# Nanny share calculator test: protocol

Version 2, 2026-11-13 (version 1 the same day; changes are listed at the end). Written before any tool was scored for the article. Conflict of interest: Sharefully publishes one of the calculators tested. The protocol, the raw outputs and the screenshots are kept so anyone can repeat the test.

## Tools
- DaycareCalc: https://daycarecalc.com/vs/nanny-share
- ChildCareCostCalc: https://childcarecostcalc.com/calculators/nanny-share-calculator
- ChildCareCostFinder: https://childcarecostfinder.com/nanny-share-calculator/
- Nest Payroll: https://nestpayroll.com/nanny-share-calculator
- FullCalculator: https://fullcalculator.com/finance/nanny-share-cost-calculator
- Sitter Rank: https://www.sitterrank.com/tools/nanny-share-cost-calculator
- CalculatorFoundry: https://www.calculatorfoundry.com/nanny-share-calculator/
- Cubb Nanny: https://www.cubbnanny.com/guides/nanny-share-cost-calculator
- BestBabyCalculators: https://bestbabycalculators.com/nanny-share-calculator/
- Daycare Cost Guide: https://daycarecostguide.com/nanny-share-calculator
- Fair Share Rate Calculator by Sharefully: https://sharefully.app/calculator . Test it exactly like the others, through its page, in both runs (inputs: single-family rate $24, extra child $3, the two-thirds rule). `sharefully-reference.json` holds API outputs for cross-checking only.
Add any other free public nanny share rate or cost calculator you find, up to five more.
- NannyFinder: https://nannyfinder.io/charlotte/nanny-share#calculator . Added later on 2026-11-13 at the request of Sharefully's founder, after the first eleven tools were scored. Same cases, tests and evidence rules; nothing else changed.

## Cases
Use the tool's real page in a browser, as a visitor would. Enter exactly these inputs. Where a tool needs something else (state, weeks, hours), use 40 hours a week and the tool's default for the rest, and record every default you left in place.
- **C1** Nanny's one-family rate $24 an hour. Two families, one child each.
- **C2** Same, but Family A has two children and Family B has one. If the tool has an extra-child amount, use $3.
- **C2s** C2 with the families swapped (A one child, B two), or run from "the other family's" seat if the tool has a "you" and an "other".
- **C3** Three families, one child each, $24.
If a case cannot be entered, write "not supported" and say why. Never guess a number: record only what the page shows.

## What to record per case
The nanny's total hourly pay (or weekly or annual wages before taxes, if that is what is shown), each family's cost, whether taxes or fees are included in the figures, and a full-page screenshot.

## Tests, answered from the recorded numbers (pass, fail, not shown, not supported; one sentence of evidence each)
- **T1 More for more work.** In C1 the nanny's hourly pay is higher than the $24 she would earn from one family.
- **T2 Less than on your own.** In C1 each family pays less per hour than $24.
- **T3 It adds up.** In C1 and C2 the families' parts sum to the nanny's pay (within a cent per hour, before taxes).
- **T4 Same answer from either seat.** C2s gives the mirror image of C2 and nothing else changes.
- **T5 Siblings.** In C2 the family with two children pays more than the family with one, but less than twice as much, and each family still pays less than it would alone (if the tool shows a solo figure).
- **T6 A third family.** In C3 the nanny's pay is higher than in C1 and each family pays less than in C1.
- **T7 The share sets the rule.** The share premium or split rule can be changed by the user, and the page says what it is.
- **T8 Starts from the nanny's rate.** The main input is the nanny's own rate for one family (not a family's budget or a market average the user cannot change).

## Features to note (yes or no, with a short quote; these are not fairness tests)
F1 hours with one child and hours with both in the same week; F2 guaranteed hours; F3 hosting credit or shared expenses; F4 taxes or an annual budget; F5 states its assumptions; F6 free with no account or email; F7 shows a date or version.

## Rules
Public pages only, no accounts, no email addresses entered, no contact. Do not judge tax or legal content: record only whether it is present. Be fair: a tool built for a different job (an annual budget, say) is "not supported" on a rate test, not "fail".

## Evidence rules (added in version 2, after an internal review of the evidence)
- One output file per tool and case: `work/<tool-slug>-<case>.json`. Never overwrite a file; a rerun gets a new name (`-r2`).
- Each output file holds: `utc` (ISO time of the capture), `url` (the final URL after redirects), `http_status`, `browser` (name and version), the inputs entered, the values read back from the fields, the result text as shown, and the file names of the screenshot and the saved page HTML.
- Save the page HTML after the result is shown (`page.content()`) as `work/<tool-slug>-<case>.html`, and any script the result depends on.
- Write `manifest.json`: SHA-256 of every screenshot, HTML file, script and output file.
- Rerun any case whose output file was overwritten or lost.

## Wording of results (added in version 2)
Each test is a plain question with a factual answer: yes, no, not shown, or not supported. A tool that answers "no" uses a different method; it is not called wrong. The word "error" is used only for arithmetic that does not add up or for a page that contradicts itself, and only after a hand check in an ordinary browser on the day.

## Changes
- v2.1 (2026-11-13): NannyFinder added to the tool list after the first eleven were scored. No change to cases, tests or rules.
- v2: Sharefully's calculator is tested through its page like the others (v1 said not to test it); evidence rules; wording of results.
