The November 2025 SAT: a geometry exam, and a calculator that could not help

Harder than average, and concentrated. Difficulty reports clearly outnumbered reassurances, with the weight falling on math. Reading and Writing was called fair, and vocabulary was the one place it bit.

What makes November distinctive isn’t the level of difficulty. It is the composition. Geometry and measurement dominated the math talk to a degree no other recent sitting comes close to: triangles, angle relationships, tangent lines, perimeter and volume. On most forms geometry is a supporting domain. On this one it was the main event.

Alongside it ran the heaviest talk about the graphing calculator we have seen after any test date. Most of it was students working out afterwards what they should have typed. If that was you, the next section explains why nothing you could have typed would have helped.

Desmos is genuinely powerful on the Digital SAT. For one band of questions it is close to a cheat code: solving systems, finding intersections, locating a vertex, checking whether two expressions are equivalent. If you have practiced those moves, you gain real time.

November’s math sat mostly outside that band, for a structural reason. A graphing calculator works on equations you have already written. A tangent-line question hands you a point of contact and a slope. The work is recognizing that the radius meets the tangent at a right angle, which turns the geometry into a slope relationship. Once you have that, the arithmetic is one line and you don’t need a calculator. Before you have it, there is nothing to type.

The same goes for the angle questions November was full of. A triangle’s interior angles sum to 180. A second relationship between two unknown angles only helps if it is independent of that sum. No calculator hands you either of those. So learn Desmos, but learn it second. If you reach for the calculator first, you have no method at all for the questions that decide the top of the section.

Geometry and trigonometry is the smallest of the four math domains. That quietly shapes how you prepare. It is the one people leave until last, treat as formula recall, and assume will cost them a question or two at most.

November is the counterexample. On a form that weights geometry heavily, treating it as a footnote costs you a whole cluster instead of one isolated point. The loss lands in Module 2, where your ceiling gets set.

The content isn’t deep. What makes it expensive is that geometry is the one math domain you can’t reason out from scratch under time pressure. Say you don’t already know that a tangent meets the radius at a right angle. No amount of general math ability gets that back in ninety seconds. It is the most memorizable domain, which makes it the most wasteful one to be caught out by.

One pattern ran through November’s hard math. A question hands you a configuration, and you identify the single geometric fact that unlocks it. A tangent and a radius, so a right angle. Two angles in a triangle, so a fixed sum. A relationship between two solids, so a scale factor applied to volume rather than length.

The scale-factor items punish a specific guess. If one solid is twice another, its volume is eight times, not twice. Lengths scale by the factor, areas by its square, volumes by its cube. Doubling looks like the safe answer and it is wrong by a factor of four.

There was also a run of questions asking which of several expressions would be enough to pin down two unknowns. That is a condition question in geometry clothing. It turns on whether a given expression is independent of the one you already have, not on solving anything. If your two angle facts both reduce to the same statement about the sum, you have one equation and two unknowns, and no amount of algebra saves you.

Reading and Writing came back as fair, with two pressure points. Vocabulary was the first, and the words were exactly the sort the exam now favors. Not rare, but asking you to judge register and stage. Incipient and embryonic both mean early, so a word meaning early and undeveloped is right only when the sentence has already told you the idea was unfinished.

Punctuation around inserted material was the second. One item put a word introducing an example inside a dash construction. These look like questions about which mark to use. They are really about finding where the inserted material starts and stops. Once you know that, the mark follows on its own.

The test is quick. Cover the insertion and read the sentence without it. If it still works, the insertion needs matching marks at both ends: two commas, two dashes, or two parentheses. A choice that opens with a dash and closes with a comma is wrong however good it sounds.

Original questions written around what this exam tested, one for each argument above. Nothing here is a real test question.

November is the argument against leaving geometry until last. It is small by share and large by variance, so the cost of skipping it is unpredictable. On most forms it costs you a question or two. On a form like this one it costs you a cluster, in the module that sets your ceiling.

It is also a useful correction on the calculator. Desmos rewards you for thinking you have already done. November was full of questions where the thinking was the entire task, and there was nothing left for the tool to do.