From a company name to a societal P&L: how ImpactAccounting.ai works

Financial accounts tell you whether a company creates economic value. They say almost nothing about the value it creates — or destroys — for people and the planet. A business can grow revenue while also affecting wages, health, taxes, emissions, water, land and communities. Those effects normally live in different reports, in different units, with very different levels of evidence. That makes them impossible to compare and hard to act on.

ImpactAccounting.ai closes that gap. You give it an organization or a project. It gives you back a complete set of impact accounts: the positive and negative effects across the whole value chain, in physical units and in one comparable monetary measure of societal value.

It is not a dashboard and not an ESG rating. It works like an accounting system: it gathers evidence, builds a model, applies one consistent method, shows you where the value sits, and leaves the assumptions under your control.

This article follows exactly what you see on screen while an analysis runs, and explains what each step is doing and why you can trust its output.

The five steps of an analysis

When you press Start analysis, the processing view opens and shows the analysis progressing step by step. Those steps group into five simple movements.

The last arrow is the important one. The first result is a starting point, not a verdict. From then on the analysis is a working model you can open, question, correct and recalculate.

Here is how the steps in the processing view map onto those five movements.

You can open any step while it runs, or after, and read what it concluded. Nothing in the chain is hidden.

Step 1) Understand the organization

An analysis starts with a name, not a questionnaire. You identify the organization or project, add context, and optionally upload material such as annual reports, sustainability reports, financial data or operational figures.

Before anything is calculated, the software works out what it is looking at. What does this organization actually do? Where does it operate? What does it sell, and to whom? What does its supply chain look like? How big is it, in revenue, volumes and people?

This matters because two companies in the same sector can have completely different impacts, coming from different sourcing, technology, geography, workforce and customers. A sector average cannot see any of that. So the analysis is built on a profile of this organization.

The same step also separates today from tomorrow. Current-year figures form the impact account as it stands. Future potential is handled later, as a projection. A hoped-for future is never presented as a current result.

Not all evidence is equal

An audited report is not a marketing claim. A company measurement is not an industry proxy. A current operational figure is not a five-year-old estimate.

So the first thing the software does is judge the material it has been given. Your uploads become searchable inside the project, so the analysis can pull the exact relevant passage instead of skimming a whole report. Where your documents are silent, it can research external sources, and it records where each number came from.

Missing data does not stop the analysis. It changes how confidently a number is presented. Where there is no direct figure, the model uses a documented proxy or a range, tells you so, and tells you which missing piece of information would improve the result most.

Step 2) Map the impacts, not just the good story

Next, the software identifies the impact pathways: the traceable chains between an activity and a consequence for society.

It looks across the whole value chain (the supply chain, the organization's own operations, its products and services, and end of life) and across the three capitals:

• Natural capital: climate, pollution, resource use, water, land, ecosystems.
• Human capital: health, safety, skills, income, wellbeing.
• Social capital: taxes, avoided public costs, wider contributions to society.

This breadth is deliberate. Looking only at the mission or the flagship benefit hides the trade-offs. A product can create real value for customers while relying on a damaging supply chain. A low-carbon technology can still have land or workforce effects. A good employer can create value through wages and taxes while imposing environmental costs elsewhere.

The question is never "what good does this company claim to do?" but "what are the material positive and negative consequences of the whole business?"

The list of pathways is then verified for coherence, double counting and gaps before any number is attached to it.

Step 3) Quantify and value each impact

Each pathway becomes a transparent, bottom-up calculation. The logic is always the same four ingredients.

Take a training program. The output is the number of people trained. The outcome is the improvement in employment, income or wellbeing per person. Additionality removes what would have happened anyway, accounts for how long the effect lasts, and attributes only the share that belongs to this organization. The value factor converts the outcome into comparable societal value.

The same four ingredients work for emissions, injuries, wages, taxes, education, health, ecosystems or customer benefits. Geography and time are always explicit, because a tonne of pollution or a year of better health does not mean the same thing everywhere.

