How the number comes about.

A portfolio-level case and a door opener. The tool answers whether an adaptation programme pays off for the company as a whole, and it opens the deeper conversation. Single projects need their own appraisal. It covers physical climate effects such as heat, flood, storm and drought.

The model assumes that adaptation at your sites can work. All eight levers calculate inside the Adaptation Sweet Spot, where adapting on site is possible and economical. When a site becomes uninsurable and relocation is the only answer, the decision changes and this model stops applying.

Enter only what your company carries itself. Leave out losses your insurer pays and costs you pass on to customers or tenants. Where cover is capped or pass-through is limited, the part above the cap is yours.

Under the headline sentence the result shows the contribution of each lever in EUR per year and as a share. Users told us the profile is the most useful output: it shows where the effect hits.

Seven levers add up to a gross annual benefit: the adaptation benefit of A1, B1 and B2, plus A2, C, D1 and E. D2 works through the discount rate. Benefit before OpEx = gross × confidence − maladaptation. The confidence haircut books less of the calculated upside when the inputs are estimates (100% = no haircut). Maladaptation is spend that delivers little or no benefit.

Payback rule: budget = payback target in years × benefit before OpEx. Hurdle rule: budget = benefit before OpEx × annuity factor at the hurdle rate over ten years, with the hurdle rate lowered by the D2 effect. Example company: a benefit of EUR 102.1 M a year carries about EUR 510 M at a five-year payback and about EUR 630 M at a 10% hurdle rate. The headline reads: it may be worth investing about EUR 510 M in climate adaptation.

Many effective measures cost money every year instead of once: forecasting, early warning, monitoring, maintenance. A split slider sets the share of the budget that goes into such running measures. Yearly OpEx = share × benefit before OpEx; CapEx = rule × (benefit before OpEx − yearly OpEx), so the CapEx still pays back within the target. The example value is one part OpEx to five parts CapEx (17%), a simplifying assumption without a robust benchmark: about EUR 420 M CapEx plus about EUR 17 M a year. Every EUR 1 M of yearly OpEx uses the payback target in EUR M of the CapEx budget.

The result splits the benefit into avoided cost (A1, the extra cost in A2, D1), protected revenue and output (the lost sales in A2, B1, B2, E) and protected asset value (C). Only the first two reach the P&L, so the dashboard also states the figure without C.

Every figure at the result carries a range. The tool calculates every lever once lower and once higher than the user's setting, by as much as its basis allows: ±20% where the figure rests on loss records and the user's outlook (A1, B1, B2), ±30% for estimates (C, D1) and ±50% for expert judgement (A2, E). A lever never exceeds its unadapted exposure. The range shows how far the result could plausibly move. It is no statistical interval, and the percentages are working assumptions of the Working Group.

It is an upper bound for what the assumed benefit justifies and makes no spending recommendation. It applies to new, planned measures and to the climate-driven increment only: for a roof renewed anyway and built stronger, the extra cost counts.

Users with a budget in mind can still enter it and read payback, NPV at the adapted WACC over ten years, and IRR. Headline figures are rounded to two significant digits, because the inputs carry no more precision than that.

A1 captures cost-side effects of a disruption event (repair, rework, scrap, expediting, deductibles) hitting OPEX. B2 captures revenue-side effects of the same event (lost output, delivery failure, customer or contract loss) hitting the topline. Together they describe the full P&L impact.

B1 is lost working time, valued at personnel cost (reduced shifts, slowdowns, heat-related absence). B2 is interrupted operations and lost sales measured against revenue. The same event can trigger both, but they hit different P&L lines; do not double-count.

A1 covers single-hazard, recurring, observed-frequency events at known exposure points. A2 covers what reaches the company from outside: compound, cascading and up- and downstream value-chain events including ecosystem and supply chain risk. Different evidence bases.

Both can hold lost sales. B2 counts sales lost because the company’s own operations stand still. A2 counts sales lost while its sites run, because a supplier, a customer or infrastructure fails. Book each lost sale in one place only.

A2 is the cash-flow effect of avoided value chain and tail losses (hits NOPAT). D2 is the discount-rate effect of reduced volatility priced by capital markets (hits WACC). Linked through volatility, financially distinct.

A1 is avoided losses, the damage that does not happen. D1 is avoided premium escalation and retained insurability, what insurers charge for residual risk. Both real and distinct.

D1 covers insurance counterparty effects (premiums, coverage, insurability). D2 covers lender, bond-market and rating-agency effects (cost of debt and equity). Priced independently by separate counterparties.

B2 is protected revenue under disruption: did existing revenue come through? E is the preserved or expanded revenue base: is the base still there and growing?

C is preserved asset value (balance sheet). E is revenue and growth (income statement and growth rate). Distinct cash-flow channels that reinforce each other in enterprise-value terms.

Loss records are a reference point. The lever runs on the expected future loss without adaptation, suggested from the records and the climate pathway until the user enters an own outlook. The adaptation benefit is the only figure that flows into the result.

The business case looks ahead, so A1, B1 and B2 run on the loss a company expects per year over the next ten years. The climate scenario supplies the starting suggestion for that figure: average loss in the records multiplied by a single physical-hazard intensification factor. Users overwrite the suggestion when they have their own outlook or modelled figures, and the scenario then no longer moves that lever. The other levers (A2, C, D1, D2, E) are not scenario-dependent in the current model. Stranding (C) and insurability (D1) therefore do not yet react to warming; this is a stated limit for 2026 and a candidate for the next model round. The dashboard shows the chosen pathway and its factor at the result, and users can overwrite the factor with their own.

Run SSP2-4.5 as the central planning case, SSP3-7.0 as a stress test and SSP1-2.6 as the optimistic bound. We attach no probability to any pathway, and neither does the IPCC: these are internally consistent futures. Run the case on at least two and report the range. If your climate reporting already anchors on a specific pathway, use that one.

The dashboard offers three IPCC SSP pathways, chosen for comparability with how companies typically run their own climate scenario assessments. Each is mapped to a warming level and an intensification factor: SSP1-2.6 (1.7°C by 2060, 1.8°C by 2100; factor 1.3); SSP2-4.5 (2.0°C by 2060, 2.7°C by 2100; factor 1.5, the default); and SSP3-7.0 (2.1°C by 2060, 3.6°C by 2100; factor 1.9). An earlier 1.5°C-by-2050 option was dropped, since leading with a "nice" pathway understates a future where above-2°C is more likely than not.

They are anchored to each pathway's warming level and to IPCC AR6 evidence that extreme-event frequency rises steeply with warming: a 10-year heat extreme is about 2.8 times more frequent at today's ~1°C, 4.1 times at 1.5°C, 5.6 times at 2°C and 9.4 times at 4°C, relative to 1850-1900 (IPCC AR6 WGI, Figure SPM.6). Because economic losses scale sub-linearly with hazard frequency (adaptation, exposure limits and insurance absorb part of the increase), the factors are set well below the raw frequency multipliers. They are deliberately conservative, illustrative proxies, not official IPCC loss figures.

The factor scales losses in a simple, near-linear way with warming. This is a non-scientific simplification: real impacts likely rise non-linearly beyond about 2°C, with tipping points and a near-uninsurable world approaching 3°C, and the cumulative benefit over time is closer to a hockey stick than a straight line. We keep the linear view on purpose, to stay simple and conservative rather than to model the exact damage curve. A non-linear option is a candidate for a later revision.