How Much Hail Damage Does It Take to Replace a Roof?

Insurance carriers typically want to see 8 to 10 hail impacts in a 10 foot by 10 foot test square before they scope a full roof replacement, and the threshold generally has to be met on at least two slopes. That figure applies to homeowners everywhere, and here is the part almost nobody publishes: it is carrier practice, not an engineering standard. Haag Engineering, which developed the test square method, does not publish a hits-per-square threshold at all. Arrowhead Roofing has run these squares on Tulsa roofs since 1987, and knowing where the number comes from is worth more than the number.

Note: this is a general explanation of how hail damage is assessed. It is not a statement about any particular policy. Arrowhead Roofing is not a licensed public adjuster and we do not negotiate insurance claims. What your policy covers is determined by your policy and your carrier.

The test square method has a defined procedure, and each step exists for a reason.

The Short Answer

There are two questions inside this one and they have different kinds of answer.

Is the roof damaged? That is a technical question with a technical answer: are there fractures in the mat, and how many, and where.

Will it be replaced? That is a coverage question, decided by the policy, and the count is one input to it rather than the whole of it.

Anyone who tells you a number settles both has skipped a step.

Where the 8 to 10 Figure Comes From

The test square method was developed in the early 1960s by engineers at Haag, who applied statistical sampling to hail assessment and settled on 100 square feet because it is representative, practical to inspect, and easy to talk about.

What Haag did not do is publish a threshold. Their method is extrapolative: count the damage in the sample, project it across the slope, and describe what you found. The 8 to 10 figure is an insurance industry convention layered on top of that method. The inspection curriculum published by InterNACHI puts it plainly: the number required will vary, different insurance companies have different policies, and eight is a common number.

Andy Anderson at Arrowhead says it the way it should be said: “Insurance companies typically want to see 8 to 10 hail hits in a 10 foot by 10 foot area.” Typically. Not always, and not by rule.

What the Test Square Cannot Settle

This is the section a competitor will not write, and it is the reason to trust the rest of the page.

The method has a real statistical weakness, and it was set out in Claims Journal in May 2026 by Giancarlo Grant, a consultant at Berkeley Research Group. Three points from it are worth knowing:

• Where the square goes changes the result. Placed where the damage looks worst, or simply where the roof is easiest to reach, it distorts the count in one direction or the other.
• Finding nothing does not mean there is nothing. Modelled statistically, a square with zero observed impacts is still consistent, at 95 percent confidence, with a real density of three or four impacts per hundred square feet.
• It gets less reliable as damage gets worse. Variability rises with density, so a single square is least dependable exactly when the roof is worst hit.

Two honest caveats about that source. It is a named, dated analysis by a practising consultant in a trade journal, not peer-reviewed research, and as far as we can find no peer-reviewed study of test square sampling error exists. We would rather tell you that than dress it up.

Why Two Slopes Matters

A single damaged slope is a harder case than two, and the reason is physical rather than procedural.

Hail driven by wind arrives at an angle, so a genuine storm usually marks the slopes facing into it and leaves the others comparatively clean. Damage found on one slope and nothing anywhere else invites the question of whether something other than hail caused it, which is why the method calls for a square on each direction rather than one square in the most convincing place.

What Changes the Answer

Three things move the outcome more than the raw count does.

The age and type of the shingle

An older, more brittle shingle fractures at a smaller stone size than a new laminate. The same storm produces a different count on two different roofs, and that is a property of the roofs rather than a discrepancy.

Whether the damage is functional

Dents in metal and scattered granule loss are real hail effects that do not necessarily reduce water-shedding capability or service life. The count that carries weight is fractures in the mat.

What the policy says

An age based roof schedule, a percentage deductible, and whether the settlement is at replacement cost or actual cash value all change the outcome without changing a single thing on the roof. Your declarations page is where those are written, and your carrier or agent is the only party who can interpret them.

Related Questions

What am I actually counting?

Bruises, meaning circular indentations with a fracture in the shingle mat that feel soft. What does hail damage look like on a roof? covers what qualifies and what gets mistaken for it.

What if the hail was small?

Three quarter inch is the most common size in Tulsa and the research on it is more interesting than either side of the usual argument. Can 3/4-inch hail damage a roof? covers the impact testing and the cumulative effect.

What if they say the damage is old?

Freshly exposed asphalt is black and oxidises to grey over months, so age is checkable rather than arguable. How can you tell old hail damage from new hail damage? sets out the indicators.

See Also

The two answers on either side of this one are what you are counting in the first place, and how the count becomes a documented claim.