Every pathway is valued with the eQALY method, which expresses outcomes as changes in human wellbeing before converting them to money. Environmental pathways additionally draw on established life-cycle inventory data, including ecoinvent processes, and resolve to country-specific natural-capital value factors. For a step-by-step walkthrough of that calculation, see The eQALY method: value your first impact.

Money is not a claim that everything can be priced. It is a common decision unit, so that unlike impacts can finally be compared, ranked and discussed in the same conversation. The physical measures stay visible underneath.

Step 4) Check whether the result makes sense

A correct equation can still produce an implausible answer. Two steps exist purely to catch that.

Calibration checks the scale and composition of the model against what the organization actually is, and identifies which few variables are driving the result. Deep dive looks for weak evidence, missing pathways, extreme assumptions and internal inconsistencies.

The outcome is not a single number presented with false precision. You get low, expected and high values, and confidence levels and sources that stay attached to each variable and pathway instead of disappearing into a total.

That is why the results view shows gross value created, gross value reduced, net societal value, the split by capital and by value-chain step, the biggest positive and negative pathways, physical indicators, data-quality signals, risks and opportunities — not just a headline.

Step 5) Explain, and look ahead

Prediction describes how the organization could move from where it is today toward its intended scale, keeping the projection separate from the current account.

Insights writes the interpretation: the top pathways, the trade-offs, the red flags, the data gaps and where to focus next — in language you can take into an investment committee without an analyst translating it first.

Where AI helps, and where it does not

This is the part that decides whether you can trust the number, so it is worth being precise.

AI is used where judgment and language matter: reading documents, researching evidence, understanding a business model, proposing pathways and variables, and explaining results.

The arithmetic is not left to AI. Once pathways, variables, factors and assumptions are set, the calculation is performed by fixed code. The same inputs always produce the same output. AI builds and interrogates the model; a deterministic engine does the accounting.

Everything the model relies on stays visible: every variable has a source, every assumption has a reason, every result can be traced back to the evidence behind it. If you disagree with a number, you can find it, and you can change it.

You stay in control

After the first run, the model is yours to interrogate. Open any pathway down to its variables, equations, citations and assumptions. Change a variable, adjust an additionality factor, correct a geography, remove a pathway, or add better evidence. Everything downstream recalculates from the corrected foundation — nothing is overwritten by hand.

The project chat makes this accessible without any technical skill. Ask why a result is high, request a breakdown, challenge an assumption, inspect the evidence, or build a scenario in plain language — even in the middle of a meeting. Edits are validated before they are applied, and every change is recorded in a history you can revert or restore.

For heavier questions, the assistant can work against an export of your project in an isolated environment with no network access, so pivots, breakdowns and what-if calculations run on your own data.

From one company to a portfolio

At project level you get an impact statement: a P&L-like view of societal value across the three capitals and the four value-chain steps, plus forecasts, scenarios, physical KPIs, an insights summary and optional CSRD- and SFDR-oriented readouts.

At workspace level, projects come together into a portfolio. Compare holdings on the same basis, see which companies drive value or risk, spot concentration, review the coverage and quality of the underlying models, and look at societal value against invested capital. Different forecast horizons are respected, and holdings that cannot be included in an aggregation are flagged rather than silently counted as zero.

Results export as Excel, PDF and structured data, and a portfolio can be shared read-only. The same model then serves screening, due diligence, engagement, board material and reporting.

The point is better decisions, not a better report

The output looks like a societal P&L. The value is the model behind it.

One comparable unit makes scale and trade-offs visible. Physical indicators keep the real-world meaning. Sources and confidence levels tell you how much weight to put on each result. Scenarios turn the account into a forward-looking tool. Portfolio views make different investments comparable. Editable assumptions keep expert judgment where it belongs: with you.

ImpactAccounting.ai does not remove uncertainty and does not replace your judgment. It makes the reasoning explicit. By connecting evidence, activities, outcomes, additionality and value in one traceable system, it makes impact information as usable as financial information.

Where to go next

• Assessing an organization — run your first analysis end to end.
• The eQALY method: value your first impact — the valuation logic, worked by hand.
• Impact valuation: questions and answers — methodology, evidence and limitations.