How Microsoft Stock Reacted to Its Q2 2026 Earnings, Measured Minute by Minute

On paper Microsoft's quarter cleared the bar. On the tape, measured from 20:02 UTC, the stock dropped from $401.83 to $396.99 inside fifteen minutes — -1.2% — which is what it looks like when expectations sat above the published consensus.

This page is the measured anatomy of that print on 29 July 2026: every earnings-related headline that crossed for Microsoft, in order, each with the price reaction measured from its own instant.

The print, row by row

The shape matters as much as the size. One minute in, the stock had moved +1.27%; by minute fifteen the sign had flipped to -1.2%. The first algorithmic read and the considered read disagreed, and the considered read won — a reversal inside the first quarter-hour, which is the single strongest argument for measuring at more than one horizon.

Which line actually moved it

Each row above is measured from its own instant — the price immediately before that specific headline crossed — so the rows can be compared directly. The largest single reaction, +2.98%, belongs to the 22:04 UTC line. By the time the 20:55 UTC line arrived, the measured response was -0.12% — the marginal information in each successive headline decays fast, and the decay is measurable.

Did the first read hold?

No — and that is the most useful fact on this page. The first fifteen minutes measured -1.2%, yet by the close of the next session the stock stood at +15.51% against the prior close. The instant reaction and the settled session disagreed about what this report was worth, which is exactly why both are worth measuring. A reader acting on the first print alone had the wrong sign.

A beat that gets marked down is the market saying the whisper sat above the estimate, or that something inside the release — guidance, margin, a segment line — mattered more than the headline. The measured sequence above shows where that judgement landed first.

How these numbers were measured

We record the price of the instrument immediately before each headline crosses our tape, then again one minute and fifteen minutes later, together with traded volume against that instrument's own normal. Every figure on this page comes from those recordings — the timestamps are the actual instants the lines crossed. Nothing here is an estimate, a close-to-close change, or a model's guess. The method, the gates every page must pass, and how we correct errors are documented in our editorial standards.