.svg)
From DQ 5 to DQ 2: what actually moves a PCAF data quality score?
A weighted PCAF data quality score of 4 and a bit is not a portfolio of secretive companies. It is usually a portfolio whose disclosures have not been attached to it. Most of the counterparties carrying your exposure have published something, and the score reflects whether anybody went and found it.
DitchCarbon provides verified emissions data for over 2 million organisations, so procurement, sustainability and finance teams can measure and act on supply chain and portfolio emissions from one source. This page sets out what each score on the PCAF scale actually requires, which move is cheap, which is expensive, and where the effort stops paying.
What do the PCAF data quality scores mean?
The PCAF data quality score, which most finance teams shorten to DQ, runs from 1 for the best available data to 5 for the weakest. The definitions below are from the Global GHG Accounting and Reporting Standard, Part A: Financed Emissions, Third Edition, December 2025, for listed equity and corporate bonds. Business loans and unlisted equity use the same structure with total company equity plus debt in place of EVIC.
| Score | What the calculation uses | What the counterparty itself had to provide |
|---|---|---|
| DQ 1 | Reported emissions of the company, verified | A published figure, plus verification. The assurance evidence has to exist and has to be found and attached |
| DQ 2 | Either the company's own unverified reported emissions, or, where it has reported none, primary physical activity data on its energy consumption | A published figure, or primary energy data. Two separate routes reach the same score, and the activity route covers Scope 1 and 2 only |
| DQ 3 | Primary physical activity data on the company's production. PCAF's wording is explicit that "Reported company emissions are not known" | Production data, and no reported emissions. Not a rung between 4 and 2, for the reason in the next section |
| DQ 4 | The company's revenue, multiplied by a sector emission factor per unit of revenue | Nothing but its revenue, which comes from filings or a financial data feed rather than from the company |
| DQ 5 | The outstanding amount, multiplied by a sector factor per unit of asset, or per unit of revenue with a sector asset turnover ratio | Nothing at all. Your exposure and a sector classification are the whole input |
Condensed from PCAF Part A, Third Edition, Table 5.1-2, December 2025.
One thing to read carefully in that table, because it is commonly restated as a strict hierarchy. PCAF says that "Options 1 and 2 are preferred over Option 3 from a data quality perspective", which separates reported and primary activity data from economic estimates. It does not rank reported emissions above primary activity data for corporate exposures. The real estate asset classes do state an order of preference; the corporate ones do not. DitchCarbon records which of these produced every figure it holds, so the mix in your own book is a view rather than an estimate.
Why is DQ 3 not a step between DQ 4 and DQ 2?
Because DQ 3 requires something DQ 2 does not need and rules out something DQ 2 relies on. PCAF's definition of score 3 states that reported company emissions are not known, and that emissions are calculated from primary physical activity data on the company's production.
So the route from a modelled figure to a reported one does not pass through it. A counterparty that has published an inventory goes from DQ 4 to DQ 2 directly, and the work is finding the disclosure and matching it to the right legal entity. A counterparty that has published nothing can only reach DQ 3 by supplying production data, which means a data request, a response, and a calculation you build yourself. That is the expensive path, and it lands one rung lower than the cheap one.
Treating the scale as a staircase leads teams to plan a climb they never needed. The sequence that works is to take every holding that has already published straight to DQ 2 first, then decide what to do about the ones that have not. DitchCarbon is built around that order, which is why the disclosed part of a book is scored on the day you start.
What actually moves a holding from DQ 4 to DQ 2?
A published inventory, correctly attached to the counterparty you hold. Nothing else, and nothing from the counterparty.
The size of the opportunity depends on the universe you are looking at, which is why disclosure statistics are so often quoted misleadingly. Of the S&P 500, more than 88% disclose Scope 1 and Scope 2, per the UCLA Institute of the Environment and Sustainability in June 2025. Across the whole MSCI ACWI Investable Market Index, which reaches far down into small caps, the rate has been well under half. Over 22,100 businesses disclosed through CDP in 2025. A corporate loan book weighted by outstanding amount sits closer to the first of those numbers than the second, because exposure concentrates in larger companies, and larger companies disclose.
