Explain a calculator's assumptions beside the result
Specify calculator inputs, units, invalid states, rounding and result explanations so users can understand what the output supports and what it excludes.

What must users understand before trusting the output?
A calculator needs a result contract as well as a formula. State what each input means, which values are valid and what the output can support. If “margin” means contribution after variable costs, say so beside the field. Users should not have to infer a financial definition from a short abbreviation.
Google’s people-first guidance emphasizes satisfying the reader’s task. Here, the useful original artifact is a result-copy specification and reproducible fixture. The guidance does not certify the calculation, and a calculator interface does not transform an illustrative model into business evidence. [Creating helpful, reliable, people-first content]
A result-copy specification for break-even ROAS
Illustrative model: break-even return on ad spend equals 1 divided by the contribution margin fraction before advertising. This simplified model assumes the contribution boundary already includes relevant variable non-ad costs. It excludes fixed overhead, financing, retention effects and uncertainty unless those are explicitly incorporated elsewhere.
| State | Required behavior | User-facing explanation |
|---|---|---|
| Valid margin: 40% | Calculate 1 ÷ 0.40 = 2.5 | Break-even ROAS is 2.50× under the stated cost assumptions |
| Margin: 0% | Do not divide by zero | No positive contribution is available to cover advertising |
| Margin: negative | Reject for this simplified model | This model requires a positive contribution margin |
| Margin above 100% | Reject | Enter a percentage within the supported range |
| Blank or nonnumeric input | Do not substitute zero | Enter a contribution margin to calculate a result |
| Display rounding | Round output only | Calculation uses the unrounded input |
A 2.50× result means $2.50 in attributed revenue for each $1 of ad spend under this model. It is not a promise of profitability for the company. Google Ads defines ROAS through reported conversion value relative to ad cost; your chosen value and margin boundaries still determine whether a business interpretation is valid. [About Target ROAS bidding]
Test inputs that expose ambiguity
- A 40% input returns 2.50× and 250% only if the display clearly labels both units.
- A 25% input returns 4.00×, with the same cost boundary and explanation.
- A blank field leaves the result unavailable rather than showing 0.00×.
- A decimal-comma paste is either parsed deliberately or receives a recoverable explanation.
- Changing an input updates the result and its accessible announcement without moving focus unexpectedly.
Keep test numbers independent of presets. A suggested margin might look authoritative even when it is only a teaching default. Label it illustrative, allow replacement and avoid implying that it is typical for a reader’s industry. If a result depends on revenue being net of refunds or taxes, put that boundary in the visible instructions.
What should the result tell users to do?
Connect the output to a bounded decision. A break-even model can help compare a planned target with included variable costs. It cannot establish incremental demand or predict a campaign’s response to more spending. Give users the assumptions they need to review before taking the next action.
For a downloadable result, preserve the timestamp, inputs, units, formula version and exclusions. Do not export a standalone number that loses its meaning outside the interface. The same principle applies to a shared screenshot: the key assumption should remain visible with the value.
I would test the explanation with the awkward states before polishing the chart. Ask someone to explain what a zero, blank and unavailable result mean. If those answers differ from the specification, fix the copy or behavior. A technically correct division is only one part of a trustworthy calculator.
Sources
AI-assisted writing and an AI-generated editorial cover. Source links and checked dates are listed below; illustrative examples are labeled in the text.
- Creating helpful, reliable, people-first content — checked 2026-10-18
- About Target ROAS bidding — checked 2026-10-18