The moment you go from owning one roadmap to owning several, the job changes from optimizing a plan to allocating a portfolio: you're distributing bets, risk, and — critically — people across areas that will never all deserve equal attention. The math shifts from "how do I make this better" to "how much of my scarce team should this get at all."
Quick Answer: Stop treating every product area like a startup you must save. Instead, run your areas like a venture portfolio — some are core bets deserving heavy investment, some are steady cash generators to maintain, some should be harvested for value before they fade, and some should be killed so their people can go where the returns are better.
Why "one roadmap, perfected" stops being the job
A single-product PM's job is to make one plan as good as possible. A group PM's job is to decide which plans deserve to exist at their current size at all — a fundamentally different kind of thinking that most PMs are never taught before they're handed it.
The failure mode is almost universal: new group PMs keep running each area the way they ran their one product, trying to make every roadmap excellent. That's not leadership — it's just doing five single-PM jobs badly in parallel, and it guarantees under-resourcing your best bet to keep your weakest one limping along.
Portfolio thinking is the antidote. Venture capital firms don't expect every investment to succeed — they expect a distribution of outcomes and size their bets accordingly. A firm following the "power law" pattern documented by investors like Peter Thiel and firms like Andreessen Horowitz assumes the vast majority of returns come from a small number of positions, while most investments return little or nothing. Your product areas follow the same shape, whether you've acknowledged it or not.
The mindset shift, concretely
- From "fix everything" to "size everything." Not every area needs your best people; some need almost none.
- From "fairness" to "expected value." Equal headcount across areas is not fair if the areas have wildly different upside.
- From "protect my roadmap" to "protect the portfolio's return." Sometimes protecting the portfolio means starving a roadmap you personally like.
This is the conceptual foundation the rest of this piece builds on — and it's also the mental model that separates a manager who's genuinely operating at the group level from one still doing IC-style product work with a bigger title. For the fuller context of what that level shift entails, see our complete guide to the group product lead role.
The venture-capital mental model, applied to product areas
You should treat each product area the way a VC treats a portfolio company: as a bet with a distinct risk profile, expected return, and time horizon, not as a plan you're obligated to perfect. The practical shift is that you now optimize across bets, not within one.
Three ideas from venture practice map directly onto a product portfolio:
- Not every bet is meant to return the same. A VC fund expects most positions to underperform and a few to carry the fund. Your incubating area is allowed to look unprofitable right now — that's the point of a bet, not a failure of it.
- Diversification is deliberate, not accidental. Funds spread bets across stages and risk levels on purpose. If every one of your areas is a "mature, low-risk, low-growth" bet, you have no upside left in the portfolio — and if every one is an unproven bet, you have no cash flow to fund them.
- Follow-on investment is a decision, not a default. VCs decide at every round whether to double down or let a position ride without more capital. You should be making that same explicit call at every planning cycle, not silently re-staffing everyone at last year's headcount.
The single biggest mindset error group PMs make is treating "an area is under-resourced" as always a problem to fix. Sometimes it's a signal working exactly as intended — it's telling you that area doesn't deserve more.
What this means for how you talk about the portfolio
Stop asking "is this roadmap good." Start asking "is this the right amount of investment for the return I expect, given everything else I'm holding?" That question only makes sense once you can see all your areas side by side — which is exactly the reframe the rest of this article gives you the tools for.
The three-horizons framework for time-horizon allocation
The three-horizons model — introduced by McKinsey consultants Mehrdad Baghai, Stephen Coley, and David White in The Alchemy of Growth — buckets every initiative by how far out its payoff sits, so you can fund near-term, mid-term, and long-term bets deliberately instead of by accident. It gives you a second axis beyond simple risk/return: time.
