which is why the pre-work takes real effort the first time. That is also why a review built only from what the ERP can export tends to conclude that every line is fine: the inputs that would have contradicted it were never gathered.
Obligations are the input that gets forgotten and then blows up an exit three months later. Every supply contract, regulated item, and service-part commitment belongs on the sheet with the date it ends, because those dates set the earliest possible timing for any exit verdict and they are almost never what people assume.
Contribution deserves one caution. If your costing spreads overhead by volume or labor hours, the contribution figures in the grid will flatter complex low-volume segments and understate simple high-volume ones, sometimes badly enough to reverse a verdict. Fix the allocation roughly before the first review, or treat contribution as the least reliable of the five inputs and lean harder on the advantage and the capacity consumed.
Include the engineering hours even when the number is rough. It is usually the input that changes minds, because a line that consumed a fifth of your engineering capacity while contributing a twentieth of the profit has made an argument that no amount of market commentary can answer.
The pre-work is the review
By the time people are in the room, the work is either done or the meeting is theater. Build and circulate the grid at least a week ahead, with a deadline for corrections that closes before the meeting starts.
Set that expectation explicitly, because it inverts the habit most organizations have. In a presentation culture, the meeting is where the analysis is revealed and then debated for the first time, which guarantees that the discussion is about whether the numbers are right rather than what to do about them. Closing corrections beforehand removes the most effective delaying tactic available to anyone who does not want a verdict on their line.
Expect the first cycle to generate a lot of corrections, and treat that as the process working. Setup times in routings are often years stale, cost allocations often miss freight or packaging, and segment revenue is often assembled by hand for the first time. Every correction improves the data the business runs on, and the second cycle takes a fraction of the effort.
Circulating the grid early has a second effect worth having. Line owners who see their own numbers a week ahead tend to arrive having already accepted the weak cells, because the argument they were going to make has had time to lose to the evidence in private. The meeting then spends its time on the cells that are genuinely contested rather than on the ones that were merely unexamined.
One discipline makes the corrections honest. A correction has to change a number and name its source, not merely register disagreement with the conclusion. “That setup is ninety minutes, not two hours, here is the last run record” is a correction. “These figures do not reflect the strategic value of the line” is an opening argument, and it belongs in the meeting.
What actually happens in the meeting?
Line by line, each gets its verdict: fund, harvest, or exit, per segment. The decision owner states the verdict, the room tests it against the five inputs, and anything unresolved gets an owner and a date rather than another meeting. Two hours is usually enough for a mid-sized portfolio.
Start with the cells nobody disputes, not the contested ones. Clearing the obvious verdicts in the first twenty minutes establishes that the meeting produces decisions, and it shows the room what the evidence standard looks like before anyone has to apply it to a line they care about. Opening on the hardest cell does the opposite and usually consumes the whole session.
Run the agenda against the grid rather than against the org chart. Going line by line through the grid keeps the comparison in view, which is the entire point: a verdict on one line is a statement about what the others do not get. Going owner by owner turns it into a series of unrelated defenses, and the resource question never gets asked.
Timebox each line and hold the box. Lines with a clear position take three minutes. The argument will concentrate on the borderline cells, which is where management attention should go, and the timebox is what protects that time from being consumed by a line that nobody actually disputes.
Handle the borderline cells with a date rather than a debate. When the room cannot resolve a cell, the useful output is not a longer discussion. It is a named person, a specific question, and a deadline: find out by the fifteenth whether the two customers in this segment would pay a higher price, and the verdict follows automatically from the answer. Most stuck cells are stuck on a missing fact, not on a difference of judgment.
End every line with a recorded verdict, including the ones where the verdict is no change. A recorded decision to keep a line as it is, made by a named person with the grid in front of them, is a real decision that can be revisited when its assumptions break. Silence is not.
How often should you run it?
Once a year for the full grid, on a fixed date that sits ahead of the budget cycle, plus an immediate single-cell review whenever a named advantage changes. Quarterly full reviews sound rigorous and mostly produce reaction, because they judge verdicts before any evidence about them exists.
The reason to keep the full cycle annual is that portfolio verdicts take longer than a quarter to prove out. Fund a line and the engineering work, the launch, and the customer response run well past ninety days. Harvest a line and the cash shows up over a year. Reopening those verdicts every quarter means judging them before the evidence exists, and the result is a portfolio that oscillates.
