The changes are described in a November 3 Outlook Blog post on Microsoft's Tech Community titled "How Outlook Search is changing: ranking, resilience, and the feedback loop." It covers several separate updates. Some are already rolling out, one is still in internal testing, and the update for each Outlook client arrives on a different schedule. That last point matters, as you'll see below.
The grading part: thumbs up, thumbs down
The headline feature is a thumbs-up or thumbs-down prompt on the search results page. Windows Central reports that Microsoft says Outlook search now surfaces more relevant emails and lets users rate results with a thumbs up or thumbs down.
Microsoft explains why it wants your rating instead of just counting clicks. Behavioral data can mislead. If you open an email from your results, that looks like a successful search, even though you may have opened it only to check whether it was the right one and then given up. In its blog post, Microsoft says an explicit rating exposes cases where clicks and dwell time don't show what the user actually wanted. Microsoft also says the rating is sent "with appropriate diagnostic context," which helps engineers find the types of queries that fail.
Be realistic about what the button does. Microsoft presents feedback as evidence that helps engineers decide what to investigate next. It does not say that a thumbs-down will fix your own future results. You're filing a report, not training a personal assistant.
Where the feedback prompt is available, according to Microsoft:
- Classic Outlook for Windows with Exchange Online and Outlook.com accounts on Current Channel with a subscription
- Outlook on the web
- New Outlook
- Outlook Mobile
- Outlook for Mac
- More search surfaces getting the prompt in November and December 2026, for both commercial and consumer users
In short: The thumbs prompt gives Microsoft a direct signal when a search fails. It doesn't let you correct your own results.
Seven ranked results instead of three
The most visible change is how results are laid out. Windows Central puts it simply: when you search for an email within Outlook, you will see seven ranked results rather than three. According to Microsoft's post, Outlook now shows up to seven relevance-ranked results at the top, and the rest of the results follow in date order. The ranking looks at signals such as how recent a message is, who sent it, and how you've interacted with it before.
Microsoft says it is still adjusting both the ranking logic and the point where relevance order switches to date order. Other sort options remain available.
Availability for the seven-result block:
| Client | Status |
|---|---|
| Classic Outlook for Windows (Current Channel and Monthly Enterprise Channel, subscription) | Available |
| Outlook on the web | Available |
| New Outlook | Available |
| Outlook Mobile | Available |
| Outlook for Mac | Rolling out in November 2026 |
Microsoft says this applies to both commercial and consumer accounts. It hasn't given a completion date for the Mac rollout.
If you'd rather keep strict date order
Not everyone wants Outlook choosing what's "relevant," and a bigger ranked block means more of that. The setting to turn it off already exists. In a Microsoft Q&A support thread, the suggested fix for Outlook on the web is to go to Settings > General > Search, clear the checkbox for "Show top seven most relevant results," select Save, then refresh the page and perform the search again. The same answer notes that disabling this option usually makes the results appear closer to chronological order. You can also open Sort in the upper‑right corner of the results pane and select Newest on top after running a search.
For Classic Outlook on Windows, an older Microsoft Q&A answer gives the path Outlook > File > options > search > uncheck - Show most relevant search results on top. That answer dates from 2023, so the exact wording may be different in current builds.
Before you switch it off, think about whether the new ranking might work better than the old three-result block did.
In short: Most clients now get more ranked results, and you can still turn ranking off if you prefer date order.
The stop-word fix is already live
This is the change that actually addresses the "the" and "is" problem. Microsoft says it has rolled out service changes that improve how Outlook handles common stop words such as "the," "and," and "is." According to the company, the change affected about 3% of searches, and those searches saw more result clicks where users opened and engaged with the mail. Microsoft didn't say how large that increase was.
Windows Central illustrates the change with Microsoft's before-and-after example of a search for "the FY27 budget." The example shows how Outlook's updated search ranks relevant emails higher in the results instead of surfacing messages that only happen to contain common search terms.
Microsoft lists the stop-word handling as available on all Outlook endpoints, for commercial and consumer users.
The AI model isn't shipping yet
Windows Central's coverage gets one thing wrong. It states that the improved rankings in Outlook are the result of a small language model ranker built specifically for Outlook. Microsoft's own post doesn't support that.
