The exhibit arrives. The sponsor wants the meaning.
A carrier’s claims experience report is twelve pages of tables built for an underwriter. What the plan sponsor is asking is simpler and harder: is this plan running hot, which line is doing it, and is it a trend or one bad year. blankit reads the exhibit and answers that in their own advisor’s branding.
Paid and incurred, kept apart
Incurred is paid plus the reserve for claims that happened and have not been billed yet — and it is what an underwriter rates on. Both are carried through every figure, because a plan judged on paid claims alone looks better than it is.
Period by period, not year by year
Experience windows rarely line up with anniversaries, and a multi-year summary often discloses detail for the current year only. Each period is read on its own terms rather than flattened into an annual figure that was never on the page.
However the carrier lays it out
Every insurer formats an experience report differently, and several split a line into pooled and non-pooled pages that have to be reconciled before any ratio means anything. Reading it is the job; you upload the document you were sent.
Documents are read the Canadian-first way the rest of the platform reads them, and the ladder is set out in full on Trust & security.
A blank on the page is unknown — it is never zero.
A zero invents a year that did not happen
Where a document states paid claims and not incurred, the incurred figure stays empty rather than being set equal to paid. Copying it across would flatter every loss ratio built on that period — and those ratios are what a renewal argument is made of.
The denominator is the dangerous half
Missing claims make a report say less. A missing premium once made one say the opposite — a health line carrying real claims reported as running at a 0% loss ratio, comfortably below break-even, because the denominator was absent and got written down as genuine. That shape is now refused outright.
The distinction costs something: you will sometimes see “not disclosed” where another tool would have shown you a confident number. That is the number being honest about what the carrier actually printed.
Claims with no premium behind them stop the whole document.
An insured line that reimbursed a dollar of claims was billing premium to do it. So a period showing claims against no premium is never a real year — it is a figure the reading missed, and the fix is to read the document again rather than to render around it.
Nothing renders when that happens, and every affected period is named. A partial report would be worse than none: the sponsor cannot tell which pages were built on a figure nobody had.
A line showing no claims and no premium is a different thing entirely — that is the carrier reporting a benefit nobody used, and it reads as exactly that.
A benefit the carrier printed and the reading lost stops the document too. Every other check asks whether the lines that came through hold up, so a line that never came through at all passes them by default and looks exactly like a plan that does not carry the benefit. blankit reads the carrier’s own page alongside the extraction and names the benefit, and the page it is printed on, rather than producing a report about the rest of the plan.
And where the carrier prints its own total over a block of rows, blankit adds up what it read and checks the two against each other. A figure read wrongly leaves nothing missing to notice — every column is there, and four real numbers add up to an answer that is simply not the carrier’s. The totals are held back, and the cover note gives you both figures and the difference between them, so you can settle it against the page rather than take our word for it.
A fully pooled group has no experience to report at all.
A pooled carrier prices the group against a book of similar groups rather than against its own claims, and issues no experience exhibit for it. So there is no report to produce, and blankit does not pretend otherwise: a claims document arriving for a group you have marked pooled is set aside, naming the group, rather than asking you for a target loss ratio the carrier is never going to issue.
Every refusal above says a reading went wrong and a re-read would fix it. This one says the question does not apply to this group, so it is set aside for good rather than retried on every sweep. Only you can say a group is pooled — nothing infers it, and nothing clears it on your behalf.
“Is that high?” is a question about other plans.
A sponsor told paramedical services are a given share of their health spend has no way to judge it. Set beside the same share across the wider benchmark, it becomes a finding they can act on — and the beginning of a conversation about plan design rather than about the rate.
Both sides of that comparison are recomputed on the same category map rather than taken from a percentage a carrier printed, because two figures built on different definitions make a difference that is not there. A comparison is only shown where there is genuinely enough behind it, and a firm can keep its own book out of the benchmark entirely.
The report goes out with a short cover note a sponsor can read without opening the attachment: what is in it, where each line landed against its target, the direction of travel, and anything about the data worth flagging up front. It is written to be forwarded as it stands.
Every figure in that note comes from a block blankit builds in code before any prose is written, so the numbers in the sentence are the numbers in the report by construction. A benefit line the note cannot properly narrate is named rather than quietly left out — a sponsor counting their benefits notices the gap, and a missing line reads as an advisor who did not look.
If the prose cannot be produced at all, the note falls back to a plainer written version and still sends. The report is not held up for its own covering email.
The figures do not stop at the report. The same history is what a renewal is argued from when the carrier’s number arrives, and what an underwriter is sent when a group goes to market — read once, kept once, rather than re-keyed into three places.
An experience report can reach blankit without anyone uploading it: forwarded to your firm’s own intake address, or picked up from a cloud folder the carrier drops into. The latest figures mean the most recent period, incidentally — not the most recent upload, which is a different thing the moment someone files a back year.
Send the meaning, not the exhibit.
Bring a claims experience report you have already sent on, and we will show you what your sponsor would have read instead.