Review loop
Let changed evidence reopen the choice.
Original choice
Result, assumptions and commitments.
What changed
Customers, economics, capability or constraints.
Decision now
Continue, adjust or replace the choice.
Return to the choice when the evidence changes. Record the reason, owner and next review trigger.
Read the diagram
Compare the original choice with changed evidence; decide whether to continue, adjust or replace it, then set the next review trigger.
01
Start with a choice the plan already made, not the whole plan
A review can walk the entire plan and still reach decisions. This guide takes one choice instead, because one choice can be tested to a decision inside one meeting, and the record below can be repeated for the next choice. Pick the choice with the most money or the most people behind it, or the one whose result is furthest from what was expected. Write it as the plan wrote it: the result it promised, the measure, the date, and the owner. If the plan did not state a measure and a date, the review's first finding is that it cannot yet be tested, and the decision requested is to set them.
02
Separate the assumptions from the facts, then test each one
Every choice rests on a small number of beliefs about customers, costs, capacity and timing. List them as the plan held them, in one line each, with the figure the plan used. Beside each, put what is now known from the operation's own records: orders, costs, throughput, spend and results to date. An estimate that was written down as a fact is not a failure of the plan; finding it is the reason the review exists. Mark each assumption held, not held, or not yet testable, and say which record settled it.
Where a record does not exist yet, the assumption stays not yet testable, and the review asks for the measurement rather than treating silence as success.
03
Price the gap and find the conflict it creates
Where an assumption did not hold, work out what the difference costs at the volume the plan expects, in the plan's own units, and say from which date the volume, the cost and any correction apply. State the components so anyone can repeat the calculation. Say whether the figure is cash or an operating-cost comparison: a cost allocation that falls does not release cash until the spending behind it stops. Then look for the resource the response would need: a team, a budget line, a system change, a supplier. Correcting a belief can compete with something already committed, and the competing commitment has its own value.
Name it, its owner, its date, and what its delay would cost, so that the decision compares two consequences rather than one saving.
04
Ask for one decision and set the trigger that reopens it
Write the decision in one sentence: what changes, what it costs, who approves it, and what it protects. Where a consequence is still unknown, ask for it as a bounded figure before approval rather than deciding around it. Offer the alternative honestly, including keeping the plan as it is with the target moved, and say on what basis any moved target was chosen. Then set a trigger, a measured condition and a date, that reopens the review if the correction does not work, with the threshold on the same side as the target so that meeting the target does not fire it.
Record who watches it, at which existing meeting it is reported, and what the authorized decision-maker does if it fires.
Example review
A direct online channel
A specialty products company approved a three-year plan in January 2026 with three choices. The review, in July 2026, takes one: grow direct online sales from 6 percent of revenue to 25 percent by the end of 2028. The plan's measure is online share of revenue, the owner is the commercial director, and the first milestone is 12 percent at the end of 2026.
- The choice as the plan wrote it:
Direct online sales to reach 25 percent of revenue by December 2028, first milestone 12 percent by December 2026, owner the commercial director.
- Assumptions and what is now known, six months in:
Online share would reach 12 percent by December 2026: online revenue was 9 percent of total revenue for January to June. The plan does not specify whether its December milestone measures that month or the full year, or give a comparable half-year target, so progress against the milestone is not yet testable. The commercial director confirms the measure and comparison (not yet testable).
Cost to serve an online order would be 9 dollars: the operation's own costing of packing labor, parcel handling and returns for the half year puts it at 14 dollars (not held).
The existing warehouse could handle the parcel volume: peak days reached 900 parcels against a measured limit of 1,200 a day, so measured capacity covered the observed first-half peak; the rest of 2026 is not yet tested.
The plan's 2027 volume of 300,000 orders over the operation's 250 dispatch days is 1,200 parcels a day on average if every order ships as one parcel, which is the limit with no room for peaks, and the plan gives no peak projection, so 2027 capacity is untested (not yet testable).
Marketing spend would bring in new online customers at 20 dollars each: 600,000 dollars in the half year brought 21,000 new customers, 600,000 divided by 21,000 is 28.57 dollars each (not held).
