Every AFL club’s media team is happy to tell you about their advanced data analytics setup. They’ve got the dashboards. They’ve got the wearables. They’ve got the consultants. The annual report mentions “performance analytics” three times. The coach gives an interview about how the modern game is data-driven.

What actually happens with the data once the ball bounces is a different question. And after a couple of conversations with people who work inside AFL clubs, the honest answer is that the gap between “we have a data team” and “data is meaningfully shaping in-game decisions” is enormous, and varies hugely between clubs.

What every club actually has

The baseline in 2026 is more sophisticated than people outside footy realise. Every AFL club has access to Champion Data’s full feed, which provides essentially every measurable event on the field — possessions, disposals, pressure acts, contested ball wins, run distance, the lot. The data is granular enough to know which boot a player kicked with on a particular disposal. It’s there. It’s standard.

On top of that, every club has GPS tracking from training and matches, providing distance, speed, acceleration, deceleration, and recovery metrics on every player. Most clubs also have some form of biometric tracking — heart rate, sleep data, subjective wellness scores entered by players each morning.

So in terms of raw data, the modern AFL club is drowning in information. The question is what they do with it.

The training and recovery layer is genuinely sophisticated

This is where the data work is unambiguously paying off. Load management has been transformed by what clubs now know about player physical state. A player who’s accumulating fatigue markers across multiple metrics gets pulled out of contact training. A player whose recovery scores tank after a match gets adjusted minutes the following week.

This has had visible effects on the league. Soft tissue injury rates have dropped meaningfully across the AFL over the last five years. Career lengths at the top end have extended. Players coming back from injury are being managed more conservatively and the data shows it’s working.

The bit where this hits a ceiling is that load management is essentially a defensive use of data. You’re trying to avoid bad outcomes. The harder question is whether data is helping clubs achieve good outcomes — specifically, whether it’s giving coaches an edge during matches.

Pre-match planning is data-heavy

Coaches now go into matches with detailed opposition profiles built from data analysis. They know which opposition players run hardest in the last quarter. They know which opposition midfielders have lower pressure act numbers when they’re tagged. They know which kicking patterns the opposition forwards use coming back to the ball.

This stuff genuinely shapes tactical preparation. A defensive coach studying an opposition forward’s data history can identify habits that help his backline. A midfield coach can identify which opposition mid is most likely to break a stoppage with a clearance versus which one is more likely to lay a tackle. These are useful inputs.

The constraint is that AFL matches are too dynamic for pre-match data plans to survive contact with reality. You go in with a plan, the opposition responds to your plan, and by the second quarter the data-driven preparation is moot. So coaches use the data to set the starting conditions and then revert to football instincts for everything that happens after the first bounce.

In-game use is patchy at best

This is the bit where it gets interesting. The promise of analytics in any sport has always been real-time, in-game decision support — telling the coach what’s happening that he can’t see from the box, suggesting interventions before the patterns become obvious.

The AFL reality, based on conversations with people who’ve sat in coaches’ boxes, is that real-time analytics rarely changes a decision the coach was about to make. The coach knows when his mids are getting smashed at the stoppage. He can see when the opposition forward is finding too much space. He doesn’t need a dashboard to tell him to adjust.

What the dashboards can do is flag things the coach isn’t watching. Specific match-ups that have shifted, momentum patterns that aren’t obvious from the box view, fatigue patterns in opposition players. Whether the coach actually pays attention to those flags in the chaos of a match is genuinely variable. Some do. Many don’t.

The most useful in-game data application I’ve heard about is interchange management. With clubs running tighter rotation patterns and the bench reduced over the years, optimising when to bring a player off based on physical metrics has become a real decision support area. That’s a problem with a clear data signal and a clear decision output. It works.

The post-match analysis is where the gold is

If you ask the analysts inside clubs where they actually move the needle, most will tell you post-match. The week between matches is where the data team genuinely shapes coaching decisions — what worked, what didn’t, what’s the prep priority for next week, who needs personal feedback on specific patterns.

This is the part of the workflow that’s evolved most dramatically. A senior analyst at an AFL club now does work that would have taken a small team of video analysts a decade ago. They can run queries across multiple seasons of data, find specific patterns, build visualisations, and present findings to coaching staff in formats that are actually digestible.

The clubs that win the analytics game aren’t the ones with the fanciest dashboards. They’re the ones with analysts who can communicate findings to coaches in ways coaches actually use. That’s a human skill, not a technology problem. The technology has been adequate for at least five years.

The next frontier and who’s actually working on it

Where the actual innovation is happening at the moment is in machine learning applied to tactical pattern recognition. A few clubs — I won’t name them, the ones who’ve talked to me have asked for discretion — are working on systems that can identify tactical patterns the human eye misses. Specific spatial relationships, timing patterns in defensive structures, micro-patterns in contested ball situations.

This is genuinely interesting work and it’s at a very early stage. The clubs that get it right first will have a real edge for two to three seasons before everyone else catches up. The clubs that get it wrong will spend a lot of money on consultants and end up no further ahead.

One of the AFL data leads I spoke with mentioned that they’d been working with Melbourne AI consultants to help with the ML side of pattern recognition specifically because the off-the-shelf sports analytics products don’t handle AFL’s spatial complexity well. It’s a niche enough problem that you need people willing to build something from scratch for your specific data, and the bigger global sports analytics firms aren’t focused on Australian football.

The Sydney Swans have been one of the more publicly aggressive clubs on the analytics front and their results have been consistent enough to suggest the approach is working. Geelong’s data work has been quieter but their on-field consistency speaks for itself. Brisbane has invested heavily, though the on-field translation has been less obvious.

What I’d watch for the rest of 2026

A few things I’ll be tracking through the back half of the season:

Whether any club genuinely makes an in-game adjustment that’s visibly data-driven rather than instinct-driven. We haven’t really seen this yet and it would be a meaningful step.

How interchange optimisation evolves as the season fatigue accumulates. The clubs who get this right in August and September will have a finals edge.

Whether the new generation of younger coaches coming through — most of whom grew up in a more data-literate footy environment — start making genuinely different decisions to the older generation.

For now, the data revolution in AFL is real but it’s quieter than the marketing suggests. The training room has been transformed. The coaches’ box less so. We’ll see whether that gap closes over the next few years.