Sustainability and Climate Disclosure Reporting Assessment for Finance, ESG and Operations TeamsYour disclosure answers everything except the one question the assurer will ask.
Forty exercises for people who prepare, review or sign off sustainability and climate disclosures. Ten print a short extract and six questions a reader might need answered, three it answers and three it does not, and you mark which. The two ways of getting that wrong are counted apart and never averaged, and the report is a heat strip: one row per area, so you can see where your reading is even and where it is spiky.
A report that never averages the two ways you can be wrong
The Sustainability and Climate Disclosure Reporting Assessment for Finance, ESG and Operations Teams is a thirty-minute knowledge and judgement check of what a sustainability or climate disclosure actually states, supports and leaves open, for people who prepare, review or sign off disclosures, reported as a heat strip across five areas with the two ways of misreading it counted apart.
Ten of the forty exercises print a short extract exactly as a reader would meet it, then ask six questions a reader might need answered: three the extract answers on its own and three it leaves open, balanced within every extract. Marking every question and marking none of them both land at exactly three right calls, the same arithmetic, because the only way through is to read what the words actually state rather than take a stance on the format. The other thirty exercises test what a figure supports, what would make a claim checkable against an independent record, and how a disclosure is assembled, reviewed and put forward for outside checking.
Being wrong about an extract runs two different ways, and they need opposite fixes. Reading a gap as answered is the direction whoever checks the disclosure later finds, and it is fixed by adding the missing figure, year or threshold. Reading something the extract already states as open is the direction the person preparing the disclosure pays for, in data re-requested and sentences rewritten that were already there, and it is fixed by deleting the duplicate. A single accuracy number would average these into a figure that tells you to do neither, so the report keeps them apart, counts each, and prints a worked example of both from your own answers.
The report is a heat strip: one row per area — what is in scope, measurement and boundaries, evidence and assurance readiness, targets and claims, and omissions and comparability — and one shaded, labelled cell per exercise, in the order you met it. A mean cannot show whether a reading is even or spiky, and the two look identical on a single number; the strip shows it directly, in words as well as shading, so a report can say your boundary work is even and your targets work is spiky in the same sentence a mean would erase.
Every figure on the page is corrected against a declared prior on every option, pair, position or value — the share of respondents expected to make that call, authored before anybody sat the instrument — so zero means an ordinary reader answering the way people typically answer, not half marks and not a coin toss on a select-all. Every figure carries its 68 and 95 per cent band, drawn and written in words, and the extract you read least accurately is printed beside the mean rather than folded into it. An area with too few answered exercises or too little reliability carries a three-way placement instead of a number, and the refusal is printed where the figure would have been.
No law, filing threshold, regulator or deadline is the key to any exercise, so the same sitting is fair wherever it is taken, and no real company, auditor, rating agency or reporting platform is named anywhere. This is a reading check, not assurance: it does not measure whether any organisation's disclosure complies with anything, and it is not legal, regulatory or accounting advice. It measures whether you can tell what a disclosure states, supports and leaves open — the read an assurer will do anyway, done here before the report goes out.
What you walk away with
Whether a scope statement is read as two directions and not one, and the difference between what a disclosure decides to cover and what it merely mentions.
Control versus equity share, the two accepted views of purchased electricity, and keeping a boundary, a method and a figure apart so a moved number is not read as performance.
What a claim needs behind it to be checked, and matching a claim to the record that exists independently of it, at the level of checking it actually needs.
Base year, absolute against a rate, whether credits count, and reading a target's terms before its headline number.
What a disclosure leaves out, whether it says so, and sizing an exclusion against the total rather than naming it and stopping there.
How often you read a gap as answered and an answer as open, each with its own cost, its own fix, and a worked example from your own calls.
Inside your report
Illustrative sample — your report is generated from your own responses.
| Area | Figure or placement | Texture |
|---|---|---|
| What is in scope | +34 | Even |
| Measurement and boundaries | +40 | Even |
| Evidence and assurance readiness | near typical | Mixed |
| Targets and claims | −18 | Spiky |
| Omissions and comparability | +4 | Mixed |
Dark ● Full is every call right on that exercise; hatched ◐ Part is some; light ○ None is none; dashed □ No answer was not attempted. The texture is the finding: an even strip and a spiky strip can share the same mean.
