Innovation portfolio management: the framework, the 70-20-10 split, and how to balance a portfolio

Innovation portfolio management is the practice of treating every innovation bet your organization is making as one balanced set, and allocating resources across that set deliberately, instead of approving projects one at a time as they arrive. The question it answers is not "is this project any good?" but "given everything we are already funding, is this the bet we are missing?"

Most companies do the first thing and believe they are doing the second. Projects get approved individually, on individual merit, by whoever has budget. Each decision looks defensible. The aggregate is almost always the same shape: a large pile of safe, incremental improvements to the current business, and nothing that would matter in five years. Nobody chose that. It is what you get when a portfolio is assembled by addition rather than by design.

What is innovation portfolio management?

It is portfolio thinking applied to innovation. The word portfolio is borrowed from investing on purpose, and the analogy is the useful part: an investor does not evaluate each stock in isolation and buy everything that looks good, because a set of individually attractive holdings can still be a badly concentrated portfolio. The same logic that stops a serious investor from holding one position and calling it a strategy is what innovation portfolio management applies to product and R&D bets.

Three things distinguish it from ordinary project management. It looks at the whole set at once rather than at projects individually. It classifies bets by ambition, not just by size or department. And it sets a target allocation up front, so the portfolio has a shape you are steering toward instead of a shape you discover afterward.

That last point is where the practice earns its keep. Without a target allocation, every individual funding decision quietly favors the safe option, because the safe option always has better numbers attached. Incremental work has a forecast; transformational work has a hypothesis. If the two compete head to head on projected return, the forecast wins every time, and it wins every time for a hundred consecutive decisions, which is how portfolios end up at ninety-five percent core without anyone ever deciding to be conservative.

What is the innovation ambition matrix?

The innovation ambition matrix is the classification most portfolios are built on. It comes from Bansi Nagji and Geoff Tuff, in "Managing Your Innovation Portfolio," published in Harvard Business Review in May 2012.

It is a simple two-axis grid. One axis is how new the offering is, running from using existing products and assets to developing genuinely new ones. The other is how new the market is, running from serving existing customers and markets to entering new ones. Plot your initiatives on that grid and they fall into three zones.

What is the 70-20-10 rule for innovation?

Nagji and Tuff found that companies outperforming their peers tended to allocate innovation resources in roughly a 70-20-10 split: about 70 percent to core initiatives, 20 percent to adjacent ones, and 10 percent to transformational bets.

The more interesting half of their finding is that the returns ran close to the inverse. The transformational 10 percent tended to account for something like 70 percent of the eventual return. That asymmetry is the entire argument for portfolio management as a discipline. The smallest slice of the budget produces the largest share of the payoff, and it is also the slice that loses every individual funding comparison it is entered into. Left to project-by-project decisions, it gets squeezed out. The only reliable defense is to ring-fence it as an allocation before the comparisons start.

Treat 70-20-10 as a reference point rather than a target to copy. The authors were explicit that the right split varies by industry, and reported meaningful variation: consumer goods companies skewed much harder toward core, on the order of 80-18-2, while mid-stage technology companies ran closer to 45-40-15. A stable business in a slow-moving category genuinely should sit more conservatively than a company whose market is being redefined every three years. What travels across industries is the structure of the argument, not the three numbers.

What is the difference between innovation portfolio management and the three horizons model?

They are close relatives and are often used together, but they measure different things. The three horizons model comes from The Alchemy of Growth (1999) by Mehrdad Baghai, Stephen Coley and David White, out of McKinsey. It sorts initiatives by time to material contribution: Horizon 1 is the current business defending and extending today's earnings, Horizon 2 is emerging opportunities that will contribute in a few years, and Horizon 3 is options on genuinely long-term possibilities.

The ambition matrix sorts by distance from what you know, not by time. That difference matters more than it sounds. A project can be transformational in ambition and land quickly, or be an incremental extension that takes four years. Sorting purely by timeline lets a long, dull project masquerade as a bold one, which is a common way Horizon 3 budgets get consumed by things that are merely slow.

In practice, use ambition to decide the allocation and horizons to check the phasing. A portfolio that is well balanced by ambition but has nothing arriving for four years still has a revenue problem. If you are formalizing how these bets move through review points, our guide to the stage-gate process covers the decision mechanics, and the innovation funnel covers the attrition math that determines how much has to enter for the portfolio to stay full.

How do you build an innovation portfolio?

Five steps, in an order that matters.

1. Inventory what you are actually funding. Not the projects on the innovation team's slide, everything: the R&D line, the product roadmap, the skunkworks somebody runs on Fridays, the partnership nobody has cancelled. Most first inventories surprise people, usually by revealing that between a quarter and a half of innovation spend is maintenance work that was reclassified to get funded.

2. Classify each item by ambition. Core, adjacent or transformational. Do this with more than one person in the room, because sponsors systematically overrate their own project's ambition. A useful test: if you can produce a credible revenue forecast for it, it is almost certainly not transformational.

3. Add up the actual allocation. Use money and people, not project counts. Ten small transformational experiments and one enormous core program is not a 90-10 portfolio by headcount even though it looks like one by count. This is the step that produces the uncomfortable number.

4. Set a target split and write down why. Start from the benchmark, adjust for how fast your category is actually moving, and record the reasoning so the number survives the next budget round.

5. Close the gap deliberately. Usually this means killing or shrinking core projects, because the gap is almost never a shortage of core work. This is the step that fails most often, and it fails for an unglamorous reason covered below.

