CPA BAR Forecasting: Read Assumptions Before the Model (2027 Prep)

Before comparing forecasts, identify the period, the result being forecast and the assumptions that connect the inputs to that result. A larger sales total can accompany a smaller operating result when price, unit cost and fixed costs change together. The arithmetic can be correct. You still need evidence for the assumptions.

Our original Cedar Bookmark example compares two fictional forecasts for the same coming quarter, then changes one input at a time. No probabilities are assigned. The original forecast inputs and outputs (CSV) retain each calculation, and the fictional assumption register (CSV) distinguishes supplied teaching assumptions from facts that would need verification in real work.

Define the result before comparing the totals

Cedar's simplified model calculates sales revenue as units sold times selling price, subtracts units sold times variable cost per unit, then subtracts quarterly fixed costs. The output is an illustrative operating result. It is not cash flow. Collection dates, inventory movements, financing and taxes are outside this small model, and no claim about a complete financial forecast follows from its result.

In Forecast A, Cedar assumes 1,000 bookmarks sold at $12 each, a variable cost of $7 per bookmark and $3,000 of quarterly fixed costs. Revenue is $12,000. Variable costs are $7,000, leaving $5,000 before fixed costs and an illustrative operating result of $2,000.

Those numbers are invented inputs, with no historical data supplied to estimate demand or costs. Capacity is also unverified. Treating the output as a forecast that has been validated would require work beyond multiplying the figures in our record.

Follow the assumptions behind the second forecast

Forecast B assumes 1,200 bookmarks sold at $11.50, a variable cost of $7.25 per bookmark and $3,300 of fixed costs for the same quarter. Revenue reaches $13,800. Variable costs become $8,700, leaving $5,100 before fixed costs and an operating result of $1,800 in this simplified model.

The sales increase is $1,800, or 15 percent of Forecast A's $12,000 revenue. The operating result falls $200. That is a 10 percent decrease from A's $2,000, despite selling 200 additional units, so a claim that B improves both sales and this result would contradict the supplied arithmetic.

Both forecasts are conditional. Nothing in the example establishes that discounting will generate the assumed extra sales, that all those sales can be fulfilled or that the changed costs are complete. The register leaves the actual supporting evidence and validation result blank.

Separate a one-input check from a changed scenario

For a one-input volume sensitivity check, keep A's price, variable unit cost and fixed costs unchanged while increasing units sold to 1,100. Only volume changes. Revenue becomes $13,200 and variable costs $7,700, producing a $2,500 operating result after the unchanged $3,000 of fixed costs.

For a separate price check, return every input to A, then increase price to $12.50 while keeping sales at 1,000 units. The result is again $2,500. Equal dollar changes in these two chosen checks do not make price and volume interchangeable, and different test increments would produce different comparisons.

Forecast B changes several assumptions together, whereas each sensitivity check isolates one chosen input within A's model. Neither estimates demand response. Holding volume fixed while changing price is an arithmetic condition of our sensitivity check, not evidence that actual customers would buy the same quantity at the higher price.

Explain the difference with an ordered bridge

The ordered result bridge (CSV) starts at A's $2,000 operating result and changes volume first, then price, variable unit cost and fixed costs. Order is stated explicitly. Increasing volume by 200 at A's $5 contribution per unit adds $1,000, bringing the result to $3,000 before any other input changes. At the new 1,200-unit volume, reducing price by $0.50 subtracts $600, and increasing variable cost by $0.25 subtracts another $300. Fixed costs rise $300. After those changes, $2,000 plus $1,000 minus $600 minus $300 minus $300 equals B's $1,800.

That bridge reconciles the totals. Its allocation depends on the chosen sequence because price and volume interact in the revenue calculation, so the individual bridge amounts are not unique causal contributions. It explains this arithmetic path without proving why sales, prices or costs would actually change.

Give each assumption a source and a reason to revisit it

The authors recommend documenting and justifying assumptions in their forecasting principles and scenario discussion, retaining forecasts for evaluation when observations become available. A scenario can be unlikely. Their scenario discussion considers plausible alternative futures without requiring each particular case to describe a likely outcome. Our two cases have no assigned likelihood, and their endpoints do not form a statistical prediction interval.

Start an assumption record with the quantity, unit, period and model version so another reader can reconstruct the calculation.

The source field has a purpose. In real work, an approved price list could support an offered price while leaving expected sales volume unresolved, and an old supplier quote could require a new validity check before supporting a coming-quarter cost.

Our fictional register keeps proposed evidence checks separate from actual evidence and validation. Every actual result is blank. No price list, demand study, supplier quote or capacity record has been supplied, and an author-created number cannot become verified business evidence because its formula reconciles.

Keep the model's unanswered questions in the explanation

Use the blank assumption record (CSV) to record which assumption would change your interpretation and what evidence would let you reassess it. Cedar's volume increase needs support. The example also leaves capacity, the relationship between price and demand, and the completeness of quarterly costs unverified, so choosing B solely for its higher revenue would skip questions the model cannot answer.

The arithmetic checks are complete. Actual forecasting accuracy remains unknown until a defensible forecast is compared with relevant observations for the same scope and period. No historical series, error measure or realized outcome is supplied here, and the example supports no model-accuracy ranking or business recommendation.

The official exam-content page lists editions with separate effective dates, so use dated, authorized BAR instruction for the attempt you are preparing for. Detailed task alignment remains unverified. We have not inspected linked blueprint PDFs or certified this tutorial against a particular forecasting task, and a preparation-year title cannot establish exam coverage.

2027 preparation edition. Public informational HTML checked October 12, 2026. Cedar Bookmark, inputs, forecasts, sensitivities and bridge are original fictional teaching material. Same coming quarter assumed. Outputs are illustrative operating results, not complete financial forecasts, cash flows or business recommendations. No demand, capacity, price response, probability or source validation established. Actual evidence, validation results and review dates remain blank. No linked blueprint PDF, worksheet, sample or assessment inspected. Matching BAR hub, detailed task alignment and current-source/CMS/live checks required before publication.