What This Year's CAO Points Actually Teach Our 5th Years

Every September, once the CAO madness has actually died down a bit, I sit our 5th years in a room and show them the points data from the summer just gone. Not because it tells them anything about what they'll need next year — it genuinely doesn't, not directly — but because I have never found a better way to knock one specific idea out of their heads: that CAO points are some kind of exam mark colleges hand down from on high. They're not. They're just a queue. And this year's numbers make that argument for me better than I ever could on a whiteboard.

Here's the bit that should be confusing, and isn't once you get it

Grades were down this year. Properly down. The State Examinations Commission's post-marking adjustment — the bit of standardisation that happens after marking to keep results roughly consistent year to year — dropped from 5.8% of grades bumped up in 2025 to just 4% this year, and H1s fell from 12.1% of all grades to 10.1%. So on paper, this cohort did worse than last year's.

And yet — according to CourseCompass's own breakdown of the 2026 vs 2025 numbers, which I'd genuinely just tell any student to go look at themselves rather than take my word for it — of the 1,168 courses you can fairly compare across both years, 627 went up in points, 451 went down, and only 90 stayed put. Average change, plus 5.9. Median, plus 2.

If you're only looking at grades, that makes no sense at all. The second you look at demand instead, it makes complete sense. 95,710 people applied to the CAO this year. That's 6,402 more than 2025. And 20,527 of those applicants were repeat or deferred students sitting on grades from earlier, more generously marked years, going up against a cohort that scored lower on paper. More people fighting for the same number of seats pushes the cut-off up. It really is that simple, and it happens every single year whether or not anyone notices.

The moves I actually point out, because the averages hide the interesting stuff

Construction and engineering courses had a real jump in points this year — no surprise, given how strong the graduate job market is in that space right now. Computer science points kept falling, including a 72-point drop at UCC, which tells you the demand spike from a few years back is properly correcting. Veterinary applications went up 82% and the points still fell, which I honestly find fascinating and I'm not entirely sure I can explain to you why. Primary teaching fell at every single college that runs it. And TUS and Dundalk IT both saw increases across the board, which points to those colleges themselves getting more popular, not just one course within them.

On top of all that, there are 61 courses brand new to the CAO this year, spread across 29 institutions. The list of what even exists changes year to year, never mind the points needed to get onto it.

The actual lesson, a full year before anyone needs it

I say this to every 5th year and I'll keep saying it: stop treating last year's points like a number carved in stone. A course that needed 450 points in 2025 could genuinely land anywhere between 420 and 480 in 2026, purely depending on who applies and how they do, and there is no way — none — to know that a year out. So don't build your entire plan around a single figure you read on a forum.

What actually helps, instead, is understanding the shape of it. Falling points doesn't mean a course got softer — computer science proves that. Rising applications with falling points, like veterinary this year, is a course to watch, because the points there could well climb back up once demand catches up.

And honestly, if a student wants to actually check a course rather than trust whatever their older cousin told them at a family dinner, CourseCompass's points tool lets them search it properly and see the real numbers, not a rumour. If they're weighing up whether they'd even have a shot given a certain grade spread, the points calculator does that in about thirty seconds. And if a course looks completely out of reach on points alone, I'd always point them at the backdoor routes tool before they write it off — there are usually more ways in than students assume. That's five minutes of actual homework before a guidance appointment, and it changes the whole conversation.