Materiality Assessment Guide

How to use this guide

This page explains the topic, a proportionate scope, useful outputs and questions to ask. It is not a fixed consulting package; the actual work should follow your reporting basis, business boundary, maturity and decision needs.

What materiality decides

Materiality is the filter. Out of everything a business could conceivably report on — water, waste, emissions, labour practices, data security, supply chain conditions, community impact, board composition — materiality determines which topics get measured, managed, targeted and disclosed.

Get it wrong and everything downstream is wrong with it. You collect data nobody needs, miss the topic your investors actually care about, and set targets against the wrong things. It is also the part of a report that a knowledgeable reader checks first, because a materiality outcome that does not match the business is a strong signal about the rest.

The two lenses

Traditional materiality asked one question: what matters to the company. Double materiality asks two, and requires both to be assessed.

The two dimensions
	Impact materiality	Financial materiality
The question	How does the business affect people and the environment?	How do sustainability matters affect the business's cash flows, access to finance and cost of capital?
Direction	Inside out	Outside in
Primary audience	Affected stakeholders, communities, regulators, civil society	Investors, lenders, other creditors
Assessed by	Severity — scale, scope, remediability — and likelihood	Magnitude of financial effect and likelihood, over short, medium and long horizons
Evidence	Stakeholder engagement, incident records, site data, sector research	Financial modelling, scenario analysis, risk register, management judgement

A topic can be material under one lens and not the other. Community noise from a construction operation may have serious impact materiality and modest financial materiality. A carbon price in a key export market may be financially material long before it shows up as an impact. Both belong in the assessment, tagged for which lens brought them in.

How the assessment is actually run

Define the boundary and the value chain. Which entities, which upstream and downstream activities, and over what time horizons. Same boundary as the rest of your reporting.
Build the topic universe. Framework topic lists, sector standards, peer reports, your risk register, past incidents, media and regulatory developments. A long list before any filtering.
Screen for relevance. Remove what genuinely does not apply, and record why. The record is part of the deliverable.
Gather evidence. Internal data and documents, plus stakeholder engagement. Both dimensions need input from people, not only from desk research.
Assess impact materiality. For each topic, actual and potential impacts, positive and negative, scored on severity and likelihood. Severity itself has components — how bad, how widespread, and how reversible.
Assess financial materiality. Risks and opportunities, magnitude and likelihood, across short, medium and long term. This is where finance must participate.
Apply thresholds. Decide what score makes a topic material, and be able to justify the line.
Validate. Test the outcome with management and, where practical, with stakeholders. Get formal approval — usually board or committee level.
Document the whole thing. Method, participants, scoring criteria, thresholds, results, and the reasoning behind borderline calls.

Why it is more technical than it looks

From the outside it looks like a workshop and a chart. The difficulty is in the judgements nobody sees.

Two incompatible scales. Impact severity and financial magnitude are not measured in the same units. Combining them into one ranking requires a defensible normalisation, and most disputes about a materiality outcome trace back to it.
Threshold setting. Where the line falls determines whether you have eight material topics or eighteen. That decision has direct cost consequences, which creates a quiet pressure to draw it conveniently.
Weighting stakeholder input. Employees, customers, regulators and community groups will rank things differently. Averaging them flattens exactly the disagreement that was most informative.
Value chain reach. Impacts you contribute to through suppliers may be material even where you have limited control and poor data.
Time horizons. A topic immaterial this year can be highly material at ten years. The horizons must be defined and applied consistently, not left implicit.
Connecting to the financials. Once you say a matter is financially material, a reader will look for it in your risk disclosures and your accounts. Inconsistency between them is one of the first things an assurance provider will pick up.

The test of a good assessment

Someone independent should be able to read your documentation and arrive at broadly the same topic list. If the reasoning cannot be reconstructed, what you have is an opinion with a chart attached.

Common mistakes

Copying a peer's material topics. Fast, and it produces a report that describes a company you are not.
Surveying only management. You get the management view reflected back with added confidence.
Scoring without criteria. If “high” is not defined, different participants are scoring different things.
Letting the matrix be the deliverable. The chart is a summary. The reasoning is the work.
Never revisiting it. A materiality outcome carried unchanged for five years is a signal that nobody is looking.
Excluding a topic because it is inconvenient. The pressure is real and usually unspoken. It is also the fastest way to lose credibility with an informed reader.

How often to redo it

A full reassessment every two to three years is typical, with a lighter annual review that checks for new risks, incidents, regulatory changes, and shifts in what stakeholders are raising. Rebuild sooner if you acquire or divest something significant, enter a new market, or experience a serious incident.

Stakeholder engagement — how to run it so it produces something you did not already know.