Strategy as a Living System

The annual ritual

Every year, the same thing happens. Senior leaders disappear into an offsite for two days. They argue about priorities, negotiate resource allocations, and produce a strategy deck. The deck gets socialised. Town halls are held. Posters go up. OKRs are cascaded. And then the organisation does what it was already doing, with slightly different labels.

We’ve watched this cycle play out in dozens of organisations, and the pattern is remarkably stable. The strategy deck is maximally relevant the day it’s published. From that point forward, it degrades. Market conditions shift. Competitive dynamics change. Customer behaviour evolves. Internal capability develops in ways that weren’t anticipated. By month three, the deck is a historical artefact. By month six, it’s a fiction.

But the organisation keeps executing against it, because the plan is the plan.

What we did differently

We worked with three organisations, two in financial services, one in healthcare technology, to test an alternative model. Instead of an annual strategy cycle producing a static document, we helped them build a continuous system. Strategy became something the organisation does, not something it has.

The core shift: every strategic decision was reframed as a falsifiable bet with explicit evidence thresholds and live data connections.

This sounds straightforward. It was not.

The mechanics of a strategic bet

A strategic objective like “accelerate digital adoption” is untestable. You can’t falsify it. You can’t measure evidence for or against it with any precision. It’s a direction, not a hypothesis.

A strategic bet looks different. One of our financial services partners had “accelerate digital adoption” as a strategic pillar. We worked with their leadership team to decompose it into specific bets:

Bet 1: “If we reduce the digital onboarding flow from 14 steps to 5, first-time digital completion rates will increase from 23% to above 50% within four months.”

Bet 2: “If we introduce proactive digital servicing alerts, customers who receive them will generate 30% fewer inbound calls within six months.”

Bet 3: “If we sunset three legacy paper-based processes, the cost-to-serve for affected customer segments will decrease by at least 15% without increasing complaint volumes.”

Each bet has a thesis that can be proven wrong. Each has a timeframe. Each specifies the evidence that would confirm or deny it. And critically, each has a kill threshold: the point at which the evidence says stop.

Evidence thresholds versus status reports

The difference between a bet and a traditional strategic initiative comes down to how you know whether it’s working.

Traditional initiatives report status. Green, amber, red. On track, at risk, off track. These assessments are subjective. They’re produced by the people doing the work, who have every incentive (consciously or not) to present progress favourably. We’ve documented this pattern extensively: initiative owners rate their work an average of 23 percentage points more favourably than external evidence-based assessment would support.

Bets use evidence thresholds instead. For Bet 1 above, the evidence stream was straightforward: weekly digital completion rates by step, segmented by customer cohort. The team didn’t need to “report status.” The data spoke. At week six, completion rates had moved from 23% to 34%. Positive trajectory, but below the pace needed to hit 50% by month four. The system flagged this automatically. No status report required. No filtering through management layers. The evidence was visible to everyone who needed to see it.

This is where it gets interesting. In a traditional model, that initiative would have been reported as “on track” at the quarter mark. The team was making progress. The direction was positive. The absolute number was growing. A reasonable manager would have written “green” on the status slide.

The evidence threshold told a different story. At the current rate of acceleration, the bet would reach roughly 41% by month four, well short of the 50% threshold. This wasn’t failure. It was information. The leadership team used it to diagnose what was happening (the five-step flow had a friction point at step three that the original design hadn’t anticipated) and make an adjustment two months earlier than a quarterly review would have surfaced the issue.

Continuous confidence scoring

Each bet carried a confidence score, updated continuously based on incoming evidence. We used a simple framework:

High confidence (above 70%): Evidence is accumulating in the direction the thesis predicted. Assumptions are holding. Continue and consider increasing investment.

Medium confidence (40-70%): Mixed signals. Some evidence supports the thesis, some doesn’t. Assumptions may be weakening. Diagnose before deciding.

Low confidence (below 40%): Evidence is contradicting the thesis, or key assumptions have broken. The bet needs to be restructured or killed.

The scores weren’t generated by algorithms. They were human judgements, but human judgements anchored to specific evidence streams rather than to narratives, gut feelings, or political dynamics.

Our healthcare technology partner found this particularly powerful. Their strategic portfolio contained 11 active bets. At the start of the engagement, leadership would have rated eight of them as “on track” based on their existing reporting. After six weeks of evidence-based scoring, four were in the medium confidence range and two had dropped below 40%.

One of the low-confidence bets had been running for 14 months. It was a $3.2 million investment in a partner integration platform that the quarterly review had rated as “on track” every single quarter. The project team was hitting their milestones. Code was being written. Features were being delivered. But the bet’s thesis, that the partner integration would drive a 20% increase in platform stickiness, had zero supporting evidence. Partner adoption was flat. The features being built weren’t being used. The milestone-based reporting had created an illusion of progress by measuring output (features delivered) rather than outcome (the thesis being validated).

They killed it. Redirected the team and budget to a bet with strong evidence. Fourteen months of quarterly reviews had missed what six weeks of evidence-based scoring made obvious.

What changes about governance

When strategy becomes a living system, the governance model has to change with it. The three organisations we worked with all converged on similar patterns.

Cadence shifts. The quarterly strategy review didn’t disappear, but it changed in character. It became a portfolio review: which bets are we making, what does the evidence say, what adjustments are needed? The real strategic governance happened in shorter cycles. Fortnightly evidence reviews for high-priority bets. Monthly portfolio rebalancing conversations. The quarterly session became a synthesis point, not the primary governance mechanism.

