ESG
Ask once.Answer everyone.
A reporting season is one exercise, not three procurement decisions. The questionnaire that goes out in January is what computes your PAI statement in June and what prints the board report in September — the same answers, never re-keyed, never re-asked.
On this page
A questionnaire builtfor your fund.
Response rates collapse when a company receives a hundred questions and forty do not apply to it. Datasets here are scoped by sector and NACE code, phrased in both languages question by question, and split across the people who actually hold the data.
What it does
Scoped by sector and NACE
A software company is never asked about process emissions or hazardous waste. This is the single biggest driver of whether a questionnaire comes back at all.
Per-question ownership
Finance, HR and operations each answer their own part in the same workspace, so no single person blocks the return.
Bilingual, question by question
French and English side by side, with guidance and units on the questions a respondent could reasonably answer wrongly.
Live completion and red flags
You see who is behind by company, dimension and question — so you chase the three that are late, not all twenty-two.
Frozen when the campaign opens
Every indicator is snapshotted into the campaign at launch. Editing the dataset for next year never rewrites what a company answered last year — which is what makes a year-on-year series mean anything.
GP, fund and company
The three levels each answer their own questionnaire on the same entity tree, so the manager's own disclosures and the portfolio's roll into one consolidation rather than two exercises.
How a campaign runs
- 01 · Week 0
Build the dataset
We start from the standard you report against and add the indicators your LPs and your thesis need. Scoping by NACE code happens here, not later.
- 02 · Week 1
Invite the portfolio
Each company gets its own workspace. Questions split between finance, HR and operations so no single person carries the whole return.
- 03 · Weeks 2–8
Chase what is missing
Completion and red flags update live. You chase the three companies behind, and the advisory team chases alongside you.
- 04 · Close
Export and reuse
The same answers become the consolidation, the PAI inputs, the EDCI submission and the board report. Nothing is re-keyed.
What trips a flag
| The answer | The rule | What you do |
|---|---|---|
| Scope 2 reported as 0 | A company with sites and employees reports no purchased energy | Ask before it reaches the consolidation |
| Headcount down 60% | Moves more than a threshold against last period with no note | Confirm a disposal, or a typo |
| Policy: yes, document: none | An affirmative answer with no evidence attached | Request the file while the campaign is open |
Start from a standard, then tailor
| Layer | What it is | Why |
|---|---|---|
| Harmonised core | France Invest, Invest Europe, CSRD, EDCI or ISO 37001 | Answers stay comparable and reusable across disclosures |
| Sector scoping | Indicators filtered by NACE code | Nobody is asked a question that cannot apply to them |
| Thesis indicators | The three to ten measures you act on at board level | Makes the data useful for something other than compliance |
| Destination tags | SFDR PAI code, EDCI metric, report section | Collected once, consumed everywhere |
Almost every dataset we run is a fund-specific derivation of a master rather than the master itself. Tailoring is the normal case here, not the exception — and the harmonised core underneath is what keeps the answers comparable anyway.
“Our companies will not respond. They never do.”
They will not respond to a hundred questions when forty do not apply — that is a scoping failure, not a motivation problem. Scope by sector and NACE code, split the questions across the people who hold the data, put each one in the respondent's language, and add guidance where a wrong answer is plausible. Then chase with them rather than at them.
The PAI statement,computed not assembled.
The statement is not a separate collection exercise. Every input is already an indicator in the questionnaire, tagged to its PAI code, combined with the financial data entered for the reporting period.
What it does
Annex I, Table 1
All eighteen indicators, scoped to the asset classes you hold, with at least one from Table 2 and one from Table 3 as the regime requires.
Financed emissions
Attribution on enterprise value including cash, named explicitly. Get this wrong and the headline number moves by a factor, not a percentage.
Coverage per indicator
If eleven of twenty-two companies reported Scope 1, the statement says so. Undisclosed coverage is the most common reason a statement comes back.
Prior period alongside
From year two, every indicator carries the previous period on the same basis — and the platform stops a methodology change from silently breaking the comparison.
How the statement gets built
- 01 · Once
Tag the dataset
Each questionnaire indicator is mapped to the Annex I input it feeds. Tags live on the dataset, so wording can differ between companies.
