The Organizational Pendulum

Strategy has a thesis. This is just a reflex with a budget.

Something goes wrong. A product ships without adequate review. An audit surfaces a gap. A compliance event makes the news. The response is predictable: add process. Add review layers. Add sign-offs. The organization swings toward rigor. And rigor works. For a while.

Then the cost shows up somewhere else. Decisions take quarters. Good ideas die in committee. The people closest to the market can't move without approvals from people who've never spoken to a buyer. The organization starts losing the operators who actually build things.

So the mandate shifts. Move faster. Fewer gates. More autonomy. And that works too. Until scope creeps, portfolios fragment, and two teams ship products that contradict each other because nobody was coordinating. Then something goes wrong again. And the cycle restarts.

In healthcare, the cost isn't efficiency. It's lives. A protocol that takes a decade to adopt, a clinical team gutted by a platform company chasing margin, a governance vacuum that nobody notices until the incident report. The pendulum's price is paid by the people who never see the committee slide.

Both positions feel right from the inside. The people adding governance are right that it was missing. The people cutting it are right that it calcified. Each fix becomes the next problem.

Drag the bob along the calibrated arc to where your organization sits right now. Not where leadership says it is. Where the people doing the work would place it. Most organizations can't see their own position. The people closest to the work can.

O'Reilly and Tushman's argument is structural: organizations that sustain performance over time are structurally designed to explore and exploit simultaneously. Not sequentially. Not in alternating phases. At the same time, within the same entity, with different operating models for each.

Most organizations respond to the swing by choosing a side. New leadership arrives, reads the pain, and prescribes the opposite of whatever the organization just experienced. That feels like progress. It's phase two of the same cycle.

Ambidexterity breaks the cycle by refusing to choose. It builds separate structures for rigor and agility, connects them through shared strategy, and lets each operate with its own cadence, its own metrics, its own definition of speed.

The explore side operates with small teams, fast iterations, short feedback loops, and tolerance for failure. The exploit side operates with defined processes, compliance checkpoints, measurement rigor, and predictability. The mistake most leaders make is trying to run both sides with one operating model. When you apply exploit rules to exploration, you kill it. When you apply explore rules to exploitation, you lose control of it.

March showed in 1991 that exploitation naturally crowds out exploration. Without structural protection, organizations default to what already works. The gravity pulls toward rigor, toward measurement. Exploration dies slowly and invisibly. By the time leadership notices the pipeline is empty, the explorers have either conformed or left.

The goal isn't equilibrium. Weiser and Laamanen argued that equilibrium is dissipative — it requires continuous energy. The moment you stop attending to it, the system drifts. The goal is abatement: reduce the amplitude, shorten the recovery time, build structures that hold both positions.

Post-M&A integration. The acquiring company centralizes for control. The acquired company loses the operational autonomy that made it worth buying. Three years later, they decentralize. The original talent is gone.

I ran a health system through acquisition and stayed as CEO for three and a half years after the close. The centralization pressure was immediate. The talent loss took eighteen months to show up. I watch the same swing from the payer side now, as Staff VP of Carelon Growth at Elevance Health, running product across six specialty risk books: MSK, oncology, CHF, maternity, autoimmune, and dementia.

Clinical operations. Balas and Boren estimated 17 years for evidence to reach practice (Balas & Boren, 2000, Managing Clinical Knowledge for Health Care Improvement). That was in 2000. The number hasn't moved much. Clinical innovation happens at the bedside and gets punished by compliance.

PE-backed healthcare services. The 100-day plan demands discipline. The operators who survive the plan demand flexibility. Three-year hold periods don't allow for a full swing, so the damage compounds faster. I've watched a platform company hit 40% EBITDA margin in year two and lose its best clinical team by year three. The hold period ended. The damage didn't.

Product lifecycle management. The sprint toward launch ignores the governance needed after launch. The governance installed after a failure kills the next sprint. The product organization oscillates between "ship it" and "control it" every 18 to 24 months, and the people in the middle absorb the whiplash.

Ambidexterity is not the answer to everything. A ten-person company has no business running two operating models. A business in genuine crisis should centralize and choose rigor, deliberately, until it survives. A real pivot demands a hard swing with a thesis behind it.

The difference is intent. A swing you chose, for a reason you can name and defend, is strategy. A swing you backed into because the last one hurt is a habit wearing a decision's clothes.

O'Reilly and Tushman proposed structural separation. Kotter proposed dual operating systems. What I couldn't find in any of it was something an operator could pick up and run on Monday.

The real question is how to prevent going so far in either direction that recovery becomes the full-time job. Ambidextrous organizations skip most of that recovery. They built for both from the start. The pendulum still moves. It just doesn't take the whole building with it.