Accuracy log
Families make real decisions on the numbers this site shows, so we treat accuracy as a public process, not a promise. This page lists the automated checks that run against every figure, the date each last ran, and a permanent log of corrections: what changed, the verified source, and why. Spot something off? Question it and it goes straight to our review queue.
An automated checker re-fetches every cited official page daily across all 34 schools. A page that moves, breaks or errors is flagged and the figure it backs is downgraded rather than left to rot.
Retention, graduation and employment figures each carry their official source, a verbatim excerpt and the period they refer to. The checker snapshots each source's content; if the page materially changes, the fact is flagged and held out of our rankings until it's re-confirmed.
A green “Sourced” chip means the exact displayed value appears on the cited page, so a verifier fetches every sourced claim and looks for the verbatim figure. Latest run: 84 confirmed on-page, 13 not machine-readable (those lose the green chip), 2 blocked from automated fetching.
Deploy-blocking checks enforce the rules that keep labels honest: a fact may only cite its publisher (government or the school's own domain, never another school's copy), and an entering average may only read “Sourced” when the school documents one (2 of 22 universities today; 9 show their documented band and 11 are labelled estimates). A build that breaks a rule doesn't ship.
Every correction, relabel or removal we've made to published figures, newest first. We log the fix even when it makes us look worse; that's the point.
84.3% updated to 86.4%.
Why: USask replaced its Academic Year Snapshot with the 2025/26 edition (published June 15, 2026), which prints a retention rate of 86.4%. Our verifier flagged that the old 84.3% no longer appears in the cited PDF, so we updated to the figure the school now publishes.
Verified source: University of Saskatchewan 2025/26 Academic Year Snapshot ↗
Enrollment figures whose cited pages render their numbers with JavaScript (so our checker cannot see the value on the page) no longer show a green Sourced chip; they now disclose that the source page is official but the value could not be machine-confirmed on it.
Why: Our own value-on-page verifier recorded these 13 enrollment values as not machine-readable on the cited pages. A green Sourced chip means the exact value is on the cited page; these were not meeting that bar, so the chip was overclaiming.
Verified source: Our citation policy
88% corrected to 86%.
Why: The 88% we showed was our own sum (2% already graduated + 86% still enrolled) and appears nowhere in SFU's report. The report prints "Still Here 86%", so we show the figure the source actually publishes.
Verified source: SFU Institutional Research & Planning, Undergraduate Retention Report ↗
Retention figure removed from the profile and rankings inputs.
Why: Manitoba publishes attrition and continuing rates, never a first-year retention percentage. Deriving one (100 minus attrition) is our arithmetic, not their published figure, so we removed it rather than present a computed number as sourced.
Verified source: University of Manitoba (institutional reporting) ↗
Upgraded from our estimate to the documented 95.4% faculty-wide mean, labelled with its faculty-wide scope.
Why: U of T Engineering's own report states "The mean entering average for the 2023 cohort is 95.4%." It is a faculty-wide mean (U of T publishes no per-program figure), so we display it with that scope stated instead of guessing per-program numbers.
Verified source: U of T Engineering, By the Numbers ↗
Green "Sourced" chips changed to "Official dataset" chips.
Why: The figures are real and come from the province's official dataset, but the cited page is a lookup tool: the numbers are retrieved through it, not printed on the page itself. Our green chip promises value-on-page, so these now carry an honest dataset label instead.
Verified source: Government of Ontario ↗
Single "Sourced" averages replaced by the documented range each school actually publishes (for example Queen's 82-91%, Calgary 70-96%), with our midpoint clearly marked Estimate.
Why: These schools publish per-program bands, not one institutional average. Showing our midpoint with a green chip presented our arithmetic as their number. The documented band is now shown and the midpoint is labelled as ours.
Verified source: Each school's own published admission bands ↗
Every entering average re-verified against the school's own pages: Carleton 82% corrected to 78%, Queen's 89% to 86%, Western 88% to 91% (uwo.ca publishes 90.9%). Schools that publish no figure (including Waterloo and U of T) were relabelled as estimates.
Why: A reader flagged that Waterloo showed 90% cited to a page that states no such figure. The root cause was systemic: the chip keyed off a source URL existing, not the value being on it. All 22 universities were re-checked and each figure is now either documented on its cited page or labelled as our estimate.
Verified source: Each school's own admissions or facts page ↗
Citations repointed from a university-hosted copy of the provincial table to the Government of Ontario publisher.
Why: The values were correct, but all 10 schools' figures cited one university's hosted copy of the shared provincial dataset, which made unrelated schools look sourced to that university. We cite the publisher, and an automated guard now fails the build if any school's fact cites another school's domain.
Verified source: Government of Ontario ↗
Six dead or wrong admissions source URLs found and fixed on launch day; schools whose pages could not be confirmed were downgraded to "Verify with school" instead of keeping a broken verified link.
Why: An earlier build had copied source links without testing them; some 404'd. Every source URL is now HTTP-checked before each deploy and re-checked daily by the crawler, and a link that breaks is flagged rather than silently trusted.
Verified source: Our citation policy
If any figure on this site doesn't match what a school or government actually publishes, tell us. Challenges go to a review queue; confirmed corrections are fixed, and logged here permanently. You can also question any stat right where it appears on a school page.
Ideas and bug reports belong on the feedback & roadmap board.