- 02 · Per period
Enter the financials
Invested amount, enterprise value including cash, revenue and headcount for the reporting period. This is what attribution runs on.
- 03 · On demand
Compute and read coverage
Every indicator recomputes as late answers land, with coverage shown per indicator rather than as one portfolio-wide claim.
- 04 · Before 30 June
Explain and publish
Add the explanations and the actions taken, then publish. Next year reopens on the same basis for the comparison.
Where statements get sent back
| Failure | What it looks like |
|---|---|
| Silent coverage | A portfolio-wide figure that actually covers half the portfolio |
| Wrong denominator | Attribution on equity value rather than EVIC |
| Missing comparison | No prior period, or a prior period on a different methodology |
| Generic actions | “Continue to engage with portfolio companies” instead of a named initiative with an owner and a date |
Because tags live on the dataset rather than on question numbers, two companies answering differently worded questionnaires still feed the same PAI input.
“Our consultant already builds this in a spreadsheet.”
The first one usually is fine. The second is the problem: the model has to reproduce last year on the same basis, survive a methodology change, and answer an LP asking where a number came from. A spreadsheet does none of those without the person who built it — which is a dependency, not a process.
A dashboard is for you.A report is for them.
Everything the platform collects resolves into one designed document you can put in front of an investment committee or an LP without touching it — and a dashboard you use while the season is still running.
What it does
Consolidation across the entity tree
Company, fund and GP levels on the same indicator keys, so the numbers add up rather than being reconciled by hand.
Benchmarks
Compare a company against its sector cohort, and a fund against the portfolio, so a figure means something on its own.
The board report
A designed multi-page PDF per fund or per company, in English or French. Not an export, not a dashboard screenshot.
Traceability
Every figure resolves to the company, the question, the respondent and the evidencing document — which is what an LP asks for six weeks later.
The action plan
Answers are scored against a sector materiality profile — which KPIs matter for that industry and what they weigh — against a regulatory floor, a sector average and a best-in-class figure with a named source. What comes out is a plan: each line with a responsible function, a cost estimate, a benefit estimate and the regulation behind it.
How a season closes
- 01 · Close
Freeze the campaign
Answers are timestamped and the period locks, so what the report says is what was known when it was published.
- 02 · Consolidate
Roll up the tree
Company to fund to GP on the same indicator keys, with coverage carried through rather than averaged away.
- 03 · Generate
Produce the four outputs
Board PDF, consolidation workbook, PAI statement and EDCI submission — all from the same answers.
- 04 · Afterwards
Answer the follow-up
When an LP asks where a number came from, it resolves to the company, the question, the respondent and the document.
One campaign, four outputs
| Output | Audience | Format |
|---|---|---|
| Board report | Investment committee, board | Designed PDF, 30–60 pages |
| Consolidation workbook | Your own model, the analyst | XLSX on stable indicator keys |
| SFDR PAI statement | The regulator, published | Annex I Table 1 template |
| EDCI submission | The benchmark | Core metric set, annual cycle |
All four come from the same answers. Nothing is re-keyed between them, which is why they cannot disagree with each other.
“Can we not just export it all to Excel?”
You can, and the workbook is one of the four outputs. But a workbook is not what an investment committee reads, and it carries no coverage, no provenance and no prior period unless someone rebuilds those by hand every year. The export is a format, not a deliverable.
Where this differs
Most platforms sell collection, then sell reporting, then sell the regulatory module.
It is one exercise. An answer given in January is the PAI input in June and a line in the board report in September, because it never leaves the same record — which is also why nobody is asked for it twice.
Most platforms stop at a dashboard and leave the interpretation to you.
Answers are scored against a sector materiality profile and come out as an action plan: each line with a responsible function, a cost estimate, a benefit estimate and the regulation behind it.
Most platforms give every fund the same questionnaire.
Tailoring is the normal case here: a recognised standard as the base, scoped by sector and NACE code, plus the handful of indicators your thesis and your LPs actually act on.
The regulation behind this
Written for someone who has to comply with it, not to rank for its name.
Bring us your next reporting season.
Thirty minutes on your actual portfolio. We will build a specimen dataset from a fund that looks like yours and show you what your companies would see, what the PAI statement would compute, and what lands on the board table.