Cost. The assurer finds this gap first.
Fix. Add the figure, the year or the threshold.
Cost. The preparer re-asks for data already on the page.
Fix. Cite the sentence that already states it and delete the second one.
| Direction | Count | Fixed by |
|---|---|---|
| Read as answered (open left uncovered) | 7 / 15 | Adding |
| Read as open (answered re-requested) | 3 / 15 | Deleting |
The two directions are never averaged into one accuracy figure. One is corrected by adding something to a disclosure, the other by deleting something from it, and a mean of the two would tell you to do neither.
Built for
- Sustainability and ESG reporting teams preparing or reviewing a disclosure before it goes out
- Finance and controllership teams picking up sustainability and climate disclosure work for the first time
- Internal audit and assurance-readiness teams checking what a disclosure would need to survive outside review
- Operations and supply-chain data owners feeding figures into a disclosure, and consultants preparing a first report
Find out which of the five areas reads even, and which one is spiky
40 exercises across six formats · about 30 minutes · two directions of misreading counted apart, the weakest extract printed beside the mean, and a heat strip that shows what a mean cannot.
₹1,199 (incl. GST) · assessment and full report, nothing further to pay
Frequently asked questions
Whether you can tell what a sustainability or climate disclosure actually states, supports and leaves open, across five areas: what is in scope, measurement and boundaries, evidence and assurance readiness, targets and claims, and omissions and comparability. It does not measure whether any organisation's disclosure complies with any law or standard, it is not assurance or legal advice, and it does not test your skill in emissions accounting or your writing.
Because reading a real extract is the actual skill. Each extract offers six questions a reader might need answered, three it answers and three it leaves open, balanced within every extract. Marking every question and marking none both earn exactly three right calls, so the format cannot be beaten by a fixed habit; the only way through is reading what the words actually state.
Because they need opposite fixes. Reading a gap as answered is the direction whoever checks the disclosure later finds, fixed by adding the missing figure or year. Reading a stated fact as open is the direction the preparer pays for, in data re-requested that the draft already carries, fixed by deleting the duplicate. Averaging them into one number would tell you to do neither, so both are counted, printed and never combined.
One horizontal row per area, one shaded and labelled cell per exercise, in the order you met it. A single average cannot show whether a reading is steady or uneven, and the two can share the same mean; the strip shows it directly, so the report can say your boundary reading is even and your targets reading is spiky in one sentence a mean would erase.
No. The concepts are used in plain, generic language — double materiality, boundary setting, value-chain data, base years and restatement, assurance levels, and absolute against intensity targets — and no country's law, filing threshold, regulator or deadline is the key to any exercise, so the same sitting is fair wherever it is taken. No real company, auditor, rating agency or reporting platform is named.
Each one takes a single capability, puts you inside the situations where it is actually tested, and scores your choices against published evidence — with a report designed for that capability alone, not a template. They span hiring, compliance, education, operations and personal skill.
Browse the catalogue →Methodology: Forty original exercises across six formats: ten coverage exercises, each printing a short disclosure extract and six questions a reader might need answered, of which three the extract answers and three it leaves open; ten single-choice judgement exercises on what a disclosure actually supports; six true-or-false claims about disclosure and assurance practice; five match-the-following exercises pairing a claim with the evidence that would make it checkable; five ordering exercises on how a disclosure is assembled or assured; and four numeric estimation exercises on quantities and boundary shares. One response instruction is declared for the whole instrument and it is a KNOWLEDGE instruction: what does this extract answer, what does this figure support, what would make this claim checkable, never what the respondent would do. Construct statement: this measures whether a person can tell what a sustainability or climate disclosure actually states, supports and leaves open, across five areas: what is in scope, measurement and boundaries, evidence and assurance readiness, targets and claims, and omissions and comparability. It does not measure whether any organisation's disclosure complies with any law, standard or framework, it is not assurance, it is not legal, regulatory or accounting advice, and it does not measure the respondent's technical skill in emissions accounting, their writing, or anything about their character. Scoring is two-sided coverage against a declared selection prior. Every coverage exercise divides its six questions into two classes, answered and open, balanced three and three