Do licensed daycare hours fit a work day?

A licensed space that closes at 16:30 is not a licensed space for a parent who works until 17:00. Opening and closing times decide whether a family can use a program at all, and they are one of the few things about child care that can be measured rather than argued about. The Working Parent Score puts a number on it.

It is not a quality score, not a safety score, and not a measure of whether a single space is free. A city full of programs with perfect hours and no vacancies would score highly and help nobody. We publish it because nobody else does, and because the gaps it exposes carry as much information as the scores: two of the six pilot cities cannot be scored at all.

The six cities

Montréal tops the pilot at 93 — 89.5% of its programs with published hours open at or before 07:00 and 92.7% are still open at 18:00, which is what a province-wide network built around full-time working parents looks like in the data. Toronto scores 60 on the strength of its evening coverage while only 24.5% open by 07:00 — a city whose centres stay late but start late. Winnipeg is the mirror image: more than half open early, just over a quarter stay past 18:00.

Two cities we will not score

In Vancouver only 115 of 504 licensed programs publish usable hours (22.8%); in Halifax it is 17 of 75 (22.7%). A score computed from 17 programs would put a figure on the page without measuring anything, and those 17 are unlikely to be a fair sample of the 75 that exist.

Their components are printed in the table above anyway, because a reader is entitled to see what the thin data says and decide for themselves. What they are not entitled to is a headline score with the sample size buried. Halifax’s 5.9% evening share rests on 17 programs; treat it as a prompt to go and check, not as a fact about the city.

Source: MapleCub Canadian Childcare Dataset — Aug 2026 · computed 2026-08-10 · N = 3,625 · CSV

What each component measures

Source: MapleCub Canadian Childcare Dataset — Aug 2026 · computed 2026-08-10 · N = 1,600 · CSV

Methodology

The formula

Among the programs with usable hours in a city:

earlyShare   = % open at or before 07:00
lateShare    = % open at or past 18:00
fullDayShare = % spanning 08:00-17:30  (open <= 08:00 AND close >= 17:30)

WPS = round(100 x (0.35 x earlyShare + 0.35 x lateShare + 0.30 x fullDayShare))

Shares are fractions of that city’s usable-hours set, not of its total listings. The weights are a judgement call, published so anyone can disagree with them and recompute from the CSV: drop-off and pickup matter equally, and the full-day span carries slightly less because it overlaps both.

The thresholds are the bigger judgement, so here is what they are worth

07:00 and 18:00 are choices, not facts, and they are cliffs: a programme opening at 07:01 scores exactly what one opening at 09:00 scores. That would be a small problem if opening times were spread out. They are not — they pile up on the half hour, so where the cliff is drawn decides which side a whole city lands on. Toronto’s modal opening time is 07:30, and 221 of its 494 usable rows — 44.7% — open at exactly that time, half an hour on the wrong side of the 07:00 bar.

So the score is published with its sensitivity rather than on its own. Below, all three thresholds are shifted together by 30 minutes in each direction, on the same rows, in the same statement.

Read that table before quoting a single score. The ORDER of the four scored cities holds at every threshold we tried, which is the one durable thing here. The GAPS do not: at the published thresholds Toronto sits 11 points behind Calgary; shift every threshold half an hour later and the gap is 1 point. Anyone reporting “Toronto 60” as a fixed property of Toronto is reporting where we drew a line.

The last two columns of the CSV carry the shifted scores, so the sensitivity can be checked rather than taken on trust. It was computed on 2026-08-11, twice, by the same two SQL constructions used for the published components, and the published column reproduced the scores above exactly. Vancouver’s usable rows read 116 in that pass against 115 in the published components an hour earlier — the hours field is live and the harvest keeps adding to it. Its coverage is 23.0% either way, still below the 25% floor, so no city changed status.

Which programs count

Visible licensed programs in the MapleCub dataset, matched to a city by province plus normalized city name (lower-cased and trimmed), the same community definition the dataset methodology uses. Montréal-Nord, Montréal-Ouest and Montréal-Est are separate municipalities and are not folded into Montréal.

