AI match analysis
Generate a written analysis of a match from your tagging — findings you can play as clips, a comparison with your recent matches, and coaching points.
On this page
AI match analysis turns the clips you've tagged on a video into a report — findings backed by numbers you can play as clips, a comparison with your team's recent matches, and coaching points. It reads your tagging, not the footage itself: tags, counts, timing, zones, and everything else you've logged. No video is sent anywhere. Open it from a video in the workbench: the … (More) menu → AI group → Generate AI match analysis (sparkles icon).
It isn't the Match document, the PDF or PNG you attach yourself (the official match sheet, for example), nor the Match stats, the numbers cliply computes with no AI. See AI in cliply.
AI match analysis lives in Broady and rolls out per organization. If you don't see it in the … menu, it isn't on for your organization yet — contact support to ask about turning it on.
Generating a report
Select Generate analysis and a panel opens on the right and the report fills in as it's written. Once done, it's saved with the video — reopen the panel any time and it shows up instantly, with a Generated {date} line.
Getting back to a generated analysis
- In the workbench: the AI analysis chip, next to the tools, opens it in one click. In the … menu, the entry now reads View AI match analysis.
- In the video list: a sparkle icon beside the name or on the poster marks the matches that have one. In the video's menu, View AI match analysis opens the workbench with the analysis showing.
Regenerating
If you've added or removed clips since the last generation, a notice suggests regenerating so the report reflects your current tagging. Select Regenerate to re-run it with the latest clips — this replaces the old report, it doesn't create a second one.
You need at least one tagged clip on the video, plus a plan that includes AI or a purchased AI credit pack — each generation spends credits. See how credits work. Generating or regenerating always needs an internet connection but once a report is finished, it stays readable offline. Members with the read-only viewer role can read a generated report but can't generate or regenerate one.
What's in the report
cliply writes the report in your interface language. It has these parts, in this order:
| Section | What it covers |
|---|---|
| Overview | One headline sentence about the match |
| Findings | Three to six findings, each marked Strength, Problem or neutral, each with its evidence |
| Compared with your recent matches | How this match stacks up against your team's recent ones — only when there is something to compare with |
| Coaching points | Exactly three recommendations, each with its evidence and a Check next match line: the number to re-check |
| Limits of this report | Caveats worth knowing: how deep the tagging is, tags with no side, small samples |
Evidence you can play
The numbers in the report are not written freehand. Before the AI writes anything, cliply works out the facts from your tagging — counts and rates per tag, per period, zone hotspots, qualifier variants, players involved, timeline events by side, the longest runs of the same side, the most-flagged tags — and each fact keeps the clips behind it. The AI can only narrate those facts, and a claim that points to a fact that doesn't exist is dropped before you see it.
In the report, every finding and coaching point carries evidence chips. Each one shows the fact, for example "Turnover: 19 (12.7/60') vs 8.1/60' last 5, +57%", and, when the fact has clips, a play icon: select the chip (it reads Play N clips on hover) and those clips play in the workbench player.
Compared with your recent matches
The comparison uses your team's last 5 tagged matches, not your whole organization:
- Matches are those of the same team set on the video, or of the same project when the video has no team.
- Only matches you can open count, and only matches dated on or before this one.
- Rates are normalized per 60 minutes, so a lightly tagged match is not read as "fewer turnovers".
- The sample size (n) is always shown next to the comparison.
If this match was tagged much more or less heavily than those matches (beyond 30% either way), the report says so, because a count that moved may be tagging and not play.
Until another match of the team is tagged, nothing is compared: the report says so and the section is left out.
What it reads
The report draws on everything you've logged against the video: tags, qualifier variants, field zones, reactions and flags, timeline events, tagged players, clip comments and event notes, the match result when it's set, and clip timing for per-period rates and runs. Tags switched off for stats in the catalog (Count in stats) are left out of the counts, as on the stats screen.
Who can see it
The report is for your staff. It is saved with the video and readable by every member of your organization in the workbench. It never appears on share links: a player or a guest opening a link sees the clips, not the coaching report.
Reports generated before this version are no longer shown. Generate one again to get the new report.
Tips
- The richer your tagging — qualifier variants, zones, players, notes — the richer the report. Sparse tagging means fewer findings and more caveats under Limits of this report.
- Set the score and opponent on the video before generating so the Overview anchors to the actual result.
- Tag both your team and the rival, and give tags a side — tags with no side are flagged as a limit and weaken the findings.
- Tag your matches with the same team set so each one has a baseline to be compared with.
Related
Rather talk to a person?
If the guide isn't enough, message us on WhatsApp or by email. Someone on the team answers, no bots.


