A client asking why “the average price is up” but their own home's assessed value feels flat is not confused — they are looking at two genuinely different measurements.
Headline figure
+0.2% vs. −3.3%
In July 2026, Canada's national average home price rose 0.2% year over year to $674,819, while the National Composite MLS® HPI — CREA's benchmark index — fell 3.3% over the same twelve months, both reported in the same release. Same market, same month, opposite direction.
What the data says
A national or city average price is a simple arithmetic mean: total sale dollars divided by number of sales, for whatever happened to sell that month. If a run of high-end detached homes closes in one month and a run of smaller condos closes the next, the average price can swing sharply between those two months even if no individual home's value moved at all. That is exactly what happened between June and July 2026: the national average dropped from $696,078 to $674,819 — a $21,259 fall — without the underlying benchmark index falling anywhere near that fast month over month. See national home sales and price, tracked for that same comparison laid out month by month.
The benchmark index is built to strip that composition effect out. It does not track what actually sold; it tracks the estimated value of a defined “typical” property in a given area, re-estimated from the accumulated pattern of what buyers have paid for specific features over many years of data. That is why it moved only 0.1% month-over-month in July — its first increase since November 2024 — while the average price fell by more than 3% over that same span.
Because CREA's own national snapshot graphic uses three different price-change methods depending on the province, a page or a client conversation that lines up percentages from that map as though they were the same measurement is making an error CREA itself does not make consistently on a single page. A +5.9% figure for Nova Scotia (benchmark-based) and a +3.8% figure for Ontario (average-based) in the same graphic are not directly comparable numbers, even though they sit side by side.
The safe practice: always name which measure a figure comes from — “benchmark price” or “average price” — rather than saying “price” alone, and check the source's own methodology note before comparing two provinces or two periods against each other.
When a client points to a national or local average-price headline, the accurate response is that it describes what actually sold, not what any specific property is worth — and that CREA's own benchmark index, built to answer the “what is a typical home worth here” question, is telling a different story this particular month. Both numbers are real; they are answering different questions.
A 30-minute call is enough to tell you whether AI pays for itself here.