A program has usable hours when its opening and closing values both parse as a real 24-hour time (HH:MM, hours 00–23, minutes 00–59) and the closing time is later than the opening time. Testing for a non-empty value is not enough. The opening-hours field is partly filled by our website harvest, and it contains values like Weekdays, Unknown, and raw HTML and CSS fragments scraped off provider sites. Every one of those rows has an empty closing time, so the strict parse throws away nothing a looser test would have kept. We ran that check across all 20,516 visible listings in the dataset, well beyond the 3,625 in the six pilot cities.

The coverage rule

Coverage is the share of a city’s visible licensed programs that have usable hours. Below 25%, the city is published with its coverage figure and no score. The floor is deliberately blunt: there is no coverage level at which an hours-based score becomes fully representative, because programs that publish hours are probably not a random sample of programs, and 25% is simply where we judged the number stops being defensible in public. Even the scored cities rest on between 39.5% and 57.7% of their listings, so read every score as a statement about programs that publish hours, not about every licensed program in the city.

Where the hours come from

Two sources, both carried in the same two fields: the provincial licensing registries, which publish operating hours for some programs in some provinces, and MapleCub’s website harvest, which reads hours off providers’ own sites. The dataset does not record which of the two supplied any individual row, so we cannot break coverage down by source, and this page does not pretend to. Both are as-of 2026-08-11. Hours change without notice and registries update on their own schedules; a program’s real hours today are whatever the program says they are.

How the numbers were checked

Every component was computed twice, by two deliberately different SQL constructions — one casting the text fields to time and comparing times, one splitting the strings into integer minutes past midnight and comparing integers. The two agreed to four decimal places in all six cities. The scores were then re-derived from the components in Python, independently of both.

We also compared the parsed times against two boolean columns in the database that flag early opening and late closing. They agreed on every one of the 1,600 usable rows — and that agreement turned out to be worth nothing, because both columns are generated by the database from the very same two text fields. They are the parsed times restated, not a second opinion, so they cannot corroborate anything and the published figures rest on the parsed times. Recording it here because a cross-check that can only ever agree with itself is the kind of thing that quietly passes for verification.

Sources

There is no external source list on this page, and that is not an omission. Every figure above is computed from MapleCub’s own registry-derived dataset, so it is cited by the dataset line under each block and by the CSV, and it is ours to defend. The Lab’s full citation registry — every government page behind every external figure we publish — lives on the methodology page.

What this score does not measure

• Quality or safety. Nothing in this calculation touches inspections, staffing, ratios, curriculum or complaints. MapleCub does not publish comparative safety scores at all, and explains why.
• Availability. Long hours on a full waitlist help no one. We have no reliable vacancy data and make no vacancy claim.
• Cost. A program open twelve hours a day at market rates and one open eight hours at $10/day score on hours alone. Fees live in the cost guides and in the $10-a-day study.
• Days, shifts and holidays. Only weekday opening and closing times are in scope. Weekend care, overnight care, shift-friendly schedules and closure weeks are not measured, though some registries hint at them.
• Anything outside these six cities. The score is a pilot. It is not a ranking of Canadian cities, and the 4 scored cities are not the 4 best.

Version and cadence

This is v1, computed 2026-08-11 — a pilot across six cities, with national rollout to follow once hours coverage improves. Coverage is the binding constraint, not the arithmetic: the formula would run on any city tomorrow, and would be meaningless in most of them. Later versions will carry a new version label rather than silently replacing these numbers, so a figure cited today still checks out after the method moves.

Download

One row per pilot city, with every component, the coverage figure and the score — empty for the two cities we refuse to score. Free to reuse with attribution under CC BY 4.0.

Source: MapleCub Canadian Childcare Dataset — Aug 2026 · computed 2026-08-10 · N = 1,600 · CSV · Working Parent Score CSV

Cite as: MapleCub, “Working Parent Score v1,” 2026-08-11, https://www.maplecub.ca/research/working-parent-score. Questions or a custom cut: hello@maplecub.ca.