The work is therefore matching rather than collection, and matching is where a score quietly fails. A DQ score describes how close the data sits to the counterparty, so it only holds if the counterparty was identified correctly. DitchCarbon resolves each row against DUNS, LEI and ISIN identifiers, with domain, email, region and industry as supporting signals, and resolves parent and subsidiary relationships, so a figure attaches to the entity you lent to rather than the group above it.
Behind that sits what DitchCarbon holds: company emissions data for over 2 million organisations, including corporate-level GHG inventories; supplier-specific emissions at spend, activity and product level; and a generic emission factor library drawn from ecoinvent, CEDA, EPA, DEFRA and EXIOBASE. Every figure carries its source document and its change history, and a recent deployment reached about 60% of a large supplier base within 2 weeks.
What does it take to reach DQ 1?
Verification evidence for a figure that has already been reported. The number does not change. What changes is that somebody independent has checked it and you can show that they did.
This is the only improvement PCAF describes that requires no new data from the counterparty, and it is worth knowing for that reason alone. It is also scarcer than disclosure itself: S&P Global Sustainable1 reported in March 2024 that of 2,590 US companies analysed, 47% disclosed Scope 1 and 45% disclosed Scope 2, while about 18% had those emissions verified externally. So DQ 1 is bounded by what exists in the market, and the ceiling on a book is lower than the disclosure rate suggests.
The practical constraint is that assurance status is a property of a document rather than a number, and it varies by scope within the same report. A company whose Scope 1 and 2 were assured and whose Scope 3 was not is the common case, and flattening that into one assured label misrepresents it. DitchCarbon lists every source document by year with three assurance columns, one per scope, so the evidence for a DQ 1 claim is a document you can open rather than a flag you have to trust.
How is the weighted score calculated, and what has to be disclosed?
PCAF requires the score to be normalised to exposure. The weighted average is the sum of each holding's outstanding amount multiplied by its data quality score, divided by the total outstanding amount, calculated per asset class and per sector.
Two disclosure expectations follow, and they carry different force. Financial institutions "should publish a weighted score by outstanding amount of the data quality of reported emissions data or should explain why they are unable to do so", which is comply or explain. Where Scope 3 emissions are reported, "the weighted data quality score of these emissions shall be reported separately from that of scopes 1 and 2", which is not.
That second one is the harder number, because Scope 3 disclosure is thinner than it looks. MSCI examined all 8,197 constituents of the ACWI Investable Market Index in July 2025 and found that only 13% had reported on their most material categories. Reporting some Scope 3 and reporting the categories that matter are different things, and a separate weighted score makes the difference visible.
Weighting by exposure also decides where effort pays. A holding with a large outstanding amount and a DQ of 4 moves the average; fifty small ones at DQ 5 barely register. DitchCarbon reports the method mix as a proportion of the total rather than as a claim, so the arithmetic of which holdings to work on is done before anyone opens a spreadsheet.
Where does the score cap out however much you spend?
Some line items cannot reach the top of the scale at all, and knowing which ones protects a budget. PCAF's motor vehicle loans method, for example, states that for a mixed line item "the highest possible data quality score this line item could receive is score 4". The asset class caps it, not your effort.
The same logic applies more loosely across a book. Commercial property and agriculture consistently score worse than heavy industry in published bank disclosures, and that is a function of what data exists about buildings and farms rather than of how hard anyone tried. PCAF and CDP put it plainly in their joint paper on data quality in June 2023: "perfect" is often sought at the expense of "good", and a low score "represents the necessary starting point".
So the useful question is not how to lift every score. It is which holdings can move, by how much, and at what cost. DitchCarbon answers the first two before the third is committed, because the disclosed share of your book is countable from what we already hold.
