Get the pack

EcoVadis and SMETA: what evidence you are actually asked for

Both frameworks reason the same way: a policy that commits, an action that implements it, a result that proves it happened. Once you see that structure, the document list stops being arbitrary.

Last reviewed 2026-09-04 · EvidenceReady, Harmony Future Holdings Limited, Dublin

Read this first — what this is and is not

This is our compilation of commonly requested evidence, assembled from the two frameworks’ publicly published methodology overviews and from the ETI Base Code that SMETA audits against.

It is not an official checklist. It is not endorsed by EcoVadis or Sedex, and neither organisation has reviewed it. The trademarks belong to their owners. We do not conduct audits, award scores, or predict scores — your rating is decided by the rating agency alone.

The structure both frameworks share

EcoVadis organises evidence into four themes and grades each on whether a commitment exists (policy), whether it is implemented (action) and whether the outcome is measured (result). SMETA audits four pillars against the ETI Base Code and local law, and asks for much of the same documentation on site.

That overlap is the useful part: assembled once, the same evidence set serves both.

The four themesWho usually holds the evidence
EnvironmentOperations or facilities lead — energy, water, waste, site permits and the environmental management system
Labor & Human RightsHR lead for policies, contracts, payroll and training records; health-and-safety lead for risk assessments and accident data
EthicsCompany secretary, finance or legal — codes, registers, whistleblowing and data protection
Sustainable ProcurementProcurement or supply-chain lead — supplier code, risk mapping and audit programme

What each of the three layers has to look like

Policy

A policy earns its mark by being written, signed, dated and scoped. Undated policies, and policies that live only on a website, are routinely treated as absent.

Action

An action must show implementation: a procedure plus evidence it was followed. A procedure with no record of use is a policy in disguise, and it is scored as one.

Result

A result must be a number, for a stated period, with its denominator and boundary. Two consecutive years beats one, because it shows a trend rather than a snapshot.

This is the single most useful thing on this page

Most suppliers have far more policy than action, and almost no result. The file feels substantial and scores thin. If you do nothing else before your assessment, go through what you hold and sort it into these three piles — the shape of the gap becomes obvious immediately.

The eight ways evidence files fail

  1. Undated policies. A policy with no date cannot be shown to have existed before the assessment period.
  2. Certificates submitted without their scope page. The scope page is the half that says which sites and activities are covered.
  3. KPIs for a single year. One number is a snapshot; raters look for a trend and discount a lone figure.
  4. Targets with no baseline year, so no one can tell what the target is measured against.
  5. Evidence in the wrong language with no translation of at least the headline and the dates.
  6. Documents dated after the assessment window opened, which read as written for the assessment rather than for the business.
  7. A grievance mechanism documented but with no evidence that workers were ever told it exists.
  8. Payroll and working-time records that cannot be reconciled to each other. The discrepancy is the finding, not either document.

The slow part is not writing. It is waiting.

The slowest part of an evidence file is rarely writing a policy. It is waiting on four other people to send you documents they already have. Start those requests before you start writing anything.

The line that makes requests come back faster

Put this in the email: “If a document exists but is out of date or unsigned, send it anyway and say so — a dated draft is more useful to me than a gap.” It removes the reason people sit on a request for a week, which is that they are quietly hoping to fix the document first.

For each document, ask for three things: the document itself, its date, and where it is normally kept. The third one is what stops you doing this from scratch next year.

If a SMETA auditor is coming on site

Documents are half of it. The other half is interviews, and these are the topics that come up. If nobody in your building can answer one of them, that is a gap no document closes:

On scores, and what we refuse to sell you

We do not predict scores, and nobody honestly can

EcoVadis’s actual scoring weights are proprietary and unpublished. We do not have them, do not claim to, and do not model them.

Our pack ranks slots by our own prioritisation heuristic — how much difference a slot tends to make relative to the effort of producing it. That is a work order, not a score forecast. Any tool that offers you a predicted rating is selling a guess as a fact.

The whole evidence file, assembled from your answers

The free check reads twelve questions. The pack maps all 55 slots to your business, ranks the gaps by the work they save, names who in your company owns each request, and gives you the emails to send them.

Get the pack — €490