Explanation
Financial value, or what your impact is worth to the company
Whatever the harm to people and nature, does any of it actually reach the P&L? How to read the Deck's Business case page, its levers and their evidence labels, and the checks worth running before the result meets a CFO.
Valued in August 2026, when each pathway carried an additionality percentage and the business case was the Financial Value screen of the Model. Results now value each pathway against its alternative where it has one, and show an absolute and an additional impact, as the net impact bridge explains. The business case is now the Business case page of the Deck, and the screenshots below show the August screen.
Nestlé's societal value for 2025 models at roughly minus 35.5 billion dollars. That is a defensible number and a hard one to take into a room where the next question is about the accounts. Whatever the harm to people and nature, does any of it actually reach the P&L?
That question has an answer, though not the one people expect. Societal value and financial value are two lenses on the same set of activities, and they are never summed. The Business case page of the Deck takes the pathways already modelled in the assessment and asks a different question of each one, namely whether it plausibly touches revenue, costs or risk, and by how much.
What the page gives you
Open the company in the Deck and pick Business case. It is prepared when a release is published, except for a non-profit, where an editor prepares it on request. One AI pass works through the modelled pathways and returns a set of levers, each tied to a business mechanism rather than a societal outcome. An editor can Rebuild it with notes that steer the pass, such as the company's real gross margin, or ask the Deck's chat to change, add or remove a single lever.
The page reads from top to bottom. The AI's summary comes first, saying how the impact makes money, which lever carries most and which assumption to test first. Below it, a chart shows the company's sales or net profit forecast for each year of the scenario, with what the levers add and take away. The Levers table lists each lever by year, with a total row, Value the impact adds, and a tick box an editor uses to leave a lever out of every figure.

The Business case page today, on the Le Cube example.

The August 2026 screen, before it became the Deck's Business case page.
In August, for Nestlé, the screen returned gains of 272 million dollars against losses and risks of 631 million, leaving a net annual financial value of minus 359 million, with the societal figure of minus 35.5 billion beside it, unchanged and uncombined. The Business case page keeps the two apart in the same way: the line under its heading says it is shown beside the additional impact and never added to it.
Two numbers, two mechanisms, no arithmetic between them. Societal value prices well-being and welfare. Financial value prices margins, avoided costs and probability-weighted risk. The gap between 35.5 billion and 359 million is not an error, it is the point. Most of what Nestlé costs society never reaches Nestlé.
Reading the result
The net figure is the least interesting number on the page. What matters is which lines produce it and how much weight each one can carry.
Start with the line that decides the answer
Twelve lines make up the Nestlé screen, and one of them carries almost the entire positive side. Brand and reputation on pathway [20], fortified nutrition intake, comes in at roughly 237 million dollars, which is 87 per cent of all gains. The only other gain on the page, the packaging recycling line, is worth about 35 million.
Opening the line shows what it rests on. On the Business case page, a click on a lever's row opens its calculation, with each figure marked by where it comes from: from the model, from company figures, or an assumption to test.

The calculation is 135 billion fortified servings at half a cent of incremental margin per serving, realised in full, with 35 per cent attributed to Nestlé. Every line also carries an evidence label, bronze, silver or gold, and this one sits at bronze, the weakest of the three, meaning it rests on a generic consumer-goods reputation margin applied to servings as unit-equivalents rather than on anything measured at Nestlé.
The source publishes the range that generic factor spans, from a tenth of a cent to two cents per serving. Run the line at each end and the whole result moves.
| Margin per serving | Brand line | Net financial value |
|---|---|---|
| $0.001 (low) | 47M | (549M) |
| $0.005 (base) | 237M | (359M) |
| $0.020 (high) | 948M | +352M |
One factor, moved within the range its own source declares, takes the company from destroying roughly 549 million dollars of business value a year to creating 352 million. The sign flips without anyone having done anything wrong.
That is the first check worth running on any business case. Find the line carrying the most weight, open its source, and see what the stated range does to the conclusion. The source note on this one says as much itself, asking for an actual category contribution margin and a measured demand response in place of the generic figure.
A risk and a cost are not the same number
Not every number in the cost column means the same thing.
Dairy and animal-derived ingredient sourcing on pathway [12] is the largest cost on the screen. It is 12.5 billion kilograms sourced at ten cents of volatility cost per kilogram, carrying a 25 per cent annual shock probability and 60 per cent attribution. The 188 million dollars shown is an expected annual loss rather than an annual bill. If the shock lands it is nearer 750 million, and in three years out of four there is none at all.
Two rows below it the screen puts a very different kind of number.

