Repair Quote
Observatory

What the observations say, by job family

One row per repair family. n is the number of currency-qualified CAD quotes in that family, not the total mixed-currency ledger. The last column is the published Greater Toronto Area market range from our own range card, market context, not a quote, so you can see where the recorded numbers sit against it.

5 families (Suspension — struts, Brakes — pads, CV axles & driveshafts, Cooling — thermostat, Battery) have fewer than 5 observations. We are not going to tell you what is typical off those, the count is published so you can watch it grow.

The observation ledger

Every observation drawn from already-public text, shown in full so anyone can check our reading of it. 1 observation is counted in the totals above but not listed here: it came from a private conversation with us, and we do not republish what a customer told us in confidence, only the number, the job family and the region ever enter the dataset.

Methodology

What counts as an observation

One observation is one repair price, stated by the person who was quoted it, attributable to exactly one job family. Amounts are captured where they already occur: an estimate pasted into our free quote checker, a number stated in the site's chat, a finding from a pre-purchase inspection, and public posts already collected by our GTA forum monitor.

What gets thrown away

• Ambiguous job scope. If a text names two repair families, the dollar figure can't be attributed to either, discarded, not guessed. Bundled multi-line estimates are therefore under-represented.
• Ambiguous amounts. If two candidate numbers appear and neither is introduced by a quote cue ("quoted 1400", "wants $2,300"), the row is dropped.
• Numbers that aren't prices. "113k miles", "201k km", "30k service", a model year, a non-CAD currency, all explicitly rejected in code.
• Implausible amounts, anything far outside the published range for that family is rejected as a parsing error.
• Work outside the job families, timing chains and belts, head gaskets, internal engine and transmission work, catalytic converters, body and collision work. Their prices live in a different universe and would distort every family they touched.
• Duplicates, collapsed on a stable key so the same post can never be counted twice.

What we store about a person

Nothing. No name, phone number, email, address, postal code, link or handle is ever written to the dataset, they are stripped before storage and the row is dropped entirely if anything identifying survives the strip. Location is recorded at city level at most. Text from private conversations is never republished. Cars With Fares customers' own jobs are excluded from the published rows completely: a customer's invoice is not research data.

How the numbers are computed

Midpoint is the median of the recorded amounts for that family. "Middle half" is the 25th to 75th percentile. Both appear only at n ≥ 5. These counts describe the dated published snapshot, not the current live database. Statistics use only currency-qualified CAD observations; region-confirmed counts are separate.

Limitations, stated plainly

• A bare dollar sign is not currency evidence. Unknown currencies are retained separately and excluded from CAD statistics. Legacy CAD labels are requalified from preserved text and city evidence; no conversion is performed.
• This is a convenience sample, not a random survey. Every observation is a price someone chose to tell us about, and people are far more likely to share a quote that shocked them — so the sample skews high by construction.
• Observations with zone "unknown" come from public North-American mechanic forums and are NOT confirmed Greater Toronto Area prices. They are counted and reported separately and are never merged into a GTA figure.
• Amounts are as stated by the person who received the quote. They are not audited, not itemised, and may or may not include tax, shop supplies, or a particular parts tier.
• A job family is assigned only when exactly one family is named. Bundled or ambiguous quotes are discarded rather than guessed at, so multi-line estimates are under-represented.
• Nothing here is a Cars With Fares price. These are other shops’ and dealers’ numbers, recorded as observations. Our own completed jobs are counted separately and are deliberately excluded from every job-family statistic and from the row-level feed — a customer’s invoice is not research data.
• No claim of statistical significance is made for any job family below the minimum reportable count.

Where the 19 observations came from

Take the data

Both feeds carry the same rows and are regenerated with the page.

• gta-price-observations.json, full export: totals, per-family statistics, stated limitations, and every publishable row.
• gta-price-observations.csv, the same rows, flat, for a spreadsheet.

Why this exists

Repair pricing in this city is opaque on purpose. You get one number, on one piece of paper, with no way to tell whether it's normal, and the only people who know the answer are the ones who wrote it. Every honest quote and every padded one passes through the same fog.

We can't fix that with an opinion. We can fix it slowly, by writing down what actually gets quoted, in the open, with the count attached, and by refusing to say more than the count supports. In a year this is a real reference. Today it is 19 numbers and an honest accounting of what they do and don't prove. Both versions are worth publishing.

Report generated 18 August 2026 from 19 observations. This is a dated publication snapshot. Counts may differ from later collection; publication is a separate refresh.