Test-square documentation done right

Ask a desk reviewer what separates a file they approve from a file they return, and test squares come up before almost anything else. Not whether you chalked them — whether the documentation proves what they showed. A test square that was counted carefully on the roof but photographed carelessly might as well not exist by the time the file reaches the desk.

This is the craft of making a test square hold up: where to put it, how to shoot it, how to count it, and how to record it so nobody has to take your word for anything.

What the square is actually for

A test square isn't a formality — it's a sampling method. One hundred square feet of roof, marked and counted, stands in for the whole slope. That's what makes the count defensible: "eleven impacts per hundred square feet on the north slope" extrapolates; "a bunch of hits up top" doesn't.

Which is why the square has to be honest. A square chalked over the worst patch of the roof isn't a sample, it's an argument — and an experienced reviewer can tell. A square placed to represent the slope is the one that survives scrutiny.

Chalking it right

• Full 10-by-10. A hundred square feet, squared off with chalk on all four sides. Partial squares and eyeballed boundaries invite the question of what was left out.
• One per slope, every slope. The square on the damaged slope proves scope. The squares on the clean slopes prove causation — directional damage consistent with the storm. Skipping the "boring" slopes throws away half the file's evidentiary value.
• Representative placement. Mid-slope, away from edges, valleys, and penetrations where damage concentrates for other reasons. If the slope's condition varies, place the square in the typical area — not the best patch, not the worst.
• Mark the hits inside it. Circle or dot each impact with chalk before you shoot the close-ups. A circled hit is countable in a photo months later; an unmarked one is a smudge someone can dispute.

Shooting it so it survives the desk

Every test square needs the same three-part sequence:

1. The locator. Wide enough to show which slope you're on — ridge line, neighboring elevation, or a landmark in frame. A square that could be on any roof proves nothing about this one.
2. The full square. All four chalk lines in frame, hits circled. This is the photo the count hangs on, so shoot it straight-on enough that nothing hides in the foreshortening.
3. The close-ups. Two or three marked impacts, shot close, with a gauge card or tape in frame for scale. This is where bruised mat and granule displacement become visible instead of asserted.

Four or five frames per slope — on a typical hip roof, call it twenty photos. Two extra minutes on the roof against a re-inspection trip across town.

Counting like it will be checked

Recount before you climb down — a count you got once is a count you guessed. Then record it twice: say the number out loud into your voice memo — "north slope test square, eleven impacts" — and type it into the claim's field data as a number. Spoken counts drift; typed counts get audited, which is exactly why you want them typed.

Two rules that separate clean counts from disputed ones:

• Count hail, not history. Blisters pop clean — exposed asphalt, no bruised mat underneath. Hail displaces granules and fractures the mat below. Mechanical damage has straight edges and pattern; foot traffic scuffs follow the walkable paths. If you counted it, you should be able to defend why it's a hail impact — the close-ups are where that defense lives.
• Zero is a count. An inspected slope with no impacts gets recorded as 0, photographed like every other square. A zero is causation evidence — it's what makes the directional-damage argument work. A blank is an uninspected slope, and uninspected slopes are what reviewers send files back for.

The number and the story have to match

The last failure mode isn't on the roof at all. The count in your field data says eleven; the narrative says "numerous." The memo says the south slope was clean; the report never mentions the south slope. Each mismatch is a thread a reviewer has to pull, and pulled threads become returned files.

Whatever you counted is what the report should say — per slope, as a number, consistent everywhere the number appears. If your process types the count once and carries it through to the narrative automatically, this failure mode disappears entirely; if you write reports by hand at night, this is the consistency check worth doing before you submit.

---

Test squares are one station in a larger on-roof discipline — the full sequence is in the 10-minute capture routine, and the desk's-eye view of why this matters is in what desk reviewers actually kick back. To see per-slope counts carried into a finished narrative, read a full sample report — LossNarrative writes the report from your photos, memo, and typed counts, and the typed counts always win.