| Horizon | What it is | Typical payoff window | Staffing posture |
|---|---|---|---|
| Horizon 1 (H1) | Core business, defends today's revenue | Now – 12 months | Protect and extend; your most experienced people, but not your largest headcount growth |
| Horizon 2 (H2) | Emerging opportunities, scaling what's working | 1–3 years | Fund selectively; increase investment as signal strengthens |
| Horizon 3 (H3) | Options and bets, mostly unproven | 3+ years, may never pay off | Small, dedicated teams; explicitly capped exposure |
A portfolio skewed entirely to H1 will hit a growth cliff in a few years because nothing is being built to replace today's core. A portfolio skewed to H3 burns cash with nothing funding it. The group PM's job is to hold a deliberate mix and rebalance it as areas mature — an H3 bet that proves itself should graduate toward H2 funding, and an H1 area that's declining should be moved toward harvest, not defended forever out of habit.
Applying it to squads, not just initiatives
Map each of your areas or squads onto a horizon before you plan headcount for the next cycle:
- List every area you own on one page.
- Tag each with H1, H2, or H3 based on time-to-payoff, not how exciting it is.
- Check the mix against your risk tolerance — a portfolio with zero H3 bets has no future upside; one with zero H1 has no present.
- Revisit the tags every planning cycle — horizons are not permanent labels.
The invest / maintain / harvest / kill allocation grid
A four-quadrant grid — plotting each area's growth potential against its current health or fit — gives you an explicit answer for what to do with every area you own, and forces you to name the areas you're deliberately not going to grow. This borrows from the Boston Consulting Group's growth-share matrix but replaces "market share" with a broader read on strategic fit, since most group PMs don't have clean market-share data for every internal area.
| Quadrant | Signal | Staffing action | Risk if mishandled |
|---|---|---|---|
| Invest | High growth potential, strong fit, working | Add your best people; protect from reorgs | Under-staffing this starves your best bet |
| Maintain | Stable, profitable, low growth | Keep lean, efficient team; automate where possible | Over-staffing wastes capacity that Invest needs |
| Harvest | Declining relevance, still generates value | Minimal team, milk remaining value, no new bets | Letting sentiment keep it overstaffed |
| Kill | No path to relevance, draining focus | Sunset formally, reassign people immediately | Slow-walking the decision drags morale everywhere else |
Placing an area in "kill" is not a verdict on the people who built it — it's a statement about the market or the business context that changed around it. Say that explicitly when you communicate the decision; conflating the two is what makes these calls feel punitive instead of strategic.
Running the grid honestly
- Score every area independently — resist scoring based on who leads it or how attached you are.
- Write the quadrant down somewhere visible; a grid nobody can see isn't a decision, it's a private opinion.
- Pair each placement with a staffing action, not just a label — "Harvest" without a headcount change is just a label you agree with in theory.
- Revisit quarterly; areas move quadrants as markets and execution shift.
This is the same discipline behind prioritization frameworks like RICE and Kano scoring at the feature level — you're applying the identical logic one altitude higher, at the area level instead of the feature level.
Your real leverage: deciding what NOT to staff
The group PM's highest-leverage decision is not which roadmap to approve — it's which area gets zero additional headcount this cycle, because that's the decision that actually reallocates scarce capacity toward your best bets. Every PM can say yes to a good idea. Very few can say no to a reasonable one on purpose.
This is uncomfortable because every area lead you support will make a reasonable case for more people — and most of those cases will be true in isolation. The portfolio-level question isn't "is this a good use of headcount" — almost everything is, locally. It's "is this the best use of headcount relative to everything else I'm holding."
Every additional person on a Maintain-quadrant area is a person not available to your Invest-quadrant bet. There is no headcount-neutral way to say yes to everything.
A practical staffing review checklist
- Name the opportunity cost out loud. For every staffing request, name which other area's growth you're implicitly slowing by granting it.
- Protect your Invest quadrant first, always. Fill Invest headcount needs before considering any Maintain or Harvest request.
- Cap Harvest and Kill areas explicitly. Set a hard ceiling on their team size and don't let scope creep quietly grow it back.