within every extract, so that marking everything and marking nothing land in the same place as arithmetic rather than by a threshold somebody chose. Each option carries a declared selection prior, the share of respondents expected to tick it, authored before anybody sat the instrument. The expected number of correct calls is the sum of the prior over answered options plus one minus the prior over open options, and the score is the number of correct calls minus that expectation, divided by six minus that expectation. A uniform draw is not a null on a select-all, because nobody answers one by tossing a coin per option. The two error directions, treating an open question as answered and treating an answered question as open, are counted and printed separately and never netted, because one is fixed by adding something and the other by deleting something. The weakest extract is printed beside the mean, because a mean across ten hides the one that would have cost a week. The class list on every coverage exercise is checked programmatically against the key across the whole file, because an option marked answered and left out of the key is a silent inversion no per-item validator sees. The judgement, claim, matching, ordering and estimation exercises are corrected against the declared prior on every option, pair, position or value. Every area is scored on all its exercises and carries its standard-error band from an assumed omega that is printed unrounded; an area with fewer than eight answered exercises or an omega below .70 carries a three-way placement and no number, and the refusal is printed. An unanswered exercise leaves the numerator, the denominator and the chance term together; it never scores zero. An empty sitting scores exactly zero on every figure, and a sitting with fewer than six in ten exercises answered is refused a report rather than given a thin one. No percentile appears anywhere, because there is no norm group yet. Every declared prior is an authored assumption stated in the open and will be replaced by observed shares once live data exist. A careless-responding flag count is computed for the operator and never shown to the respondent as a judgement. Frameworks, directives and standards bodies are deliberately not named in any exercise, because the concepts they share are what is measured: double materiality, the three emission scopes, boundary setting by control or by equity share, the grid-based and contract-based views of purchased electricity, value-chain data quality, base years and recalculation, limited against reasonable assurance, evidence sufficiency, absolute against intensity targets, and the separation of reductions from credits. The corporate accounting and reporting standard published by the Greenhouse Gas Protocol, a mark of the World Resources Institute and the World Business Council for Sustainable Development, is the public source for the scope and boundary concepts drawn on; this instrument is not affiliated with, endorsed by or derived from it or any other framework, standards body, regulator, assurance provider or certification body. No country's law, filing threshold, regulator or deadline is the key of any exercise, so the same sitting is fair wherever it is taken. Sources drawn on: Mautz and Sharaf, The Philosophy of Auditing (1961), on the nature and sufficiency of audit evidence; Power, The Audit Society (1997), on what verification can and cannot certify; O'Dwyer and Owen (2005), on assurance statement practice in sustainability reporting; Simnett, Vanstraelen and Chua (2009), on assurance of sustainability reports across countries; Hodge, Subramaniam and Stewart (2009), on how assurance level and provider affect user confidence; Matthews, Hendrickson and Weber (2008), on the importance of carbon footprint estimation boundaries; Kolk, Levy and Pinkse (2008), on the institutionalisation and commensuration of carbon disclosure; Eccles, Krzus, Rogers and Serafeim (2012), on sector-specific materiality; Edgley, Jones and Atkins (2015), on the materiality concept in social and environmental assurance; Bjorn, Lloyd and Matthews (2021), on methods for setting science-based emission targets; Bjorn, Lloyd, Brander and Matthews (2022), on renewable energy certificates and the integrity of corporate targets; Delmas and Burbano (2011), on the drivers of greenwashing; Lyon and Montgomery (2015), on the means and end of greenwash; Marquis, Toffel and Zhou (2016), on scrutiny, norms and selective disclosure; Palmrose, Richardson and Scholz (2004), on the determinants of market reactions to restatements; Hennes, Leone and Miller (2008), on distinguishing errors from irregularities in restatements; Cohen (1960), A coefficient of agreement for nominal scales, for the correction against expected agreement from declared marginals that the coverage scorer uses; Brennan and Prediger (1981), on the dependence of agreement coefficients on marginals; Macmillan and Creelman, Detection Theory: A User's Guide (2005), for keeping the two error directions apart; Gollwitzer and Sheeran (2006), on implementation intentions; and Haladyna, Downing and Rodriguez (2002), for the item-writing rules. All exercises are original works written for this instrument. No real company, auditor, rating agency, reporting platform or regulator is named anywhere in any candidate-facing text.