You cannot benchmark against a number that does not exist. Measure your own.
Your claim denial rate is the share of the claims you sent that a payer adjudicated and declined to pay. The counts are already in your practice management software. The arithmetic is trivial; getting the definitions right is the part that decides whether the answer means anything.
The formula
Denial rate = claims denied in the period, divided by claims adjudicated in the same period.
A worked example on made up round numbers, so the arithmetic is visible: payers adjudicated 400 claims in a month and declined 34 of them, so the answer for that month is 34 divided by 400. These are illustrative inputs, not a benchmark and not anyone’s real figures.
Five things that decide whether the answer is real
Pick one definition and keep it
There are two honest ways to count, and they give different answers. You can measure denials as a share of the claims you SUBMITTED in a period, or as a share of the claims that were ADJUDICATED in that period. Neither is wrong. Mixing them is.
The submitted basis version is easier to pull and harder to trust: a claim submitted in March may not be decided until May, so a recent month always looks better than it is. The adjudicated basis version answers the question you actually care about, which is what the payer did with the work you sent.
Count a resubmission once
A denied claim that you correct and resend is one claim that got denied, not two claims. Systems that count every submission will double count your worst claims, which is exactly backwards: the more diligently your team appeals, the worse your rate looks.
Decide up front whether a corrected resubmission is a new claim or the same one, write it down, and apply it the same way every month.
Separate the three things above
A rejection, a request for information and a denial are different events with different fixes, and most reports pour them into one bucket. A rate that includes rejections is measuring your clearinghouse hygiene. A rate that includes pending requests is measuring your turnaround. Neither is measuring your payers.
Confirm your own status codes before you trust any filter
This is the step that quietly ruins the exercise. Claim status codes are configured per practice: the value that means denied in one office means something else in another, and a filter built on a guessed code silently reads zero rows, or silently reads the wrong ones.
Open a handful of claims you already know the outcome of, note the exact code each one carries, and build the filter from those. Confirm, do not assume.
Read it per payer, not as one number
A single practice wide denial rate is the least useful form of this number. The reason to measure it is to find the payer, the procedure code or the missing attachment that is costing you, and a blended figure hides every one of those.
Once you have a month of counts by payer, the outliers are usually obvious, and they are usually fixable.
What about comparing it to the industry?
There is nothing honest to compare it to. We went looking for a published US dental denial rate and could not find one, and the figure usually quoted turns out to be measured on medical insurance. Your own trend, month over month and payer by payer, is the comparison that exists.
Or have it counted for you
Practice Evolved reads your practice management software, read only, and keeps this count current without anyone exporting anything: what is open, what came back denied, what is ageing, and which payer it is sitting with.