Why do innovation portfolios drift toward the core?

Three forces, all of them rational at the level of the individual decision.

The first is evidence asymmetry. Core projects come with data because they extend something that already exists and already has a customer base to measure. Transformational bets come with a story. Any process that rewards well-evidenced proposals will select against ambition, not because anyone prefers safety but because the evidence bar is impossible to clear for something genuinely new.

The second is that measurement systems are built around the core. Quarterly revenue targets, existing customer satisfaction and current-product margin are what get reported, so they are what people optimize. A transformational bet damages every one of those metrics in the short run.

The third is the supply problem, and it is the one that gets the least attention. You cannot allocate ten percent of your budget to transformational work if no transformational candidates exist. Most organizations discover, when they inventory their portfolio, that the imbalance is not a funding decision at all. There simply were no bold options on the table to fund, because the pipeline was filled by people asking "what should we improve?" rather than "what could we do that we have never done?"

That distinction is worth sitting with. If the shortfall is a funding problem, the fix is budget discipline. If it is a supply problem, budget discipline does nothing at all, and the fix is upstream: you need a deliberately wider set of candidate directions before allocation is even a meaningful exercise. Structured divergence is the mechanism, and the reason it needs structure is documented in the research on why group brainstorming fails.

How do you measure an innovation portfolio?

Measure the portfolio, not the projects. Project-level metrics on transformational bets are actively misleading, because you are asking a question the bet cannot answer yet.

• Allocation against target. The share of money and people in each ambition zone, versus the split you committed to. The single most useful number, and the one most portfolios do not track.
• Balance drift. How the allocation has moved over four to six quarters. Drift is slow and always in the same direction, so a point-in-time reading hides it.
• Kill rate. The proportion of bets stopped at review points. A portfolio that kills almost nothing is not being managed; it is being accumulated. Consistently high kill rates in the transformational zone are healthy, not a problem to fix.
• Time to first evidence. How long from funding to the first real signal from outside the building. For transformational bets this is a far better health check than revenue, which will be zero for years regardless of whether the bet is good.
• Candidate supply. How many genuinely distinct new directions entered consideration this cycle. If this is small, every other number downstream is constrained and no amount of allocation discipline will help.

For scoring individual candidates within a zone, the usual frameworks apply, though they are better suited to core and adjacent work than transformational bets: our breakdowns of RICE scoring and the weighted scoring model both work, and both should be used with the honest caveat that scoring a breakthrough bet on confidence and reach mostly measures how familiar it feels.

What software do you need for innovation portfolio management?

Less than vendors suggest, and something different from what most teams buy.

For a portfolio under about thirty active initiatives, a spreadsheet with columns for ambition zone, resource cost, sponsor, stage and last review date does the job completely. The value is in the classification discipline and the standing review, not in the tool. Buying a platform to solve a discipline problem is a common and expensive mistake.

Above that scale, dedicated platforms start to earn their price, mainly for collecting submissions across a large workforce, routing them through consistent review gates, and reporting the allocation split to an executive committee without a monthly manual rebuild. That category, and what it does and does not cover, is broken down on our innovation management software page, with published prices on the idea management software pricing breakdown.

The gap worth naming: every platform in this category is built to manage ideas that already exist. They are pipelines, review workflows and reporting layers. None of them generate candidates, which means if your portfolio is unbalanced because nobody proposed anything ambitious, the software will render that imbalance in a very attractive dashboard and change nothing about it. That is the half a review workflow structurally cannot fill, and it is where a dedicated idea generator does the work: producing genuinely different directions on a challenge, tagged by the angle they came from, so the transformational zone has something in it to fund.

Common mistakes in innovation portfolio management

Counting projects instead of resources. Twelve tiny transformational experiments against three enormous core programs reads as a bold portfolio and is not one. Always weight by money and people.

Letting sponsors self-classify. Ambition ratings drift upward when the person who wants funding assigns the category. Classify in a group, with a skeptic present.

Applying core governance to transformational bets. Quarterly revenue milestones on a five-year option guarantee it dies at the second review. Different zones need different review questions: core asks whether it is on plan, transformational asks what you learned and what it cost to learn it.

Rebalancing once and declaring victory. Drift is continuous and one-directional. Without a standing review, a portfolio rebalanced in January is back where it started by the following January.

Treating the allocation as the whole job. The split describes how you divide the candidates you have. It says nothing about whether those candidates are any good, and a perfectly balanced portfolio of mediocre bets is still a portfolio of mediocre bets.

The part the framework assumes you already did

Every model on this page, the ambition matrix, three horizons, stage-gate, the funnel, starts at the same place: a set of candidate initiatives already exists, and the job is to sort, fund and govern them. That assumption is doing an enormous amount of unacknowledged work.

The evidence on this is fairly consistent. Portfolios are usually unbalanced not because leadership chose caution, but because the transformational column was empty when the allocation meeting happened. You cannot put ten percent of the budget into bets nobody proposed. And the standard way organizations generate candidates, asking teams what they think should be built, reliably produces work adjacent to what those teams already do, since that is the honest answer to the question as asked.

Which makes candidate generation the upstream constraint on the whole discipline. If the wide end of your funnel is narrow, every framework downstream is just arranging a small set of similar options with increasing sophistication. Widen it first, then allocate: the sequence matters, and the frameworks are silent on the first half of it.