The pipeline · Ottawa
Ottawa's housing pipeline is 41% smaller than the register says
Add up the homes column in the City's development applications and you get 63,999. Group the applications by the piece of land they are about and you get 45,520. The difference is the same homes counted two, three and four times.
The City of Ottawa publishes every live development application: the address, the type, the date, and how many homes it is for. On 2026-09-17 there were 433 of them. Add up the homes column and you get 63,999.
That number is wrong by about eighteen thousand homes, and the reason is not an error in the City's data. It is that the register is a register of files, and one project files several.
Three applications, one building, three counts
4795 Bank Street, filed on 4 August 2026:
| File | Type | Homes |
|---|---|---|
| D01-01-26-0014 | Official Plan Amendment | 620 |
| D07-16-26-0017 | Plan of Subdivision | 620 |
| D02-02-26-0061 | Zoning By-law Amendment | 620 |
One piece of land, one scheme, 620 homes — and 1,860 if you add up the column. All three were filed on the same day, because that is how it works: a project that needs the plan changed, the zoning changed and the lots drawn files all three at once, and each application has to state how many homes it is for.
322 Athlone Avenue is the same shape: an Official Plan Amendment and a Zoning By-law Amendment, both filed 15 December 2025, both for 1,206 homes. 1,206 real, 2,412 in the column.
Grouped by land instead of by paperwork
Collapse the 433 files onto the 333 sites they describe and the total is 45,520. The gap is 18,479 homes — forty-one per cent of the real figure, distributed over the 41 sites that carry more than one application with a home count on it.
The biggest single instance is 4497 O'Keefe Court, which has four:
| File | Type | Filed | Homes |
|---|---|---|---|
| D01-01-24-0023 | Official Plan Amendment | 2024-10-21 | 1,500 |
| D07-12-25-0085 | Site Plan Control | 2025-06-19 | — |
| D02-02-25-0061 | Zoning By-law Amendment | 2025-08-22 | 1,493 |
| D07-16-25-0015 | Plan of Subdivision | 2025-08-22 | 1,493 |
1,493 homes, counted as 4,486. And notice the other thing in that table: the number moved. The 2024 Official Plan Amendment says 1,500; the two 2025 applications say 1,493. A project is refined between filings, so "how many homes" has more than one answer on the public record, and picking one is a judgement rather than a lookup.
The rule we use, and where it breaks
A site is the normalised address. Within a site, the home count is chosen in three steps:
1. Prefer the applications that describe the whole site — an Official Plan or Zoning amendment, or a Plan of Subdivision — over a Site Plan Control for one building on it. 2. Among those, take the most recent. A scheme that shrank should not keep the largest number anybody ever wrote down. 3. Break a same-day tie on the larger figure, rather than on whichever row the file reader happened to return first.
It has two known failure modes and both are visible in the output rather than hidden. It will split one project filed under two different addresses, which merges nothing and undercounts nothing — it simply misses a merge. And it will merge two genuine phases filed under one address: 951 Gladstone Avenue shows amendments for 1,576 homes in August 2025 and a Site Plan for 402 in May 2026, and whether that 402 is the first phase of the 1,576 or a reduction of it is not answerable from the register. We keep 1,576. A grouping rule you cannot inspect is a grouping rule you should not trust, which is why the command prints every one of the 41.
It is Ottawa's register, not our arithmetic
The same tool on San Jose: 112 files, 111 sites, 19,962 against 19,760 — about one per cent.
That is the control. If the collapse were an artefact of how we group addresses, it would show up in both cities. It does not, because San Jose's register largely does not carry the amendment applications that Ottawa's does. Ottawa is the bad case for a specific, structural reason: it publishes the paperwork of changing the rules alongside the paperwork of building, and the same homes appear in both.
We printed the wrong number too
This is not a critique of somebody else's arithmetic. A figure about forty per cent high was on the front page of our own city map, in our fortnightly report, and in cold emails we sent, until we grouped the files by site on 17 September 2026.
The map read the grouped figure; the fortnightly report did not, and kept totalling the homes column across files for a day after we knew better. It was fixed the same afternoon this piece was written — every home figure on the report now comes from the site grouping, and the per-stage table is checked to sum to the total before the report will generate, because nobody adds up eight numbers in an email.
So the practical point is narrow and it applies to anyone who has ever quoted a pipeline figure from this register, us included: the homes column is not additive. Summing it does not measure a city's housing pipeline. It measures how much paperwork the pipeline generated, which in Ottawa is a number about forty per cent larger.
What this does not tell you
Whether 45,520 will be built. It is what has been applied for and is live in the register — not approvals, not permits, not starts, and not homes. A separate count found 106 sites holding an entitlement with no site plan behind them, the oldest idle 29.3 months, so the distance between this number and a building is the next thing worth measuring rather than something to assume.
Nor does it tell you the number is now stable. The register changes weekly, and the grouping is conservative by design. Re-run the command; if the figures have moved, this piece is dated and owes a correction.