Seattle Office Vacancy in Q2 2026: Why Do the Numbers Disagree by 19 Points?

Depends who you ask, and the answers do not just differ in size. They differ in sign.

JLL, in the report GeekWire covered on July 17, put Puget Sound net absorption at roughly +372,000 sf for Q2 2026, alongside 23.9% vacancy and 25% availability, down from a 26.5% availability peak a year earlier. Kidder Mathews, measuring the same region over the same three months, reported -46,836 sf. Cushman & Wakefield reported -386,368 sf of absorption year to date through Q2, which they framed as a large improvement over the -2.41M sf posted through the same point in 2025.

Three firms measured one market and could not agree on the sign.

Nobody is lying. Kidder Mathews restricts its office series to multi-tenant buildings larger than 10,000 sf and excludes owner-user buildings. That single rule removes a large share of the Eastside campus product where the quarter's biggest move-ins landed. JLL's basket includes more of it. Cushman's number is year to date rather than quarterly, so it still carries Q1's damage inside it. Year to date is not a quarter, and the difference hides turning points.

The most useful number in the whole set is Kidder's own submarket split. East King County absorbed +248,731 sf in Q2 2026 while Seattle Close-In gave back -247,471 sf. Those two nearly cancel, which is exactly how you land on a regional figure of -46,836 sf that describes no actual submarket anyone leases in. The region is not one market. It is two markets moving in opposite directions at almost identical speed, and every regional average is a coin flip between them.

Wider than the absorption fight, and wide enough to embarrass anyone quoting a single figure.

For Q2 2026, here is what the major providers published for something they each call the Seattle or Puget Sound office market:

• CoStar, Seattle metro: 16.7%
• Kidder Mathews, Puget Sound: 23.4%, up 30 basis points from 23.1% in Q1
• JLL, Puget Sound: 23.9%
• Cushman & Wakefield, Puget Sound: 28.6%, up from 27.6%
• Savills, availability rather than vacancy: 32.1%
• Kidder Mathews, Seattle CBD: 34.9%
• CoStar, Seattle CBD: 35.6%

Top to bottom that is 18.9 points. The two ends of the range, 16.7% and 35.6%, come from the same vendor in the same quarter. Only the map changed.

Meanwhile Kidder Mathews had the Eastside at 21.0% in Q2 2026, ticking down while the CBD ticked up. A broker who quotes "Seattle vacancy" to a Bellevue tenant is off by almost 14 points and has told that tenant a story about leverage they do not have.

Four decisions produce nearly the whole spread.

Geography is the single largest driver. "Puget Sound," "Seattle metro," "Seattle city," "Close-In," and "CBD" are five different maps, and providers do not agree on the borders of any of them. Cushman's Seattle CBD boundary is not Kidder's. The gap between a healthy region and its weakest core is most of what you are looking at.

Inventory rules come second. Kidder's 10,000 sf floor and owner-user exclusion is one ruleset. CoStar counts far more small and older buildings, which is why its metro figure lands at 16.7% while everyone else clusters in the twenties. Providers also disagree on when a tower slated for conversion or sitting mid-renovation leaves the denominator. Removing dead inventory lowers vacancy without a single lease being signed.

Third, vacancy versus availability. Vacancy counts space that is empty. Availability adds occupied space being marketed, including sublease. JLL's own Q2 numbers show the gap inside one report: 23.9% vacant, 25% available. Savills' 32.1% is an availability figure and belongs in a different column entirely. Availability is always higher than vacancy, and the gap is your negotiating room.

Fourth, the period. Quarterly, year to date, and trailing twelve months all get called "absorption" in casual conversation. Cushman's -386,368 sf and Kidder's -46,836 sf are not in conflict once you notice one covers six months and the other covers three.

One more thing worth saying plainly. Every market report is also a marketing document, produced by firms that mostly represent landlords. That does not make the data wrong. It shapes which cut leads the summary.

Which provider. Which geography. Which quarter.

Run those three against anything that shows up in a client email and most of it collapses. "Seattle office vacancy is 23%" survives none of them. "Kidder Mathews has Puget Sound at 23.4% as of Q2 2026, and the Seattle CBD at 34.9%" survives all three, plus the follow-up.

Practically, match the series to the search. A client touring the downtown core should be working off CBD numbers, so Kidder's 34.9% or CoStar's 35.6%. A client weighing Bellevue is negotiating against Kidder's 21.0% Eastside, in a submarket where Databricks took 142,000 sf at Four106 and OpenAI took 247,487 sf at City Center Plaza this year. A client asking about the trend gets the honest version: leasing volume is genuinely recovering, with Savills counting 2.4M sf in Q2 and 4.6M sf in the first half, the strongest quarter since early 2019, and almost all of the strength sits east of the lake.

A blended vacancy number sounds smart and survives zero follow-up questions.

That methodology spread is normal, universal, and boring, and that treating it as a revelation is its own kind of grandstanding.

That is a fair hit. Every metro has this. Different baskets producing different rates is how the industry has always worked, and no provider hides their rules. A broker who waves the spread around to imply the data is broken is overreaching, and clients can smell it.

The direction consensus argument is stronger still. Look past the levels and all seven Q2 2026 series tell one story: leasing volume up, sublease pressure easing, Eastside tightening, CBD still bleeding. Kidder had sublease space at 4.72M sf in Q2 2026, or 11.7% of total available, the lowest share since 2018. JLL noted no new speculative construction underway. For a client who just needs to know whether to move now or wait, the shape matters more than any point estimate, and none of the providers disagree about the shape.

And the credibility play cuts both ways. Reaching for 35.6% to make a pitch land is the same move as the landlord broker reaching for 16.7%. Explaining the spread is the value. Weaponizing it is just cherry-picking with extra steps.

Build a provider map and keep it current. One page, one row per report you actually cite: the geography it draws, its inventory floor, whether the headline is vacancy or availability, whether absorption is quarterly or year to date, and the publish date. Refresh it the week the Q3 reports drop in October. That is an hour a quarter, and it is the cheapest credibility you will ever buy.

Then change how you quote. Provider and geography go in the same sentence as the number, every time, in writing. "Kidder Mathews, Puget Sound, Q2 2026: 23.4%." It reads as pedantic for about two weeks and then it reads as authority, because you are the only person in the deal whose numbers hold up when the landlord's broker pushes back.

Use the disagreement itself as the pitch. When a tower's broker hands your client JLL's +372,000 sf, you do not argue the number. You accept it and add the geography: the region absorbed space because East King absorbed +248,731 sf while Seattle Close-In gave back -247,471 sf. The region can recover while your client's tower keeps emptying. That reframe costs one sentence and it moves the concession conversation more than any rent comp.

The same discipline works against your own client when they are wrong. A tenant holding out for downtown-style concessions in Bellevue is negotiating against a 21.0% market with large blocks disappearing to AI tenants. Telling them that early is what a tenant rep is for.

Here is the part that compounds. Clients are not buying access to data. They are buying somebody who has already done the work of reconciling it. Everyone has the same reports in their inbox. Almost nobody spends the hour to know which basket produced which number, which means the broker who does spend it becomes the person the CFO calls before the CFO calls the landlord. If you can explain why two reports disagree, you stop being a reader of market reports and start being the one who interprets them.

Q3 numbers land in October. The spread will still be there. Be ready to explain it in one sentence.