Birchbungalowvire
Lens 03 · Reviews

Turning VOD Review Into a Weekly Habit

VOD review is where data becomes behaviour, or fails to. This lens lays out the weekly loop analysts install so footage turns into a short list of changes the players actually make.

Updated 12 January 2026 · 7 min read

Why Review Fails Without Structure

Most teams review. Few review well. The typical session is a full replay watched end to end, paused whenever someone spots an error, with the loudest voice deciding what mattered. Ninety minutes later the team has a dozen observations, no priority and no way to check next week whether anything changed.

A consultant does not usually replace this session; they restructure it. The core idea is that review should test a small number of hypotheses, each linked to a metric that can be checked again. "We lose retakes on B" becomes "our B retake success rate is below our A rate across the last twelve maps; let us watch the six worst retakes and agree one change." That sentence has a sample, a measure and an action.

The loop below is deliberately light. If it takes more than a few hours of analyst time per week, it will not survive a busy tournament schedule.

Flat vector analyst at a desk with two monitors showing a match replay and a round timeline chartWeekly loop

Two screens, one question

Footage on one monitor, the metric being tested on the other.

The Weekly Loop

  1. Tag

    After each official or scrim block, the analyst tags rounds or fights by type: opening, retake, post-plant, objective setup, teamfight. Tags take seconds and make every later step searchable.

  2. Measure

    Compute the handful of rates the team tracks this month. Compare with the previous fortnight and flag anything that moved beyond normal variation.

  3. Hypothesise

    Write at most three claims, each tied to a metric and a sample. Anything without evidence stays in a parking list for later.

  4. Clip

    Cut four to eight short clips per claim, including good examples. Players learn faster from contrast than from a reel of failures.

  5. Review session

    Forty-five minutes, coach-led, analyst presenting. Each claim ends with one agreed behaviour change, owned by named players.

  6. Re-check

    Next week, the same metric opens the session. If it did not move, the change was wrong, unclear or unpractised. Each answer leads somewhere different.

Tools That Keep It Light

Capture: OBS Studio for player POV recordings, which matter for titles where in-client replay tools are limited. Shooters: CS2's built-in demo playback with round navigation, plus parsed statistics from Leetify or Scope.gg. MOBAs: League of Legends replay files, which only play on the patch they were recorded on, and Dota 2's in-client replays alongside OpenDota or STRATZ parsing.

Analysis: a shared spreadsheet is enough for most teams. Python with pandas, or Tableau, earns its place once a team tracks more than a season of data.

Running the Room

The analyst presents evidence; the coach makes decisions. Blurring those roles is the fastest way to lose player trust in data. Keep criticism attached to situations rather than names, show at least one clip where the team did it right, and close by reading the agreed changes aloud.

Ask players to rate each clip's relevance on a simple scale. Low scores over several weeks tell the analyst their tagging is missing what players feel in-game.

Signals the Loop Is Working

SignalHow to read it
Action item completionShare of agreed changes visibly practised in the next scrim block. Below half means the list is too long or too vague.
Metric movementWhether the targeted rate improves over two to four weeks. Single-week swings are mostly noise.
Repeat issue countHow many claims reappear month after month. A stable list suggests the loop identifies problems but does not fix them.
Session lengthSessions creeping past an hour usually mean hypotheses were not narrowed before the room met.

Next lens: hiring a consultant