Put the date on the calendar for the following year before leaving the room. This sounds trivial and is not. An annual review with no scheduled date becomes an eighteen-month review, then a two-year review, and by then the grid has to be rebuilt from scratch rather than refreshed, which is the difference between an afternoon and a project.
The trigger reviews are what handle speed, and they are narrow on purpose. A competitor launches into one of your segments, a customer who defined an advantage leaves, a regulation changes, a supply line breaks. Each of those invalidates one cell’s evidence, so it gets one cell reviewed, in an hour, by the same decision owner.
But our business changes too fast for an annual cycle
Then your trigger list is too short, not your cycle too slow. Businesses that believe they need continuous portfolio review are usually reacting to demand swings, which are not portfolio events at all.
Ask what the last three portfolio reviews would have decided differently with perfect foresight. In most businesses the answer is one or two cells, and both were visible at the time. That is a portfolio moving at portfolio speed, whatever the trading conditions feel like month to month.
Separate the two kinds of change. A quarter of weak orders in a segment is operating news, handled by the schedule and the sales plan. A permanent shift in why customers choose you is portfolio news, and it is rarer than the operating noise sitting on top of it. Reviewing the portfolio every time the noise moves teaches the organization that no verdict is ever final, which is worse than reviewing too slowly.
There is one genuine exception. A business in the middle of a turnaround, where the constraint and the customer mix are both moving, may need two full cycles a year for a year or two. Treat that as temporary and go back to annual once the verdicts stop changing, because a permanent state of review is how an organization avoids committing to anything.
How do you keep the decisions from evaporating?
Move a resource inside thirty days, and track the movement rather than the verdict. A funding verdict that does not change the engineering queue within a month has already been overturned by the people who schedule the work, whatever the document says.
Pick one visible reallocation and do it immediately: engineering hours moved off a harvest line onto a funded one, or bottleneck hours reassigned away from products that lost their case. The first one matters out of proportion to its size, because it tells the organization which of the two systems is real, the one that produces verdicts or the one that produces schedules.
Give the reallocation a visible owner and a date in the same sentence as the verdict. “Fund the coatings line in the transit segment” is a wish. “Move two of the four engineering heads off the legacy valve line to coatings by the end of March, owner named” is a decision, and it is checkable a month later by anyone who cares to look.
Then report the portfolio on the measures the verdicts were supposed to move. Engineering hours by verdict category, constrained capacity by verdict category, and contribution per line and segment. Reporting revenue by line tells you almost nothing about whether the review worked, because revenue moves for reasons that have nothing to do with your allocation.
A funding verdict that does not change the engineering queue within a month has already been overturned by the people who schedule the work.
The month after the review is also when the verdicts get tested by events. A large order arrives for a harvested line, or a funded line hits a delay. Neither is a reason to reopen the verdict, and treating them as one is how a portfolio strategy becomes a set of preferences. Write down in advance what would genuinely change a verdict, so that when something happens you can check it against the list instead of relitigating in the moment.
Put the verdicts somewhere the organization actually reads. A portfolio decision that lives only in the review document will be contradicted within a quarter by someone who never saw it, usually in good faith, usually by promising a customer a variant on a line that was harvested.
What if the review keeps producing no decisions?
Look at the structure rather than the participants. A review that never kills anything almost always has one of four things wrong with it: consensus decision rights, missing inputs, no resource attached to the verdicts, or an owner whose own line is in the grid.
Work through them in that order, because they are ranked by how often they are the real cause. Decision rights fail more often than analysis does. A review with perfect inputs and a consensus rule will produce nothing, while a review with rough inputs and a clear owner will produce decisions that are approximately right and can be corrected next year.
The fourth is the most awkward and the most common in smaller businesses, where the person running the review also owns the largest product line. There are two workable fixes. Put the decision rights with someone whose interest covers the whole portfolio, an owner, a board member, or a general manager one level up. Or keep the decision rights and remove your own line from your verdict, letting the same person who handles the escalations rule on it.
Check the calendar as well. A review held three weeks before the budget is due will produce whatever verdicts fit the numbers already drafted, because the alternative is redoing six weeks of work. Nobody involved will describe it that way, and the outcome is the same as having no review.
There is a fifth cause worth naming, because it hides behind the other four. Sometimes the review produces no decisions because the business genuinely cannot execute one: no engineering capacity to fund anything, no cash to absorb an exit, no slack to change the schedule. That is not a portfolio problem. It is a capacity or cash problem wearing a portfolio costume, and the review will keep failing until it is solved.
