The 70-20-10 Trap: Allocating a Product Portfolio Across Horizons

Allocate portfolio capacity by time horizon, not by feature ranking: Horizon 1 (core) defends today's revenue, Horizon 2 (adjacent) extends it, and Horizon 3 (transformational) buys options on the future. The right split isn't a fixed 70-20-10; it's a ratio set from market exposure, cash runway, and board risk appetite.

Quick answer: Don't inherit 70-20-10 as a default. Use the Three Horizons model to separate core, adjacent, and transformational bets, fund Horizon 3 in small staged tranches rather than one commitment, and defend it to your board with option value and learning velocity — not this quarter's revenue.

This is portfolio-level work, a layer above roadmapping covered in the complete guide to the VP of Product role, and it's where a lot of otherwise strong product leaders get stuck. Ranking features inside a single roadmap is a solved problem. Allocating scarce engineering capacity, headcount, and executive attention across time horizons — while a board watches the quarter — is a different discipline, and 70-20-10 is a poor substitute for actually doing it.

Why the 70-20-10 Rule Breaks Down as a Portfolio Strategy

The 70-20-10 split fails when it's treated as a fixed budget line instead of an actively defended capacity allocation. Teams protect the 70% core because it's measurable and familiar, tokenize the 10% transformational slice because it isn't, and cut it first the moment a quarter gets tight — which defeats the entire reason for holding it.

Part of the problem is that "70-20-10" is actually three unrelated rules wearing the same numbers:

• Google's time-allocation rule (popularized around its 2004 IPO letter and later described publicly by Marissa Mayer) — roughly 70% of engineering time on the core product, 20% on adjacent projects, 10% on speculative ones.
• Nagji and Tuff's portfolio-investment finding, published in their 2012 Harvard Business Review article "Managing Your Innovation Portfolio" — across a large sample of companies, a roughly 70/20/10 split of capital between core, adjacent, and transformational initiatives correlated with stronger total returns.
• A team's internal shorthand for risk tolerance, invoked in planning meetings with none of the rigor of either source behind it.

These are different units — time, money, and risk tolerance — collapsed into one memorable ratio. A VP who imports the number without the underlying study loses the part that mattered most.

Nagji and Tuff also found the return distribution ran opposite to the investment distribution: transformational bets, despite drawing the smallest share of funding, tended to generate a disproportionate share of long-term value, while the well-funded core delivered dependable but comparatively modest compounding. Allocate 10% to the future, then measure it by this quarter's revenue, and it will always look like the worst investment in the portfolio — right up until the core matures and nothing is behind it.

Clayton Christensen's research on the resource allocation process (RAP), in The Innovator's Dilemma, explains why this happens structurally rather than through neglect. Capital and talent flow toward initiatives that serve known customers with measurable near-term profit, because that's what the internal planning process rewards — sustaining work wins the internal competition for resources by design, not by accident. A horizon split that isn't deliberately protected reverts to whatever RAP would have produced anyway, which is close to 100-0-0.

A quick diagnostic makes this concrete: pull your last four quarterly business reviews and check which dashboard covered your Horizon 3 initiatives. If it was the same revenue-and-retention dashboard used for Horizon 1, the split had already collapsed long before any budget line got cut — the measurement system did the damage first.

The Three Horizons Framework: Allocating Across Time, Not Just Risk

The Three Horizons model — McKinsey's Mehrdad Baghai, Stephen Coley, and David White — splits an innovation portfolio by time-to-payoff, not just risk: Horizon 1 defends the core, Horizon 2 builds emerging businesses, and Horizon 3 creates options for a future that hasn't arrived. Forcing all three under one glide path is what makes 70-20-10 feel arbitrary.

Baghai, Coley, and White introduced the model in The Alchemy of Growth, arguing that companies stall when they fund only Horizon 1 and treat Horizon 2 and 3 as optional. The table below is the version most portfolio reviews actually need:

Read across the rows, not down the columns. Horizon 3 isn't "riskier" so much as it's answering a different question than Horizon 1, and grading it on Horizon 1's metric will make it look like a failure by construction, every single time.

Horizon 2 is where portfolios quietly under-invest, because it's neither as urgent as Horizon 1 nor as exciting as Horizon 3. The best Horizon 2 candidates usually surface from the same discipline as a complete guide to jobs-to-be-done: what job are customers already hiring an adjacent product to do, that your platform is positioned to absorb instead?

Mapping where the current experience breaks down — the work covered in a complete guide to the customer journey — is often where both Horizon 2 line extensions and Horizon 3 assumptions actually originate. A journey's worst friction points tend to sit exactly on the boundary between what the core product does today and what an adjacent or transformational bet could do instead, which makes journey mapping one of the few inputs that feeds two horizons at once.

Real-Options Thinking: Fund Transformational Bets Like a Venture Portfolio

Treat a transformational bet as a call option, not a project: pay a small premium to learn whether a core assumption holds, then decide at each gate whether to expand, pivot, or kill it. Never green-light the full multi-year build on quarter-one conviction alone.

This is the real-options reasoning Timothy Luehrman described in Investment Opportunities as Real Options, applied to innovation the way Rita McGrath and Ian MacMillan applied it in Discovery-Driven Planning. Instead of committing to a single NPV projection built on untested assumptions, you fund the next test of the riskiest assumption and let the plan update as evidence arrives.