Recall-related product destruction on pathway [42] is 67 million kilograms of destroyed product at 1.97 dollars a kilogram, realised in full and attributed at 90 per cent, for 119 million dollars. It is also the best-evidenced line on the page, because that unit rate is not a generic factor at all but Nestlé's own disclosed 2025 recall inventory write-off of 110 million Swiss francs, converted into dollars and spread across the destroyed mass.
Worth noting what the 90 per cent is doing there. The unit rate is back-derived from a figure the company reported as its own write-off, so output times outcome already equals the disclosed cost. The attribution then trims a known number down by a tenth, on the basis that not every recall traces to the pathway being priced. That is defensible, and it is also the one place on this screen where the best evidence produces a figure lower than the company's own accounts. Worth knowing before a finance director asks.
Two rows apart in one table, then, and two entirely different kinds of number. One is a probability-weighted estimate resting on a generic factor. The other is money that has already left the business, evidenced by the company's own accounts. A finance director who reads them the same way will either over-provision against the first or stop trusting the screen.
Ask what was left out
Every business case keeps a reconciliation note, whose job is to say what the AI pass did not do. For Nestlé in August it confirmed that recycling avoidance, plastic packaging cost and recall write-off run through separate physical or regulatory channels rather than monetising one quantity three times. It then listed what was left out on purpose, including general wages, taxes and routine utility costs, on the grounds that these are business as usual rather than impact-created value.
The omissions matter more than the confirmations. Carbon-price exposure was absent from the Nestlé result because the pathway set produced no tonnage reconciled well enough to price, and a reader who only looks at the levers will not know that. On the Business case page, ask the chat what the business case left out and why. Levers the pass checked but found below 0.1 per cent of sales in every year are listed, folded, at the foot of the Levers table as checked and not counted.
When the answer comes out positive
Nestlé is a large company with a long value chain and most of its financial exposure on the cost side. A smaller, single-product business produces a different shape.

