EXECUTIVE SUMMARY
Select a physical climate risk tool for the decisions it must support. Compare vendors using consistent assets, hazards, scenarios and time horizons, then examine differences in metrics, geocoding, vulnerability assumptions and damage estimates. Validate location data and ask how uncertainty and model limitations are communicated. Before using results, establish data-quality checks, independent validation, permitted uses and escalation rules. Connect outputs to specific actions such as adaptation, insurance or credit decisions, with teams able to explain the underlying assumptions.
Financial institutions are under increasing pressure from regulators and stakeholders to provide detailed, asset-level physical climate risk assessments. The vendor landscape is complex, with “black box” models often producing conflicting results for identical properties. Even when vendors assess the same assets and scenarios, outputs can vary significantly in hazard estimates and damage ratios.
This variation does not mean one model is “right” and another is “wrong.” Asset-level physical risk modelling is inherently complex and uncertain, with vendors making different methodological and design choices. The key question for decision-makers is suitability: which approach best enables you and your teams to confidently stand behind every number, knowing you can explain and defend your choices to your board, regulators, or stakeholders. Success means having the clarity and confidence that your assessments are transparent, defensible, and directly aligned with your portfolio, decision needs, data quality, and governance requirements.
This article provides a user guide to help you navigate estimate dispersion, ask informed vendor questions, and select a vendor that is fit for purpose and decision-useful. We’ll walk you through the main drivers of dispersion, provide a due-diligence checklist of vendor questions, and outline six post-selection steps to effectively integrate results into your workflows. This roadmap will help anchor expectations and guide you through the full vendor selection process.
Two checks for a vendor comparison
When comparing physical risk tools, organise the review around two questions:
1 - Are the outputs comparable?
Use the same assets, perils, scenarios and time horizons where possible. Record the definitions and units of each hazard and damage metric before comparing results.
2 - Which modelling choices explain any differences?
Ask vendors to explain their downscaling, geocoding and asset-footprint methods, metric definitions and vulnerability assumptions. Document how those choices affect the intended use of the outputs.
Talking about metrics, flood vs wind is the simplest way to see the problem
Hazard metrics are a common source of miscomparison across vendors and perils.
Flooding is often reported in relatively comparable terms, such as flood depth in metres.
Wind and cyclones are where comparability can break down quickly. Wind speed may be reported in different units and, more importantly, can refer to different definitions (for example, sustained 1-minute vs 10-minute winds, or gusts). Some vendors also use probability-based measures aligned with cyclone category scales, such as the Saffir-Simpson scale.
The key takeaway is that there is no universally “correct” metric for all use cases. If stakeholders assume metrics are directly comparable when they are not, this can lead to false confidence, poor prioritization, and reduced business value from the assessment.
The “garbage in” problem: asset location data quality can change results materially
Incomplete or ambiguous asset-location data can cause different geocoding results and materially alter hazard assessments. Validate coordinates and footprints, investigate outliers and retain the evidence for any quantitative benchmarking result.
For model users, the same “asset” can be mapped to different locations, which may alter hazard intensity, return periods, and estimated damages. If you do not understand how a vendor handles ambiguous inputs, you are not selecting a tool. You are accepting a hidden assumption.
Questions to ask before you commit
The complexity, uncertainty, and diversity of asset-level physical risk assessment means there is no single “best” vendor. Suitability depends on intended use, data quality, and how outputs will be applied inside your governance and risk framework.
Experienced professionals will recognize that selecting a vendor involves choosing a modelling approach, an interpretation layer, and a data workflow, not just a score.
To select a suitable physical risk assessment vendor, use the following questions as a structured due diligence checklist. These are designed to assess relevance, explainability, and decision usefulness, which are critical for risk, credit, and board committee discussions.
A. Use case and decision integration
What is the intended use case: site-level decisions, portfolio screening, adaptation planning, underwriting, credit, or disclosure reporting?
Which decisions will the outputs directly inform, and what decision thresholds do you recommend?
B. Coverage and modelling scope
Which geographies and perils are covered, and where are known limitations?
Do you model direct hazard impacts only, or also indirect and systemic effects such as supply chain disruption? If not, how should we treat that gap?
C. Asset data requirements
What minimum asset location inputs do you require for reliable geocoding: address, parcel, coordinates, perimeter?
How do you resolve locations when inputs are incomplete or ambiguous, and what QC flags do you provide?
Which asset attributes materially affect vulnerability (construction type, age, elevation, floor area), and how do you handle missing fields?
D. Scenarios, horizons, and update cadence
Which scenarios and time horizons are available, and how frequently are models updated?
How do you align scenario choices to regulatory expectations and internal model risk management?
E. Metrics and comparability
For each peril, what exactly is the hazard metric, its unit, and its definition (for example gust vs sustained wind)?
How should we compare outputs across tools if hazard metrics differ?
F. Damage, vulnerability, and damage ratios
How are damages estimated from hazards, and what does your “damage ratio” represent?
Do you separate structure, contents, and business interruption, and for which asset types?
G. Validation and uncertainty
How do you validate the model, and what independent checks exist?
What measures of uncertainty do you provide, and how should decision-makers use them?
Once you have clear answers, you establish a disciplined basis for vendor selection and internal review. This is where sustainability risk management, compliance confidence, and business value begin to align.
At Futureproof Solutions, we recommend six additional steps to translate vendor outputs into actionable decisions:
Create a model intake standard across vendors: scenario, horizon, hazard metric definitions, geocoding approach, and damage ratio logic.
Set location data quality gates (minimum fields, acceptable uncertainty, required QC flags) before results enter risk systems.
Define comparability rules across perils and vendors to avoid blending incompatible metrics in decision-making.
Add a validation layer, including independent checks on a subset of assets and escalation triggers when dispersion exceeds internal tolerances.
Tie outputs to action pathways: adaptation planning, insurance strategy, capex prioritisation, supplier engagement, or site resilience measures, so impact and value are realised.
Train all relevant teams, not just analysts. Governance, finance, and asset teams should understand the meaning and limitations of hazard metrics and damage ratios to reduce misinterpretation risk.
Discuss your requirements
Contact us to discuss the decisions your physical risk assessment needs to support, the data available and the questions you need to put to vendors. Agree the scope of any support before selecting a tool or starting implementation.