The Strategy OS: Cadence, Reviews, and Decision Rights

Strategy dies in the gap between the offsite and the grind

Most leadership teams do not have a strategy problem. They have an operating problem. The bets are reasonable, the deck is coherent, and then eleven months later the same three initiatives are still "in progress" with no owner willing to say they are behind. In our reviews of mid-market operating rhythms, the pattern repeats: the annual offsite produces four or five priorities, they get referenced twice in Q1, and by Q3 the weekly meeting is a status parade where nobody decides anything. The cost is not abstract. A 400-person services firm we worked with was carrying 31 named "strategic initiatives" across seven functions; only 9 had a single accountable owner, and 14 had not changed status in two quarters. That is not strategy. That is a backlog with ambition. The failure is rarely in the thinking; it is in the absence of a mechanism that forces the thinking to meet reality on a schedule. When the only forcing function is the annual offsite, everything reverts to the default operating rhythm of firefighting, and the strategy becomes a document people quote rather than a system people run.

A Strategy OS treats strategy as software that runs on a schedule, not a document that gets refreshed once a year. It has three components that must be installed together: a layered cadence so decisions happen at the right altitude and frequency, decision-centric reviews so meeting time buys progress instead of narration, and explicit decision rights so people know who can commit the firm to a course of action. Miss one and the other two degrade. Cadence without decision rights produces well-attended meetings that resolve nothing. Decision rights without a review cadence produce fast calls that never get inspected. The point is the system, not any single ceremony. A firm that installs all three at once will feel the difference within a quarter: decisions that used to drift for weeks get made in the room they belong to, and the leadership team stops relitigating settled calls because the decision-rights map already answered who owned them.

Three layers, three review types, one decision-rights map

The OS runs on three nested loops. The annual loop sets the small number of bets and the guardrails. The quarterly loop converts bets into objectives with owners and a resourcing envelope. The weekly and monthly loops move the work and clear blockers. Each loop has a distinct review type with a distinct output, and each decision is tagged to a decision-rights level so the room knows whether it is informing, recommending, or deciding.

The non-negotiable design rule: every recurring meeting exists to produce decisions, not to redistribute information that could have been read. Status goes in the pre-read; the room is for the calls the pre-read surfaces. A review with no decisions on the agenda should be canceled and replaced with a written update. This single rule, ruthlessly applied, is what separates an operating system from a meeting calendar. In the 400-person firm, applying it cut standing meeting hours by roughly a third in the first month, and the recovered time went straight into the delivery loop where the actual work lives.

Install the OS in one quarter

• Cut the bet list to five. Force-rank the current initiative portfolio and formally sunset or merge everything below the line. In the 400-person example, 31 initiatives collapsed to 6 bets with named owners, and cycle time on cross-functional decisions dropped from an average of 19 days to 6.
• Write a one-page decision-rights map using a RAPID or DACI grid for the top 20 recurring decision types. Publish who Recommends, who Agrees, who Decides, and who is merely Informed. Ambiguity here is where 60 to 70 percent of stalled decisions originate.
• Convert every standing meeting to a decision agenda. Each item names the decision, the recommender, the decider, and the pre-read. If an item has no decision, it moves to async.
• Add kill criteria to every bet. A bet without a written condition under which you stop is a budget line, not a bet. Review kill criteria at each quarterly portfolio review.
• Instrument the cadence itself. Track decisions made per review, average decision latency, and initiatives with a single accountable owner. These three numbers tell you whether the OS is running or decaying.

Where Strategy OS installations quietly fail

• Cadence theater: the meetings exist and are well-run, but no decisions are actually made because decision rights were never written down. The room defaults to consensus, and consensus defaults to delay.
• Too many bets: a firm keeps 12 "priorities" because sunsetting feels like loss. Twelve priorities is zero priorities. The math is unforgiving; attention does not scale with the org chart.
• Review as report-out: pre-reads are not sent, so the first 40 minutes get spent narrating status the group could have read, leaving no time for the decision the meeting existed to make.
• Owner without authority: an initiative has a named owner who cannot commit the resources or make the trade-offs the work requires. Accountability without decision rights is a setup for blame.
• No kill discipline: bets accumulate because nothing is ever stopped, so the portfolio grows every quarter and the resourcing envelope gets thinner per bet until everything is starved.

Ship these in the next 30 days

• Publish a five-bet list with a single accountable owner and written kill criteria for each.
• Distribute a one-page RAPID or DACI decision-rights map for your top 20 recurring decisions.
• Rewrite every standing meeting agenda as a list of decisions with named deciders and pre-reads.
• Kill or merge every initiative that ranked below your top five, and communicate it plainly.
• Start a running log of decisions-made-per-review and decision latency, reviewed monthly.