Hypothesis register and prioritization: what to do first with limited hours and money
From a register of hypotheses (effect, confidence, cost, working hours, time to effect) calculates the expected net result of each hypothesis after cost and the return per hour, drops those whose effect comes after the horizon or does not repay the cost and builds a plan in descending return per hour while the team hours and the budget last. Shows which hypotheses made it, which did not because of capacity and what part of the effect arrives in the first 72 hours.
At a glance
- What question does it answer?
- Which hypotheses to take into the plan to get the most expected money per hour of work and stay within the budget.
- What data is needed?
- The hypothesis register: action, money effect, confidence, cost, hours, time.
- How to read the result?
- Example on the default values (10 hypotheses, 120 hours, a budget of RUB 800,000, a 90-day horizon): 6 hypotheses made it with an expected net of RUB 5.27 million at 114 hours; 2 lacked hours or budget, 1 dropped by timing. In the first 72 hours we expect RUB 1.81 million. Example on the default values: Hypotheses in the register: 10; Fit the horizon and pay back: 8; In the plan: 6; Expected net of the plan after cost: RUB 5.27 million. The figures are illustrative: enter your own to see your result.
Why and how to reason
In a crisis it seems you must do everything at once, but the team has few hours and little money for launches and each hypothesis takes both. So you compare not “what brings more” but “what gives more per hour”. The expected effect is the effect times an honest probability: a hypothesis worth RUB 2,000,000 at 65% confidence gives an expectation of RUB 1,300,000. Subtract the launch cost and divide by the hours of work. Moving part of the supplier orders gives RUB 1,300,000 for 14 hours, that is RUB 92,857 per hour. A reorder of fast styles looks larger (RUB 3,200,000) but at 40% confidence and a cost of RUB 600,000 it leaves RUB 680,000 for 40 hours, that is RUB 17,000 per hour. Then go in descending return while hours and budget last. With 120 hours and RUB 800,000 six hypotheses made the plan: a net of RUB 5,265,000 at 114 hours and a cost of RUB 230,000.
When to use it
Use it when there are more hypotheses than can be done and you must quickly agree in the team what to take first: in a crisis, at a season launch, in an analysis of falling sales. Do not use it when the effect cannot be expressed in money (reputation, an agreement with a key partner) or when hypotheses depend on each other: then count them as a bundle in one row. Assumption: effects add up and confidence is honestly assessed. The calculation orders, it does not replace a test: before launching a large hypothesis check it with a small one.
How to read the result
Look at three numbers. The expected net of the plan shows how much money to expect and does not promise it. “In the first 72 hours” says what the fastest brings: here RUB 1,810,000 from advertising and the price of hits. “Left out by hours or budget” is the queue of the next round: here the reorder of fast styles and the platform cards; if the queue holds hypotheses with a high return, think how to add hours or money. Those dropped by timing are not bad: they move into the yearly strategy. If hour usage is above 95% the team is overloaded and the plan risks not being completed; leave a reserve. After launch record the fact in the register and recalculate confidence.
Common mistakes
The most frequent error is to give all hypotheses 70-90% confidence: then the order is set only by effect and hours, not probability. The second is not to count hours, and the plan promises the impossible. The third is to take an effect “for a year” instead of the effect over the horizon, which makes far ideas look better. The fourth is to count cost as the only price of a hypothesis and forget that people’s time is limited too. An empty confidence is counted as zero.
What data is needed?
- The hypothesis register: action, money effect, confidence, cost, hours, time.
- The team hours and the budget for the planning horizon.
- Results of past tests to assess confidence.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Team hours for the plan | h | 120 | How many person-hours the team can really give to hypotheses over the horizon without pulling people off operations. Count honestly: in a crisis there are fewer hours than it seems. The example value is illustrative. |
| Budget for one-off costs | ₽ | 800,000 | How much money can be spent on launching hypotheses (purchase, advertising, discount, contractors) without breaking the cash plan. The example value is illustrative. |
| Planning horizon | days | 90 | The effect of a hypothesis that arrives after the horizon is not taken into the plan: in a crisis the money of the nearest weeks matters. The example value is illustrative. |
How it is calculated
| Step | Formula |
|---|---|
| Expected effect | expectation = effect × confidence |
| Net after cost | net = expectation − one-off cost |
| Screening | drop hypotheses with a net ≤ 0 and with an effect after the horizon |
| Return per hour | return = net ÷ hours of work |
| Building the plan | go in descending return, take a hypothesis if the hours and budget are not exceeded, otherwise skip it and go on |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Hypotheses in the register
- 10
- Fit the horizon and pay back
- 8The effect arrives within the horizon and the expected net after cost is positive.
