The platform · In development
Ten steps, one evidence model.
Carbonex is being built around a single journey that runs from setting your reporting boundary to submitting the report – and every step of it writes to the same record.
How evidence moves through the platform.
Nothing bypasses the evidence layer. A figure that reaches a report was assembled from values that each carry a state, a source and an owner.
Each step uses the same four-part structure: What it is · What Carbonex is designed to do · What gets recorded · Who acts here.
Step 1 — Configure
Setting the boundary. Organisational structure, legal entities, sites and facilities, reporting periods, how figures are consolidated, what counts as material, and who does what.
Hold the boundary in a structured form rather than in institutional memory. Every requirement, calculation and report that follows builds on it, so a change to the boundary is designed to carry through by itself rather than be chased by hand.
The setup itself, every change to it, who made the change and when.
Platform administrators and sustainability managers.
Step 2 — Generate Requirements
Turning the boundary into a specific list of what data is actually needed.
Work out exactly what has to be collected – which entity, which activity, which period, which method, which unit, and who is responsible for it. Nobody should be sent a blank spreadsheet and left to work out what is wanted.
Every requirement produced, the setup it came from, and who owns it.
Generated by the platform. Reviewed by sustainability managers.
Step 3 — Collect
Requirements become requests, and the requests go to the people and organisations who hold the data.
Ask each contributor for exactly what is needed from them, in their own language, with the evidence attached as it is submitted rather than chased afterwards. Suppliers take part directly, with access limited to what concerns them, instead of sending email attachments.
The submitted value, its unit, its evidence, its submitter and its timestamp.
Data contributors and suppliers.
Step 4 — Validate
Checking what arrived before it goes anywhere near a calculation.
Run structural, unit, range and consistency checks at the point of entry, and surface problems to the person who can actually fix them – the contributor, while they still remember the context – rather than to a consolidator three weeks later.
Every validation outcome, including the ones that passed, and every correction with its reason.
Data contributors, with sustainability managers overseeing.
Step 5 — Resolve Gaps
Deciding, explicitly, what to do about everything that did not arrive.
Require a decision instead of allowing a default. Where a value is missing, someone has to say whether it is unavailable, not applicable, or to be estimated. An estimate requires a stated method and a named approver before it can be used. There is no path through this step that produces a silent zero.
The state assigned, the reason, the estimation method where one was used, and the approver.
Sustainability managers and approvers.
Step 6 — Calculate
Turning activity data into results, using methods and emission factors.
Apply the chosen method and emission factors to the collected activity data, and keep a link from every result back to everything that produced it – the factor and where it came from, the method, and the activity data behind it.
The result, the method, the factor and its version, every input, and the calculation’s own timestamp.
Performed by the platform. Configured by sustainability managers and administrators.
Step 7 — Review
Someone accountable looks at the result before it becomes a figure anyone quotes.
Present the result together with where it came from, how good it is and how complete it is, so a reviewer sees not just the number but how solid it is. Comments and queries stay attached to the figure rather than living in an email thread.
The review, the reviewer, the comments, and any values sent back for correction.
Reviewers and approvers.
Step 8 — Approve & Lock
The moment a period stops moving.
Record the approval permanently, then close the period against the exact information that existed at that moment. Afterwards the underlying data can keep changing; what was reported for that period does not.
The approval, who gave it, when, and the complete set of evidence as it stood.
Approvers, acting with the authority they have been given.
Step 9 — Report
Generating outputs.
Produce every output from a closed period rather than from data that is still moving, so the same report run twice is designed to give the same numbers. Each output says which period it came from.
Every report produced, the closed period it drew on, and who produced it.
Sustainability managers and executives.
Step 10 — Share & Submit
Getting the report to the people and destinations that need it.
Share reports with boards, assurance providers and submission destinations together with the evidence pack that supports them, so that the first question an assurance provider asks does not require a phone call to answer.
Every share, its recipient, its scope, and its timestamp.
Sustainability managers, executives, and auditors or assurance providers receiving.
One record, written once.
Nothing is erased. What was written stays written. A correction is added as a new entry with its own author, time and reason, sitting alongside what it corrects rather than replacing it. This is the property that makes everything else possible.
Every figure knows where it came from. A figure carries its origin, who submitted it, what evidence came with it, and what has happened to it since.
A path all the way back. From a figure in a board pack, back through the calculation, to the method and factor version, to the activity data, to the evidence file, to the person who submitted it and the person who approved it. Designed to be followed without a phone call.
What “locked” means, and what it does not.
Locking a reporting period fixes it to the exact information that existed at that moment. Reports produced from it are designed to be repeatable: run one again next year and it should give the same numbers, because it is reading the same fixed information.
Locking does not stop the world. Data can keep arriving, corrections can keep being made, and the next period will see them. What locking does is make one reported position permanent and open to inspection – so a restatement is visible as a restatement, rather than a number quietly changing between two conversations.
Confidence expressed, not asserted.
Every figure in Carbonex is designed to carry how good it is and how complete it is alongside its status, and for those to add up. A total should be able to tell you how much of it is recorded, how much is approved estimate, and how much is simply absent.
This is what makes those four distinctions useful rather than decorative. A completeness figure that treats an approved estimate exactly like a measured value is not telling you anything.
Analysis over the same evidence the reports use.
Analysis in Carbonex is being built on the same information the reports draw on, rather than on a separate copy that drifts away from it. An analysis that disagrees with the report is a fault, not a perspective.
Sustainability and transition risk are being connected to that same underlying data, so that a risk assessment can point at the figures it rests on rather than sitting beside them as a separate document.
Who can do what, and how that is decided.
Carbonex is a cloud service shared by many organisations, and each organisation’s information is kept separate from every other’s. What a person is allowed to see and do is decided by the platform itself, not by what the screen happens to offer them.
Administrative changes are recorded in the same history as the data. Granting someone access, changing a setting or assigning a role is as traceable as a submitted figure.