Nutris, a fava-bean protein producer, returns eight lines rather than twelve, and the two gains outweigh the six costs. Avoided agrochemical production from regenerative practices contributes 222,000 dollars of avoided cost, and avoided animal-ingredient production contributes 103,000 dollars of revenue, against costs and risks totalling 40,000 dollars. Net financial value is positive at 286,000 dollars, while societal value is still negative at 9.9 million.
Two things are worth noticing beyond the sign. The gains sit in avoided cost and substitution revenue rather than in a reputation proxy, which is a sturdier place for them. And the reconciliation shows the double-counting control doing real work, holding the regenerative fava volume apart from the conventionally sourced volume so that the regenerative benefit and the sourcing cost cannot both claim the same beans.
Working with the result
The business case is not a report that arrives finished. Three things are worth doing before anyone else sees it.
1. Ask the chat to explain a line before changing it. Any figure can be interrogated in plain language, and asking why the brand line uses half a cent per serving returns the reasoning and the source faster than inferring it from the fields. This is the right first move on any number that looks wrong, because a surprising number is often correct and badly labelled.
2. Ask it to adjust what you know is wrong. An editor changes a lever through the Deck's chat directly, in the form of instructions like setting the probability of a particular line to 15 per cent, and the row is marked as edited. Where a real figure exists, give it and let the line be revised. Actual category margins, audited procurement volumes, a measured demand response. Each replacement lifts a line off its generic factor onto company evidence, and doing that for the two or three lines that carry the conclusion is worth more than improving everything else combined. The Calibrate, don't re-run workflow covers this in full.
3. Publish again after recalibrating. A lever names the figures the model holds, such as an input, a pathway's output or one of its shares, and reads them in every year of the scenario. Recalibrating the model and publishing again moves those levers, and each release gets its own business case. Rebuild only to change what the levers are; a rebuild writes every lever afresh, edited ones included.
Once the result holds up, it is ready for the committee conversation. Every line keeps its assumptions and its evidence label attached, so a reader can see what the headline rests on rather than discovering it in the meeting.
Where it stops
The business case is a screening tool, and the page says so at its foot: "A first estimate to test with company data, not a forecast of the company's accounts; costs and risks are expected values."
It is not an audited figure, a cash-flow forecast, an accounting provision or an input to financial reporting. It should not be added to reported profit either, since the costs already sitting in the accounts were deliberately kept out of it. It does not prove that every financial consequence has been found, and it cannot confirm that its factors are accurate, only that they are visible and inspectable. A result still resting on generic factors is the start of a conversation with Finance rather than a conclusion to hand them.
What sits behind the numbers
Financial value follows the ROImpact Method, published by Valuing Impact as the business-value lens of its wider framework. Where the eQALY lens asks what an activity is worth to society, this one asks only what it is worth to the company, in revenues and costs. Assets and liabilities are left out deliberately, on the grounds that they are harder to act on in everyday management decisions. The method is symmetric, so a lever can destroy business value as easily as create it, and both sides are counted.
The equation
The method writes every line as the same calculation.
value = output × outcome ($/unit) × probability × attributionOutput is a countable result the business already tracks, such as tonnes of waste, kilograms sourced or days of disruption. Outcome is the money value of one unit of that output, taken from financial data rather than from societal value factors. Probability and attribution then adjust for how much of that value is really realised, combining the chance the effect occurs at all with the share of it that can credibly be linked to the company. The method paper groups the two as a single term, additivity. It is not the same construction as additionality in the eQALY method, which is now the additional impact over the absolute impact, read against an alternative, and the two should not be read across.
The Business case page writes the same calculation in its own terms, at the foot of the page.
lever = driver × value per unit × sharesThe driver is the output, the value per unit is the outcome, and the shares hold probability, attribution and any other fraction that says how much of the value is due to the impact, such as the share of customers who would not get the benefit elsewhere. Where the model already holds a driver or a share, the lever reads it from the model rather than writing a number of its own.
The fourteen levers
Each line is assigned to exactly one lever, drawn from four groups. Market covers brand and reputation, IP and innovation, sales and market share, and price premium. Finance covers taxes and subsidies, capital expenditure, and debt and interest. Risk and compliance covers licence to operate and business disruption, fines and regulations, and sourcing costs and resilience. People and operations covers other operating expenses, payroll and employee costs, talent acquisition, and talent retention.
The one-effect-one-lever rule matters more than it sounds. A single activity often touches several levers at once, and counting the same saving under two of them inflates the total quietly rather than obviously. Nestlé's twelve lines used six levers between them, and the reconciliation note exists to show that no two of them monetise the same output and effect.
The method makes probability mandatory on three of the fourteen, namely licence to operate and business disruption, fines and regulations, and sourcing resilience, because each depends on an event that may not happen. Everywhere else it is applied only where the uncertainty is material enough to change a decision.
Where the numbers come from
Lines at the weakest evidence level start from generic factors, each with a low and a high value, so one is a consumer-goods reputation margin, another a disposal-cost factor, another a commodity price-volatility factor. Moving a line up a level means replacing that generic factor with company evidence, and the method expects that evidence to come from finance first. Contribution margins, payroll baselines and unit costs sit with finance or controlling teams, and using their figures is what lets a result survive contact with a CFO. Unit data comes from wherever it already lives, so operations and procurement for volumes and waste, HR for headcount and turnover, legal or compliance for fines and breaches.
Two habits carry most of the credibility. Measured results come before modelled ones, and every figure keeps a record of its source. That record is what turns a screening estimate into something repeatable.
The full method, including the lever playbook and a four-step protocol for estimating probabilities, is set out in From Impact to P&L, Meet the ROImpact Canvas.