Put your head in the oven and your feet in the freezer, and your average body temperature looks perfectly normal. No doctor would sign off on that chart. Yet this is exactly what a single climate risk score does to an asset. It takes a dozen very different readings, averages them into one calm-looking number, and hands you a verdict that can hide the one thing most likely to hurt you.
The single score is everywhere. It is the number that ends up in the board pack, the screening tool, and the disclosure. If it reads “Moderate,” people relax. Sometimes they should not.
A single climate score is easy to trust, and that is the trap
A composite climate risk score is one value, usually on a scale of 1 to 5, that blends every hazard at a site into a single rating. It is useful. You can rank a hundred sites with it, sort a portfolio, and show a trend over time. The problem is not that the score exists. The problem is what people do with it: they read “Moderate” as “we are fine” and stop looking.
A composite is a summary. A summary is only safe if you know what it left out.
Problem one: averaging buries the extreme
The first failure is the oven-and-freezer effect. Average twelve hazards and a single Extreme reading disappears under eleven quiet ones.
Take a real run from our worked example, a foundry in Chennai. Its composite score comes back Moderate (2.39). Read on its own, that says “nothing urgent here.” But look at the twelve hazards underneath it. Water stress is Extreme. Severe storm is Extreme. Heat is High. Two of this site’s hazards are at the top of the scale, and the average has quietly smoothed them flat. The headline is the normal body temperature. The Extreme is the head in the oven.
The fix people reach for first is to stop averaging and show all twelve hazards, the way a full physical climate risk assessment does. That is necessary, and it helps. But it is not enough, because of the second problem.
Problem two: not every hazard sits on a vital organ
A burn on your hand is not a burn on your heart. Both are real injuries, but one ends your day and the other ends you. Hazards work the same way. A foundry depends on cooling and quench water, and it runs hot, so water stress and heat threaten the work itself. A dry-goods warehouse a mile away barely notices the same heat.
A flat list of twelve hazards still treats the toe and the brain the same. Every reading counts equally, even though the business does not depend on them equally. To read the score properly, you need to know not just how high each hazard is, but how much this particular operation depends on being spared it.
The fix: weight each hazard by what the site depends on
The repair is to keep the platform’s score as the anchor and apply a tilt based on operational dependency. Each hazard gets a weight from 0 (not material to this operation) to 2 (a critical dependency), drawn from what the business actually relies on. This is the same idea sustainability reporting standards use when they ask which risks actually matter to a business. You are not inventing risk. You are saying which of the existing risks would actually bite.
Apply that to Chennai and the picture changes. The foundry moves from Moderate (2.39) to High (2.91), a tilt of +0.52, driven by the water and heat it depends on. The hazards were always there at the top of the scale. The flat average had buried them under nine quieter ones.
Two design choices keep this honest. First, the platform’s own number stays the anchor, so you are not throwing away the model, only adding a business lens on top. Second, the original and adjusted scores are always shown side by side, so anyone can see exactly what the weighting did and argue with a weight. A tilt you cannot inspect is just a different black box.
A tilt can also go down. When a site’s worst hazards are not the ones it depends on, the adjusted score falls. That is the method working, not failing.
Peak and breadth: two views, not one number
There is a second way a single number compresses the truth, this time across sites. One score per hazard tells you how bad it gets at the worst site. It does not tell you how widespread the hazard is.
These are two different decisions. The worst-case view tells you where to act site by site. The breadth view, the count of sites where a hazard is both highly rated and material, tells you what to fund across the whole portfolio. In our worked example, heat is material at 9 of 10 sites, which argues for a portfolio-wide program. Severe storm is critical at only a few, which argues for targeted, site-specific resilience. Read either one alone and you will either overspend or miss something.
A buried hazard understates your disclosed exposure
The problem is not only a management one. Climate disclosure increasingly asks companies to identify the assets and activities that are vulnerable to physical risk, and to put a financial figure on that exposure. A compressed composite quietly understates it, because the sites whose real exposure was averaged away do not show up as vulnerable. A view that keeps the hazards visible and weights them to operations is a far stronger basis for that disclosure.
One distinction matters here. The weighting described above is operational dependency: does the work rely on this resource or condition. That feeds the financial question, whether the risk is big enough to matter to investors, but it is not the same thing. Keep the two clearly labelled.
How to tell if your climate risk score is fooling you
A short diagnostic for any climate risk score you are handed, whether it arrives in a dashboard or a climate risk assessment report:
- Can you see the underlying hazard readings, or only the headline?
- Does a single Extreme show through, or does the average wash it out?
- Are hazards weighted to your operations, or treated as equal?
- Is the weighting visible and auditable, or hidden inside the model?
- Does a score of zero mean safe, or just no data?
If the answer to most of these is “only the headline,” the number is doing the oven-and-freezer trick on you.
What the adjusted score is not
The composite is still a fine tool for ranking and triage. It is just not a verdict on whether you are safe. The materiality weights are assumptions, which is why they should always be shown rather than hidden. And how much a site depends on a resource, again, is not the same as how much money is at stake. The goal is not a single perfect number. It is to stop trusting one number that cannot carry the weight.
See it on a real portfolio
You would not trust your average body temperature with your head in the oven. Do not trust a single climate score with your most material hazard buried inside it.
We worked this end to end across ten manufacturing sites, with every hazard, weight, and adjustment visible. You can read the full worked example in our manufacturing guide.
Worked example for illustration, built on a fictional portfolio. Climate projections from NASA NEX-GDDP-CMIP6; water stress from WRI Aqueduct; flood depth and damage from JRC and FEMA HAZUS. Continuuiti publishes worked examples and tools on top of these third-party datasets.
Frequently Asked Questions
What is a composite climate risk score?
A composite climate risk score blends every hazard at a site into one rating, usually on a scale of 1 to 5. It is useful for ranking sites and tracking trends, but because it averages very different hazards together, it can hide a single extreme reading.
Why can a composite climate risk score be misleading?
Averaging buries the extreme. One hazard rated Extreme can disappear under several quieter ones, so a site with a top-of-scale exposure can still read Moderate. The headline looks calm while the real risk sits underneath it.
How do you fix a composite climate risk score?
Keep the underlying hazard readings visible and weight each hazard by how much the operation depends on being spared it, on a 0 to 2 scale. Show the adjusted score next to the original so the weighting is auditable rather than hidden inside the model.
Does a low climate risk score mean a site is safe?
Not necessarily. A low or Moderate composite can still hide a hazard that is rated Extreme and material to operations. Always check the underlying hazards and how they map to what the site depends on, not just the headline number.