- Reassign, don't just cut. When you pull people from a declining area, have their next area already identified — an unstaffed transition reads as punishment, a re-deployment reads as strategy.
- Say the "no" in writing. A verbal deprioritization gets renegotiated in every 1:1; a written allocation decision is harder to slow-walk around.
This is also where the doing-to-enabling shift compounds: the leverage isn't in reviewing every ticket across five squads, it's in the allocation call that determines what those squads spend their time on in the first place. If you're still working through that transition mechanically, our piece on moving from doing the work to enabling others to do it covers the operating habits that make this kind of delegation possible at all.
Making the allocation call without flying blind
Every framework above assumes you can actually see your areas side by side — their health, their staffing, their trajectory — at the same time. In practice, that view is scattered across five different roadmap docs, a headcount spreadsheet, and whatever you remember from the last stakeholder update. Prodinja's Workspaces view is designed to hold each of your product areas as a card — status, type, funding notes — side by side, so you can reason about bets and coverage across the whole team instead of relitigating one product at a time in your head. It doesn't make the invest/maintain/harvest/kill call for you — that judgment call is still yours — but it gives you one place to hold the comparison you're actually trying to make.
Two things make that comparison reliable rather than performative: consistent standards across areas, so a "healthy" signal means the same thing in every squad, and PMs underneath you who can make sound calls without escalating every borderline decision. On the first, see our guide to setting product standards without becoming the bottleneck who reviews everything. On the second, see our piece on teaching product judgment at scale — because a portfolio view is only as good as the judgment of the people generating the data feeding it.
Key Takeaways
- Portfolio thinking replaces "make every roadmap great" with "allocate bets deliberately across areas with different risk, return, and time horizons" — the core mindset shift for any newly-promoted group PM.
- The venture-capital model expects most bets to underperform — a power-law return distribution, not a uniform one — so an underfunded area isn't automatically a problem to fix.
- The three-horizons framework (H1/H2/H3) adds a time dimension, ensuring you're funding today's core, tomorrow's growth, and options on the future simultaneously rather than by accident.
- The invest/maintain/harvest/kill grid forces an explicit staffing action for every area, including the uncomfortable ones — sentiment-driven overstaffing of a Harvest area is one of the most common portfolio mistakes.
- Your highest-leverage decision is what NOT to staff — saying no to a reasonable request is what actually reallocates capacity to your best bets.
- A single, honest side-by-side view of your areas — whether a shared spreadsheet or a dedicated tool — is a prerequisite for making any of these calls with real information instead of memory and vibes.
Frequently Asked Questions
How is product portfolio management different from single-product roadmapping?
Single-product roadmapping optimizes one plan to be as good as it can be; product portfolio management decides how much investment each of several plans deserves relative to the others. The unit of decision-making moves from features within a roadmap to headcount and risk across roadmaps.
How often should I re-score my portfolio grid?
Re-score quarterly at minimum, and immediately after any major market signal — a competitor launch, a churn spike, or a horizon-3 bet suddenly showing real traction. Quadrant placements should be living judgments, not settled once a year and forgotten.
What if every area technically deserves more investment?
That's the normal state, not a special case — almost every area can make a reasonable case for more resourcing in isolation. Your job is comparing those cases against each other and explicitly saying no to some of them, not finding a way to fund all of them a little.
Isn't it demoralizing to tell a team their area is being harvested or killed?
It can be, if communicated as a verdict on their work rather than a market-driven allocation decision. Naming the reasoning honestly — declining relevance or fit, not effort or talent — and pairing it with a clear next assignment for the people involved is what keeps the message strategic instead of punitive.
How does this connect to customer-facing frameworks like JTBD or journey mapping?
Portfolio decisions should still be grounded in real customer signal, not just internal politics — an area's growth-potential score should reflect genuine unmet jobs or journey friction, not just internal enthusiasm. Our guides to the complete jobs-to-be-done framework and mapping the full customer journey are useful inputs when you're scoring an area's real growth potential rather than guessing at it.