Many companies start out as highly innovative ones. An exciting product, a breakthrough service, a new way of doing something that the market hadn’t seen before. That early energy is what gets them off the ground.
But somewhere along the line, the focus shifts. The organisation gets good at running what it has — and all the energy goes into protecting it. Innovation quietly narrows to process improvements, cost reductions, line extensions, and matching what competitors are doing. When companies do look beyond the core, it’s usually through acquisitions or venture investments that rarely integrate well into the mainstream business.
None of this secures the company’s future. And most leadership teams know it. But knowing it and doing something about it are very different things.
MIT Sloan Management Review just published a piece by Gina O’Connor and Christopher Meyer that takes this problem seriously. O’Connor has been studying how large companies handle breakthrough innovation for over twenty years. Her earlier work — including “Grabbing Lightning” and “Beyond the Champion” — tracked how mature firms try and mostly fail to build lasting innovation capabilities. The findings were sobering: the average corporate ventures group survives about four years before being quietly dismantled. People in innovation roles get treated as organisational misfits. And when a key champion leaves, the capability often goes with them.
This new article, drawing on 640+ interviews across two decades, lays out eight principles for building what the authors call strategic innovation capability — the discipline of turning creative discoveries into new platforms of business that bring real value to both the market and the organisation. Not incremental improvement. Not innovation theatre. A systematic, permanent way of creating new growth.
It’s worth reading because it names a problem most organisations recognise but few have solved: how to keep creating new sources of value when everything about the organisation is optimised for running the current business.
The three stages of innovation — and why they need different playbooks
The article describes innovation as moving through three stages. This isn’t a proprietary framework — it’s a useful way to describe what any innovation process looks like in practice.
Discovery is about spotting and framing opportunities worth exploring. What problems are emerging? Where is the market shifting? What could we do that we’re not doing today?
Incubation is about testing. Can this actually work — technically, commercially, organisationally? This is where you run experiments, build early prototypes, talk to potential customers, and stress-test your assumptions before committing serious resources.
Acceleration is where a validated concept transitions into something the business can run and grow.
What matters is that the first two stages and the third run on completely different logic. Discovery and incubation are about navigating uncertainty. You’re working with assumptions, not facts. You need cheap experiments, fast learning cycles, and the freedom to change direction when the evidence tells you to. Leaders who judge and fund this kind of work the way they would a delivery programme — standard business cases, ROI gates, quarterly milestones — will shut it down before it has a chance to prove itself.
The article also argues — and I think this is one of its most important points — that companies should organise innovation opportunities into portfolios within defined strategic domains, rather than pursuing them as isolated projects. This changes the conversation from “should we fund this one bet?” to “are we building a balanced set of options across the areas that matter most to our future?” That’s a much more productive way to manage both investment and risk.
AI role in this process
One thing the article doesn’t touch is how AI changes the economics of early-stage innovation. The D-I-A model was developed when market probes, experimentation, and validation were expensive and slow. That’s shifted. What used to take a team weeks of desk research, competitive analysis, and customer interviews can now be scoped and pressure-tested in days. AI can significantly compress discovery and early incubation — faster assumption testing, quicker market sensing, cheaper validation loops.
But AI cannot replace human judgement on which problems are worth solving. That still requires training, structured methods, and the space to explore before committing. Teams need the conditions to investigate opportunities properly without prematurely locking in resources on the wrong problem. Get this right and you cut out enormous downstream waste — the expensive kind, where you build the right solution to the wrong problem.
The acceleration gap — where most innovation goes to die
Acceleration is where most innovation efforts fall apart. The concept has been validated. There’s evidence it could work. And then it stalls — or dies — because it has to survive contact with the operating business. O’Connor’s own earlier research documented this clearly: companies would transition incubating businesses into operating units and watch them die on the vine, under-resourced and misunderstood.
The article’s answer is a permanent innovation capability with governance, defined roles, career paths, and portfolio management — so that the system doesn’t depend on individual champions and doesn’t collapse during the next restructure. O’Connor draws an analogy to how marketing became a recognised discipline in the 1970s: innovation, she argues, needs to follow the same path.
I agree with the principle of permanence. Innovation capability that evaporates every time the business comes under pressure isn’t capability at all. Part of the reason is a definition problem: too many leadership teams still equate “innovation” with expensive, uncertain tech bets rather than a disciplined way of creating new value. Anything with that label is an easy target when priorities shift to short-term survival.
Additional perspectives
O’Connor is right that business units left to their own devices might default to incremental improvement without guidance to open up the strategic space and to consider new directions and more innovative business models. That’s well-documented.
But the solution isn’t only to build a separate function that hands validated concepts across to an operating business that didn’t ask for them and isn’t set up to absorb them. What works is a middle path: equip business unit teams to run structured innovation processes themselves — but anchor the work directly in their strategic objectives, not in an abstract innovation agenda set elsewhere. When the outputs of discovery and incubation connect to what a BU is already trying to achieve, adoption isn’t a favour. It’s in their interest.
This means giving those teams the playbooks, methods, and governance to manage early-stage work alongside their BAU responsibilities. Not as a one-off project workflow, but as an ongoing capability embedded in their operating rhythm. Combined with portfolio thinking at the leadership level — so someone is looking across the domains and asking whether the organisation is exploring broadly enough — you get both the BU ownership that makes acceleration work and the strategic breadth that prevents incrementalism.
What “building innovation capability” actually means in practice
When people talk about innovation capability it can sound abstract. In practice it means concrete things: structured processes for each stage, with governance and success criteria that match the level of uncertainty. Playbooks that teams can actually use, not frameworks that sit in a strategy deck. Training so people know how to frame problems, test assumptions, and make evidence-based decisions when they can’t rely on historical data. Performance measurement that fits the stage of the work. AI-augmented workflows that make the early stages faster and cheaper. And leadership that understands the difference between managing risk and managing uncertainty.
This is most relevant for mature organisations that lost their edge, have operating discipline but don’t have a systematic way to find and develop new sources of value especially in times of disruption and crisis. The building blocks are often there. What’s missing is the connective tissue: the process, the governance, and the shared language to make it repeatable.
The research in this article is solid. O’Connor’s two decades of work on this problem are some of the most rigorous in the field. The diagnosis is right. The gap — as always — is in the how.