Sample frameDay 337·Innovation · Structure

Good morning. The new thing isn't dying of a bad idea. It's dying of the machine you're running it on.

Innovation.

The Ambidextrous Organisation

5 min read·Apply by lunch

The question

The new thing you're building inside the existing business — is it measured on the same numbers, staffed from the same pool and reviewed in the same meeting as the part that already works, and if it is, who exactly is going to protect it the first quarter the core misses its forecast?

The idea

The canonical answer to housing a new business inside an old one is to cut it loose: a separate unit, its own metrics, insulated from the parent's immune system. That's right, and on its own it's fatal. Michael Tushman, then at Columbia, and Charles O'Reilly of Stanford's Graduate School of Business made the ambidextrous organisation a working idea in California Management Review in 1996, and then went and measured what separation actually does. With Wendy Smith, Robert Wood and George Westerman they examined thirty-five attempts to launch breakthrough innovations across fifteen business units in nine industries, and found four structures in use. Seven ran the new work inside the existing functional organisation. Nine used cross-functional teams sitting outside the management hierarchy. Four were cut loose entirely as unsupported teams. Fifteen used an ambidextrous design: structurally independent units with their own processes, structures and cultures, reporting into the existing senior management hierarchy. More than ninety per cent of the ambidextrous attempts reached their goals. The cross-functional and unsupported teams — separated, with no line back to the top — reached none. Only a quarter of the functional designs did. Nor did the core pay for it: under ambidextrous designs its competitive performance almost always held or improved, which is not what happened under the other three.

The framework

01
Separate the metrics before you separate the org chart.
Tushman and O'Reilly's list of differences runs longer than most people expect: strategic intent, critical tasks, competencies, structure, controls and rewards, culture, and the role the leader plays. Cost and profit on one side, innovation and growth on the other. Formal and mechanistic against adaptive and loose. Margin and productivity against milestones and learning. Of all of them the controls decide the outcome, because a unit judged on the core's margins in the core's monthly review will converge on the core's behaviour inside two quarters, whatever the boxes on the chart say. You don't need to move a desk to lose this; you only need to keep the dashboard.
02
One senior owner with a foot in both businesses.
At Ciba Vision in the early 1990s, president Glenn Bradley launched six breakthrough projects and cancelled dozens of small conventional-lens initiatives to pay for them. The move that made it survivable was narrow: every project leader reported to one executive, Adrian Hunter, the R&D vice-president, chosen because he knew the existing business intimately and had relationships across it. Hunter's job was to manage the conflicts between old and new rather than let them be settled by whoever had more revenue. The project leaders also sat in on Bradley's executive team meetings, because that's where the trades get made and a unit that isn't in the room doesn't get any.
03
What happens when you separate them too well?
In 1995 Tom Curley, president of USA Today, put Lorraine Cichowski in charge of an online news service, gave her free rein and put the team on a different floor. She built a genuinely different organisation. It stalled. Cut off from the newspaper, the unit couldn't draw on its reporting, and print executives who read it as a competitor had no reason to help. It ran short of cash and lost staff. Cichowski wanted to spin the business out; Curley concluded the opposite — that it needed more integration, not less — and rebuilt it on that basis. Autonomy isn't a resource. It's a permission, and it doesn't feed anybody.
04
Pay the top team on the whole, not on their parts.
Curley replaced unit-specific goals with a single bonus tied to growth across the newspaper, the website and the television stations, changed promotion criteria to reward sharing, and let go of roughly forty per cent of his executive team — the members who wouldn't back the strategy. Ciba Vision did much the same, rewarding managers primarily on company performance rather than their own unit's. Both leaders also gave people one sentence that covered both halves of the business. That's what stops organisational separation curdling into organisational fragmentation: shared rewards at the top, and a reason the two halves belong to each other.

Members only · 4 principles + template + AI mentor

364 more frameworks are waiting.

The Ambidextrous Organisation is Day 1 of 365. One framework every morning for a year — across strategy, sales, negotiation, leadership, and more.

$1 /day

Billed $365/yr · cancel any time

365 frameworks

12 topics · templates · AI mentor

Reserve your place — $365/yr →

One payment a year · the whole library · no auto-renewal.

More Innovation frameworks

All 24 innovation frameworks →