- The gap, priced:
Average online order 85 dollars at a 38 percent gross margin is 32.30 dollars of margin per order. At the assumed 9 dollars to serve, contribution per order is 23.30 dollars; at the measured 14 dollars it is 18.30 dollars, a difference of 5.00 dollars per order.
At the plan's 2027 volume of 300,000 orders, that is 1,500,000 dollars of contribution the plan counted for 2027 and the operation will not produce (300,000 × 5.00). This is an operating-cost comparison at the plan's volume, not cash lost from a bank account, because the packing labor inside the 14 dollars is on the payroll whether or not the volume arrives.
The customer acquisition gap, 8.57 dollars per new customer (28.57 less 20), is kept separate so the two causes do not merge; at the measured cost the same marketing budget buys 70 percent of the customers the plan assumed (20 ÷ 28.57).
- The conflict:
Bringing cost to serve down needs a packing-line change: 420,000 dollars one time, expected to take cost to serve from 14 dollars to 11 dollars once it is live, planned for the first quarter of 2027.
The same integration team is scheduled to deliver the ERP inventory module in that quarter, and it cannot do both; the module was approved to reduce stock write-offs, and the cost of delaying it by one quarter has not been quantified. The operations director owns the packing line; the finance systems lead owns the ERP module; both dates are in the approved plan.
- What the packing line is worth, stated carefully:
Three dollars less per order on 300,000 orders is 900,000 dollars a year of modeled operating-cost improvement at the 2027 volume, against a one-time cost of 420,000 dollars. It becomes cash only where the packing labor it replaces leaves the payroll or is not hired as volume grows, and it starts when the line is live and the volume arrives, not at approval.
Which expenditure actually falls is not yet stated by the operations director and is the fourth unknown.
- Decision requested:
Sequence the packing-line change ahead of the ERP inventory module, moving the module by one quarter, subject to the August monthly review. At the August review, compare the cost of delaying each project for the same quarter, using its expected start date, volume and spending that would actually change: the finance systems lead brings the inventory module's figure, the operations director the packing line's.
Keep capacity benefits separate. Approve the sequence only after that comparison; if either consequence is unknown, retain the existing sequence while its owner supplies the missing figure. The commercial director also brings a marketing plan restated at 28.57 dollars per customer.
Targets are management's choices, stated with their basis: the 25 percent result for 2028 is kept because no evidence yet says the market is smaller; the 2026 milestone moves from 12 percent to 11 percent because the half year is at 9 percent and the marketing budget is buying fewer customers than assumed, not because a demand model produced 11.
The alternative, stated honestly: hold the ERP sequence and cap online share at 15 percent, the level the commercial director judges the current cost to serve and warehouse can carry, until cost to serve is solved another way. The CEO decides; the commercial director owns the result; the operations director owns the cost to serve.
- Trigger:
If measured cost to serve is above 11 dollars at the end of June 2027, or has not been measured by then, the review reopens; missing measurement is not success. The FP&A lead reports cost to serve at the existing monthly review from the month the packing line goes live.
If the trigger fires, the commercial director brings the marketing plan back to the same meeting and the CEO decides whether online marketing spend is held at its current run rate; nothing stops automatically.
Keep the next decision clear.
Copy this record into the document you already use with your team. Give an unanswered question an owner and a next action.
Your working record
Strategy review: one choice
- Choice as the plan wrote it (result, measure, date, owner):
- To complete
- Why this choice now:
- To complete
- Assumption 1, the plan's figure, what the records show, status, record used:
- To complete
- Assumption 2, the plan's figure, what the records show, status, record used:
- To complete
- Assumption 3, the plan's figure, what the records show, status, record used:
- To complete
- Gap priced (components, volume, period, cash or operating-cost comparison):
- To complete
- Competing commitment (resource, owner, date, cost of delaying it):
- To complete
- Decision requested (what changes, cost, who approves, what it protects):
- To complete
- Figures still needed before approval, and who brings them:
- To complete
- Alternative stated honestly, and the basis of any moved target:
- To complete
- Result owner and cost owner:
- To complete
- Trigger (measured condition, date), who watches, reported at:
- To complete
- If the trigger fires, who decides what:
- To complete