A practical staging sequence looks like this:

1. Write the reverse income statement. Work backward from the revenue Horizon 3 would need to matter, and list every assumption that has to hold to get there.
2. Rank assumptions by uncertainty and consequence. Fund a test for the riskiest one first, not the one that's easiest to build.
3. Fund only the next test, not the full roadmap — a few weeks or a small team, sized to retire exactly one assumption.
4. Set expand, pivot, and kill criteria before you start, so the gate decision isn't made under the emotional pull of sunk cost.

A useful tell: if a Horizon 3 initiative has never had a kill review, it isn't actually being treated as an option — it's being treated as a mandate, and mandates are exactly what accumulate sunk-cost pressure until a bad quarter forces an abrupt, late cancellation instead of a cheap, early one.

Setting Your Actual Horizon Split, Not the Default 70-20-10

Set your split from four inputs — core business maturity, competitive and disruption exposure, cash runway, and the board's actual risk appetite — then re-derive it every planning cycle instead of inheriting 70-20-10. A mature, cash-rich core might run 55-30-15; a single-product company might reasonably run 85-10-5.

Either ratio is defensible provided it was chosen, not drifted into. Work through these four inputs before committing to a number:

• Core maturity — where is the core on its S-curve? A flattening core needs a heavier Horizon 2 bet sooner than the calendar suggests.
• Disruption exposure — how many plausible entrants or substitutes could make the core's economics obsolete inside three years?
• Cash runway and investor patience — Horizon 3 is the first thing a board cuts under cash pressure, so know how much runway actually exists before promising a ratio you can't hold.
• Explicit risk appetite — ask the board directly what ratio they'd defend publicly; don't infer it from what they approved last cycle.

Put a date on the next review before anyone leaves the room. A ratio without a scheduled recheck degrades the same way an unowned metric does: quietly, and almost always toward whichever horizon the organization already understands best.

Who owns this ratio matters as much as the number itself. It belongs in your product operating model that doesn't depend on you personally as a named, recurring decision — not held in one VP's head and re-litigated informally every cycle. It also needs cross-functional buy-in before it's real: the same groundwork required for getting executive alignment around one roadmap applies directly to a horizon split, since finance, sales, and engineering will each pull the ratio toward their own incentive the moment it isn't written down.

Defending Transformational Bets to a Metrics-Hungry Board

Boards cut Horizon 3 first because it's usually pitched on the wrong metric — revenue it cannot yet produce — instead of the option value it's creating. Present it as a hedge: name the future the core can't reach alone, the assumption being tested, and the cost of not holding this option.

A four-part method holds up better in a board meeting than a features list:

1. Name the strategic question, not the feature. "What do we do when core growth flattens in 18–24 months" is defensible; "we're exploring an idea" is not.
2. Show the assumption-test cadence. List what's been tested, what was killed, and what that killing saved — a visible kill record is evidence of discipline, not failure.
3. Reframe the metric ask from revenue to learning velocity and option value: how much cheaper is it to decide now than to decide in two years with less information?
4. Show the sizing discipline. Staged tranches with named gates read as controlled risk; an open-ended budget line reads as an unaccountable one, regardless of the idea's merit.

This is the same discipline as turning a product ask into a business narrative the board can act on: boards fund a narrative with named risks and a clear ask, not a spreadsheet of initiatives with a dollar figure attached.

Scoring the Bets Inside Each Horizon

Three Horizons tells you how much capacity each horizon gets; it doesn't rank the ideas competing for one horizon's slice. That's a separate scoring problem — inside Horizon 1 you weigh sustaining features against each other, inside Horizon 3 you weigh which assumption is riskiest to test first.

Each horizon needs its own defensible, repeatable scoring method rather than a gut call in a planning meeting. This is the layer where a framework like RICE (reach, impact, confidence, effort) or Kano (which separates must-haves from delighters) earns its keep — both produce a comparable number for bets that otherwise feel like apples versus oranges.

Prodinja's RICE and Kano prioritization tooling is built for exactly this layer: it can score candidate bets within a horizon on a consistent basis, so the portfolio-level split you present to the board sits on a defensible, comparable foundation rather than on whoever argued loudest in the room.

Key Takeaways

• 70-20-10 is three different rules wearing one number — Google's engineering-time heuristic, Nagji and Tuff's capital-investment finding, and an internal risk shorthand measure different things and shouldn't be swapped for each other.
• Three Horizons separates bets by time-to-payoff and funding logic, not just risk, so Horizon 1, 2, and 3 initiatives can be graded on their own appropriate metric instead of one shared one.
• Fund transformational bets as staged real options, not annual commitments — pay a small premium to retire the riskiest assumption first, and set kill criteria before you start, not after the sunk cost accumulates.
• Set your own ratio from maturity, exposure, runway, and explicit board risk appetite — a chosen 85-10-5 beats an inherited 70-20-10 every time.
• Boards cut Horizon 3 first when it's measured on revenue; reframe the ask around option value, learning velocity, and the cost of not holding the option at all.
• A visible kill record is evidence of discipline, not failure — showing what you tested and stopped funding is often more persuasive than showing what's still running.
• Horizon allocation and within-horizon scoring are two separate decisions — get the split right first, then use a consistent method like RICE or Kano to rank what competes inside it.