Microsoft says it is building an Outlook-specific small language model (SLM) ranker. It is trained through distillation: a larger "teacher" model supervises a smaller "student" model that is cheap enough to run at Outlook's scale. The company calls early offline results "promising" and says internal testing is underway. It gives no release date and doesn't say the model is generally available.
So the stop-word improvement and the seven-result layout are shipping now. The SLM ranker is still in testing. Microsoft actually describes the stop-word fix as one of several "smaller service changes" made "alongside" the model work, not as something the model produced. If your searches got better this week, credit the stop-word change and the new layout, not an AI model.
Microsoft's reasoning for distillation is sound. It says the goal isn't model quality on its own but a better balance between relevance and speed under real-world load. A large model that finds the perfect email after eight seconds would still make search feel broken.
In short: The SLM ranker is still in internal testing. The improvements you can see today come from the stop-word fix and the new layout.
Classic Outlook gets faster and handles bad connections better
Classic Outlook users get two more changes, and the details matter.
Cloud first, with automatic local fallback. Classic Outlook combines Microsoft's cloud search service with Windows Desktop Search, which keeps a local index on the PC. Microsoft says some query types used to stay local, including Find Related, parts of the advanced search syntax, and queries in a broader set of languages. Eligible searches now go to the cloud, so they get the same service-side improvements as other Outlook clients.
Local search is still needed offline and for content that isn't in the cloud, such as PST files and POP or IMAP accounts. Classic Outlook now switches to the local index automatically when the connection or a slow service response prevents a timely cloud result. Microsoft lists automatic fallback for Classic Outlook for Windows (Current Channel, subscription), Outlook Mobile and Outlook for Mac. For new Outlook, the feature is still "in development."
Latency numbers, with their limits. Microsoft credits two changes for improvements in the slowest 5% of Classic Outlook searches:
- Authentication-path optimization: a 40% latency reduction for the slowest 5% of searches, for commercial mailboxes only.
- Migration to cloud search: for query types moved from the local index to the cloud, latency for the slowest 5% fell from 4.5 seconds to 1.3 seconds.
These are tail-latency figures for specific groups of searches, not averages for every search. Microsoft argues that the slowest searches shape whether search feels dependable, and anyone who has watched the "searching..." spinner while a meeting starts will agree. Microsoft lists these gains for Classic Outlook on Current Channel and Monthly Enterprise Channel with a subscription.
For admins: Users on PSTs or POP/IMAP accounts still rely on Windows Desktop Search, so a healthy local index still matters. And if new Outlook users complain that search fails on weak hotel Wi-Fi, the automatic fallback hasn't reached that client yet.
How Microsoft tests ranking
Microsoft says it starts with offline tests against generated mailboxes built to resemble real enterprise inboxes: thousands of messages, meeting invites, newsletters, group mail and overlapping project topics. Using generated data means no real customer mailbox becomes a test fixture. The evaluation set has thousands of queries with known correct answers, based on real search patterns such as keywords, names, email addresses and combinations of those.
Microsoft admits that offline tests can't cover every mailbox or every user's intent. That's why it runs controlled live experiments before broad rollout, and why it wants the thumbs feedback. As the post notes, two people can type the same query while looking for different emails.
Also new: Calendar search spotlight
The same post announces Calendar search spotlight, which shows matching events directly in the calendar view so you can see what's before, after or overlapping them. You can switch back to the old list view with one click. It's rolling out to Outlook on the web and new Outlook in November 2026, and Microsoft says it is reviewing feedback before bringing it to other clients. It's a calendar feature and has nothing to do with grading email results.
Verdict
This is a worthwhile set of updates. The stop-word fix addresses a long-standing complaint, the Classic Outlook latency gains are measured where users notice them most, and asking for explicit feedback is a reasonable admission that click data doesn't tell the whole story.
Keep expectations in check, though. The AI ranker isn't shipping yet, automatic fallback hasn't reached new Outlook, and a thumbs-down tells Microsoft about a problem without fixing your own results. Rate your searches honestly anyway. Microsoft says it will use those ratings to decide which search failures to investigate next.