The question operators actually ask
If you run a funeral home, a cemetery, or you sell into them, the question is rarely national. It is:
> Is the county I am in going to have more funerals in ten years, or fewer?
National mortality is projected carefully by the federal statistical agencies. County mortality is where it gets hard, because the answer depends on the age structure of that specific county and how it is changing — and two counties with identical population can be a decade apart in median age.
What the dataset is
50,304 rows. 3,144 counties. Every projection year from 2025 to 2040.
Each row carries the county's projected total population, its expected annual deaths, and the implied deaths per 1,000 — plus, on every row, the named source for both the population input and the mortality input.
That last part is deliberate and it is the part we would ask about if someone were selling this to us. A mortality projection is a chain of assumptions. If the chain is not visible on the row, you cannot tell whether you are looking at a measurement or a guess.
How the projection works, in plain terms
Two inputs, combined:
1. Population by age, projected forward per county. 2. Age-specific mortality rates — the probability of death within each age bracket.
Multiply the second by the first, bracket by bracket, and sum. That is indirect standardisation, and it is the standard demographic approach for exactly this problem.
The reason it works is that mortality within an age bracket changes slowly and predictably, while the age structure of a county changes faster and is the thing that actually drives local demand. A county that is ageing will see rising deaths even with flat or falling population. A growing county full of young families can see falling deaths for years while its population climbs.
Those two counties look identical on a population chart and opposite on a demand chart. That gap is the entire reason this dataset exists.
What it will not tell you
Stated plainly, because a forecast sold without its limits is a liability:
- It is not a revenue forecast. Expected deaths is not the same as funerals served, and it says nothing about price, market share, or what share is cremation versus burial in your county. - It will not anticipate a shock. Projections are built from demographic trend. A pandemic, an opioid wave, or a large facility opening or closing is not in them. - County boundaries move and merge, rarely but not never. FIPS codes are the join key for a reason. - The further out, the wider the honest error bar. 2027 is a much stronger claim than 2040.
Why the methodology travels with the data
We publish the method alongside the numbers, and every row names its own sources. Partly that is because a buyer should be able to audit what they bought. Mostly it is because we have watched what happens when a derived number loses its provenance: it gets cited, then re-cited, and eventually someone makes a siting decision on a figure whose origin nobody can reconstruct.
A projection you cannot audit is a rumour with a decimal point.
Try it before paying
Connect an MCP-capable assistant to `product.kcalbinmcp.uk/mcp` and query death-care demand directly. No account needed. County names stay masked until you subscribe — so you can confirm the data exists, and see the population and the rate, before paying for the detail.