- In the plan
- 6
- Expected net of the plan after cost
- 5.27 m ₽The sum of “effect × confidence − cost” over the plan hypotheses: an expectation, not a promise.
- Hours in the plan
- 114 hOut of the available team hours.
- Hours used
- 95 %
- Plan cost
- 230,000 ₽Out of the budget for one-off costs.
- Expected effect per ruble of cost
- 23.9×
- Left out by hours or budget
- 2Good hypotheses that lacked capacity: candidates for the next round or outside help.
- Dropped: effect after the horizon
- 1Not bad ideas but not crisis ones: they move into the yearly strategy.
- Expected effect in the first 72 hours
- 1.81 m ₽Plan hypotheses whose effect is visible within three days.
- Best return per hour
- Move part of the supplier orders to the next season · 92,857 ₽/h
Data as a table
| Measure | Net after cost |
|---|---|
| Move part of the supplier orders to the next season | 1.3m |
| Narrow advertising to channels where the ad share is not above the target | 1.26m |
| Raise the price of the hits by 3% | 550k |
| Sell dead stock through channels outside the main one | 1.29m |
| Preorder for the loyal base with a discount | 415k |
| Renegotiate commission and logistics with the platform | 450k |
| Reorder fast styles in the selling sizes | 680k |
| Fix product cards and photos on the platform | 265k |
| Reduce the rented space | −75k |
| A new product line | −300k |
Data as a table
| Number of the hypothesis in the plan | Cumulative net |
|---|---|
| 0 | 0 |
| 1 | 1.3m |
| 2 | 2.56m |
| 3 | 3.11m |
| 4 | 4.4m |
| 5 | 4.82m |
| 6 | 5.27m |
Register after prioritization
| Hypothesis | Expected effect | Net after cost | Hours | Return per hour | Days to effect | Status |
|---|---|---|---|---|---|---|
| Move part of the supplier orders to the next season | 1.3 m ₽ | 1.3 m ₽ | 14 | 92,857 ₽ | 21 | in the plan |
| Narrow advertising to channels where the ad share is not above the target | 1.26 m ₽ | 1.26 m ₽ | 16 | 78,750 ₽ | 3 | in the plan |
| Raise the price of the hits by 3% | 550,000 ₽ | 550,000 ₽ | 10 | 55,000 ₽ | 3 | in the plan |
| Sell dead stock through channels outside the main one | 1.44 m ₽ | 1.29 m ₽ | 30 | 43,000 ₽ | 14 | in the plan |
| Preorder for the loyal base with a discount | 495,000 ₽ | 415,000 ₽ | 20 | 20,750 ₽ | 10 | in the plan |
| Renegotiate commission and logistics with the platform | 450,000 ₽ | 450,000 ₽ | 24 | 18,750 ₽ | 30 | in the plan |
| Reorder fast styles in the selling sizes | 1.28 m ₽ | 680,000 ₽ | 40 | 17,000 ₽ | 45 | no hours or budget left |
| Fix product cards and photos on the platform | 315,000 ₽ | 265,000 ₽ | 36 | 7,361 ₽ | 20 | no hours or budget left |
| Reduce the rented space | 325,000 ₽ | −75,000 ₽ | 60 | −1,250 ₽ | 75 | does not pay back |
| A new product line | 1.2 m ₽ | −300,000 ₽ | 160 | −1,875 ₽ | 120 | effect after the horizon |
2 hypotheses with a positive net did not fit the limits: add hours and budget or leave them for the next round.
The calculation answers “what to do first”, not “what will bring the most”: a large hypothesis with a low return per hour and a long time yields to small fast ones. Assess confidence without a test strictly: overstated confidence distorts the order most. For footwear, accessories and children’s the hypotheses and timings differ: footwear has a long order cycle, children’s has a school season.
Nuances and limits
- Giving every hypothesis 80-90% confidence: the order loses meaning.
- Taking hypotheses with an effect in half a year into a crisis plan.
- Not counting hours: the plan overloads the team and nothing gets finished.