Decision rights are what make the hard verdicts possible
The clearest portfolio decisions I have been part of were the ones that would not have survived a committee. In side-by-side refrigeration at Whirlpool, a category losing $175 million a year, part of the recovery was a deliberate retreat from unprofitable volume that took market share from 53 percent down to 29 percent on purpose.
Consider what that decision looks like inside an organization. Share is the number everyone is measured on, and a plan to cut it almost in half will be opposed by people with entirely legitimate concerns, in a consensus process, indefinitely. It only happens where someone owns the verdict and can carry it.
The same pattern shows up in smaller decisions constantly. Somebody has to be able to say that a line the organization is fond of will not get the next engineering hire, and to say it in a way that survives the following week. That capacity to make a verdict stick is not a personality trait. It is a structural property of how the review was set up, and it can be designed in or left out.
That is the practical argument for the decision rights described above, and it is worth more than any refinement of the analysis. The grid tells you which lines have a right to win. It cannot tell your organization who is allowed to act on that, and a review without an answer to the second question will produce correct analysis and no change for as long as you run it.
Keep the record of what each verdict was based on, too. When a funded line disappoints two years later, the useful question is whether the advantage was real and faded or was never there, and only a written record of the original evidence can answer it. That answer is what makes the third review better than the first.
The shorter version for a smaller business
A portfolio review does not need a program office. For a business with a handful of lines, the whole thing fits on one page and takes an afternoon once a year, and the smaller you are the more each misallocated engineering hour costs you.
Put the lines down the side and the segments across the top, and fill only the cells where you have sales. For each one, write contribution, whether it is rising or falling, and the advantage in a customer’s words. Mark it fund, harvest, or exit. Then name the one resource you are moving this quarter as a result, and put the next review on the calendar before you leave the room.
The one thing the short version must not drop is the resource movement. A small business that marks three lines fund, harvest, and exit and then changes nothing about who works on what has run a labeling exercise. With a handful of engineers, moving even one person for a quarter is a larger reallocation in proportion than most corporate portfolio shifts, and it is entirely within the owner’s control.
That version is not a compromise. It contains every element that makes the large version work, which is the grid, a single decision owner, a verdict per cell, and a resource that moves. Everything else that accumulates around portfolio review in larger organizations is coordination overhead, and coordination overhead is not what makes the verdicts good.
The questions that come up every time
How long should the meeting be? Two hours for most mid-sized portfolios, with the grid circulated a week ahead and corrections closed beforehand. If it runs longer, the pre-work was skipped or the room is too big.
What if a line owner refuses to accept the verdict? They implement it anyway, and their objection is recorded with a date to revisit. Disagreement is fine and is worth capturing, since it occasionally turns out to be right. Reopening the decision every week is not.
Should customers be involved? Not in the meeting, but their words should be. The advantage input is supposed to come from customer conversations, and the best time to have those is during the pre-work, not after a line has been marked for exit.
What about new products still ramping? They get a ramp period set at launch with a defined result that counts as success, and they are exempt from verdicts until it expires. Without that, new products are protected forever by the argument that it is too early to tell.
What if two lines both deserve funding and we can only fund one? That is the review working correctly, and the tiebreaker is which advantage widens faster per unit of the scarce resource. If both look equal on that test, fund the one whose customers can already name the advantage over the one where you would have to build it first.
Do we need market share data? Not for this. Reliable share figures rarely exist for niche industrial segments, and the estimates people substitute are usually produced by the person whose line is being scored. What a customer can name as your advantage is both more available and more decisive.
Where does SKU-level work fit? Below this. The portfolio review decides which lines get resources; the SKU work decides the shape of each line. Running the two together in one meeting buries the portfolio question under a hundred part numbers, and the method for the SKU layer is in how to cut a product line without losing revenue.
One limitation of this design, stated plainly. It concentrates authority in a single decision owner, and that only works if the owner has the standing to carry an unpopular verdict and the judgment to be talked out of a wrong one. In a business where that person does not exist, this review will produce fast decisions of uneven quality, which is better than no decisions but not by as much as you would like. Fix the decision rights before you refine the grid.
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.
If your last review produced slides instead of verdicts, the problem is usually upstream of the agenda. Request a free Decision Velocity Audit and I will measure how long a portfolio decision takes in your business, and which step it waits at.
See also: How to Build an Operating Cadence, The Blunt Guide to Product Portfolio Strategy