EMR is the only number on a vendor's insurance that predicts future losses

Every number on a certificate of insurance describes what happens after a loss. Limits, deductibles, aggregates. All of it is about the payout.

Experience modification rate is the one number that describes how likely the loss is in the first place, and it is not on the certificate.

What the number actually measures

EMR, also written EMOD or experience mod, is a multiplier applied to a vendor's workers compensation premium based on their own loss history compared to other employers doing the same work at the same size.

1.0 is exactly average for the class. Below 1.0 means fewer or cheaper claims than peers, and the vendor pays less than manual premium. Above 1.0 means more, and they pay more.

The comparison is against their own class code, which is what makes it useful. A roofing contractor at 0.85 is being compared to other roofers rather than to office employers, so the number is not telling you that roofing is dangerous. It is telling you this roofer is safer than most roofers.

Four bands, and the trend matters more than any of them

Broadly, with the caveat that interpretation varies by trade and by state:

Below 0.80 is a well-run safety program, and it is uncommon.

0.80 to 1.00 is better than average. Most established vendors you want are here.

1.00 to 1.25 is worse than average and worth a conversation rather than a rejection. A single bad claim moves a small employer's mod substantially, and that is not the same as a pattern.

Above 1.25 is a real signal, and in construction it is a common contractual disqualification threshold. Above 1.50 is a vendor with a loss pattern rather than a bad year.

The bands matter less than the direction. A vendor moving from 1.30 to 1.05 over three years has fixed something. A vendor moving from 0.90 to 1.20 is developing a problem, and the current number alone hides that. Ask for three years, not one.

The three-year lag is why the number is backward-looking

An experience mod is calculated from a three-year window that ends about a year before the current policy period, so the most recent year of losses is not in it yet.

Two consequences.

A vendor who had a bad year eighteen months ago may still show a good mod, because the bad year has not entered the calculation. Their next mod will be worse and they know it.

A vendor who fixed their safety program last year still carries the old number for two more years. Punishing them for it is how you lose a vendor who is now genuinely safer than their mod suggests.

That lag is the reason to ask about it rather than to score it mechanically. "Your mod is 1.31, what happened and what changed" is a two-minute conversation that tells you more than the number does.

Ask for the rating worksheet, not the number

A vendor telling you their mod is 0.92 is telling you a number you cannot verify.

The document is the experience rating worksheet, issued by NCCI or by the independent state rating bureau. It shows the mod, the three years in the window, the individual claims that drove it, and the split between frequency and severity.

That split is the part worth reading. A mod of 1.20 driven by one large claim is a different vendor from a mod of 1.20 driven by eleven small ones. Frequency predicts future losses better than severity does, because a pattern of small injuries is a pattern of behavior, while one catastrophic claim can be genuine bad luck.

The worksheet comes from the vendor's broker and takes about a day to get.

An entity change breaks the whole signal

The clean way for a vendor to escape a bad mod is to form a new entity.

New FEIN, new experience, mod resets to 1.0, and the same crews and the same management continue. Rating bureaus have combination rules meant to prevent this when ownership is substantially the same, and the rules do not catch everything.

What to look for: a vendor whose entity is under three years old, whose principals have been in the trade for fifteen, and whose mod is exactly 1.0. That combination is worth one question.

What IDCore does and does not do here

Being straight about the boundary, because this is a gap rather than a feature.

IDCore reads insurance certificates and matches them against your requirement set. Experience modification rate is not on the ACORD 25, so the document reader does not extract it, and we do not pull EMR from rating bureaus.

What the platform does hold is the vendor record and the onboarding requirement set, so an EMR document and a stated value can be collected and tracked as a required item alongside the certificates, with the same expiration and reminder behavior. Collected and tracked, not verified.

If EMR-based prequalification with automated bureau lookup is central to what you are buying, that is a construction prequalification product rather than a COI compliance product, and it is worth being clear about which problem you are solving. Anyone selling you both in one platform should be asked to show the bureau integration.

Ask three vendors for three years

Pick your three highest-risk vendors. Ask each for their experience rating worksheets for the last three years.

You will learn two things: what the numbers are, and which vendors can produce the document at all. The second one is nearly as informative as the first.

What did the trend look like?