Accountability shifts. In the traditional model, initiative owners are accountable for delivering outputs. In the bet model, the accountability question is different: is evidence accumulating? If not, why not? This changes the incentive structure. An initiative owner in the traditional model is rewarded for completion. A bet owner in the new model is rewarded for learning, including learning that the bet is wrong.

One of our financial services partners struggled with this transition. Their culture had deeply embedded “delivery equals success” as a value. Telling a senior leader that killing their bet was a good outcome required real cultural rewiring. It took about four months before the first leader voluntarily recommended killing their own bet based on the evidence. After that, the cultural shift accelerated.

Information flow changes. The most profound shift was in how information moved through the organisation. In the traditional model, information flows up through management layers, getting summarised and smoothed at each boundary. In the bet model, evidence streams are visible to everyone with a stake in the bet. A frontline team member working on Bet 1 could see the same completion rate data as the CEO. The management layer still added interpretation and context, but the raw evidence wasn’t filtered through it.

The zombie problem

All three organisations discovered the same thing within the first three months: they were investing significant resources in strategic initiatives that had no evidence of working.

We started calling these zombie initiatives. They weren’t dead, they were still consuming resources, producing outputs, filing status reports. But they weren’t alive either. No evidence was accumulating that their thesis was correct. They were the walking dead of the strategic portfolio.

Across the three organisations, zombie initiatives represented between 18% and 31% of total strategic investment. One organisation was spending $8.4 million annually on four initiatives that had produced no evidence of thesis validation in over a year.

The traditional governance model couldn’t see them because it measured the wrong things. Milestones hit? Yes. Budget on track? Mostly. Team engaged? Absolutely. The people working on zombie initiatives were working hard. They were just working hard on things that weren’t producing the outcomes the organisation needed.

Live evidence made the zombies visible. That visibility is uncomfortable. Killing initiatives means telling people their work didn’t pan out. It means reallocating resources and restructuring teams. It means admitting that a decision made 12 or 18 months ago was wrong. No executive enjoys that. But the alternative, continuing to invest millions in work that isn’t producing results because the reporting system can’t see the problem, is worse.

From deck to system

The core argument is simple. Strategy stored in a deck is static. The world it describes is dynamic. The gap between the two widens every day after publication. Within months, the deck is a fiction that the organisation treats as fact.

Strategy as a living system closes that gap. Bets replace objectives. Evidence replaces status reports. Continuous confidence scoring replaces quarterly red/amber/green assessments. Kill thresholds replace the sunk cost fallacy.

The mechanics are straightforward. The cultural change is hard. The three organisations we worked with all experienced resistance, particularly around killing initiatives and around the transparency that live evidence creates. When everyone can see the same data, the political dynamics around strategic governance change. Leaders who previously controlled the narrative through selective reporting found themselves working with shared evidence. Some found that liberating. Some found it threatening.

What we learned

Six findings from the three engagements:

1. Evidence availability is the bottleneck. Most organisations don’t lack strategy. They lack the evidence infrastructure to test it. Building the data connections to feed evidence into bet scoring took longer than reframing the bets themselves. In one case, it took eight weeks to wire a single customer behaviour metric into the evidence system because the data sat across three different platforms with no integration layer.

2. Kill thresholds must be set in advance. If you wait until the evidence is bad to decide what “bad enough to kill” means, you’ll never kill anything. The negotiation over what constitutes failure is politically charged. Setting that threshold at the start, when the bet is still abstract and nobody’s identity is tied to it, is dramatically easier.

3. Confidence scores create better conversations. The language of confidence (72% and falling, 45% and stable, 61% and rising) gave leadership teams a shared vocabulary for strategic discussion that “on track” and “at risk” never did. It made disagreement productive. Two executives could look at the same evidence and disagree about whether 52% confidence warranted continued investment, and that disagreement could be grounded in specifics rather than intuitions.

4. Zombie initiatives are universal. Every organisation we’ve worked with has them. The proportion varies, but the pattern doesn’t. Static strategy governance, measured by output rather than outcome, will always accumulate zombie initiatives. It’s a structural inevitability, not a management failure.

5. Transparency changes politics. When evidence is visible to everyone, the dynamics of strategic governance shift. This is mostly positive, but it requires deliberate management. Some leaders will feel exposed. The transition needs support.

6. Annual strategy isn’t dead, but it needs a different role. The annual cycle still serves a purpose: stepping back, reviewing the portfolio holistically, making major directional choices. What it can’t do is govern. Governance needs to be continuous, because the evidence is continuous and the world doesn’t pause between quarters.

The structural argument

Strategy treated as an event (the offsite, the deck, the annual plan) will always degrade. This is not a solvable communication problem. It’s a structural limitation of static artefacts in dynamic environments.

Strategy treated as a living system, with falsifiable bets, live evidence, continuous scoring, and explicit kill thresholds, evolves with the environment it operates in. It’s harder to build. It requires infrastructure, cultural change, and governance redesign. But it’s the only model we’ve seen that reliably identifies zombie initiatives, surfaces evidence early enough to act on it, and turns strategic governance from a quarterly theatre production into an ongoing discipline.

The three organisations we worked with are still early in this transition. But two of them identified and killed multi-million dollar zombie initiatives within six months, initiatives that their traditional quarterly reviews had rated as “on track” for over a year. That alone justified the investment. The longer-term value, a strategic portfolio that learns and adapts continuously, is still being proven. But the evidence is accumulating.