What are banks actually reporting today?
Most of them are reporting the score. KPMG's benchmark of sustainability disclosures from 33 major banks, published in June 2025 and covering the 2024 reporting cycle, found that 82% use the PCAF framework to report data quality for financed emissions, up from 75% the year before. Most disclose scores by sector, and some split them by scope.
The finding worth acting on is a different one. KPMG reports that some banks do not disclose financed emissions targets at all where data quality scores are higher than 4, because of the risk of restatement. At that point the score has stopped being a reporting metric and become a blocker on the transition plan: the target cannot be published, so the sector cannot be steered, so the work does not start.
KPMG also records what makes this hard, and it matches what the ECB observed across supervised institutions in May 2026: proxies are used as an interim solution where client data is not available, and data projects get prioritised rather than run in parallel. Prioritising means knowing which holdings would move. DitchCarbon produces that list from the disclosures it already holds against your own counterparties, rather than from a survey campaign that has to run first.
Where should the effort go first?
Four moves, and they do not cost the same. Only one of them requires nothing from anybody, which is why it belongs before the others rather than after them.
| The move | What it takes from you | What it takes from the counterparty | What it is worth |
|---|---|---|---|
| DQ 5 to DQ 4 | The counterparty's revenue, from filings or a financial feed | Nothing | One notch, and no new information. Both scores are a sector average with your exposure attached, so the estimate is no closer to the company than it was |
| DQ 4 to DQ 2 | Nothing beyond the holdings list. The disclosure already exists and has to be found and matched | Nothing, where it has already published | Two notches, on the largest share of a typical book. The figure becomes the counterparty's own, so it moves when they reduce. This is the move most portfolios have not made |
| DQ 2 to DQ 1 | Finding the assurance evidence and recording it per scope | Nothing, where the report was assured | One notch, bounded by the market. Verification is scarcer than disclosure, so this lifts part of the book and not all of it |
| Nothing reported, to DQ 3 | A data request, a response, and a calculation you build from primary production data | Real work, from a team that has not published before | One rung below the cheap route, and the right move only for holdings that matter and have disclosed nothing |
Assessed August 2026 against PCAF Part A, Third Edition and the DitchCarbon platform.
The second row is the one that changes a weighted score, and it is the one most programmes skip, because a data quality problem sounds like a data collection problem. It is a matching problem. DitchCarbon supplies the disclosed part of your book already matched and already scored, with the source document behind every figure, and the calculator that produced those figures is verified to ISO 14064-3, limited assurance, by UL Solutions, renewed annually. DitchCarbon data has been used in emissions reports subsequently assured by ten different third-party assurance providers, including Big Four firms.
Where nothing has been published, the fourth row is a real option and DitchCarbon supports it: requests go out prepopulated with what that organisation has already disclosed elsewhere, aimed by a four-step maturity ladder sorted against embodied emissions, so the campaign runs on the holdings where an answer would move the weighted average rather than across the whole book.
What if you are the company being scored?
If your organisation has not published emissions, your lenders and investors are not leaving you out of their financed emissions total. They are estimating you from your revenue and your sector, and that figure reaches their reporting and their sector targets whether or not you have seen it.
Your DitchCarbon profile already exists and already carries whatever you have published, listed by year with the assurance provider and standard found for each scope, and the document each figure came from. A disclosure to-do list shows year by year what is held on you, marked complete or partial, and names the categories investors expect and cannot find. Upload an assurance statement or an updated report and we extract from it rather than asking you to retype figures you have already published, and where extraction has read something wrongly there is a route to say so. It is free, and there is no fixed form to complete. Claim your profile.
How do you find out where your portfolio stands?
Three questions, answerable against the book you already hold.
What share of your outstanding amount sits on a figure the counterparty published itself, rather than on its revenue times a sector factor? For the holdings that are modelled, how many of those companies have actually published something that nobody has attached? And what would your weighted score be if they had?
The second question is the one that usually surprises people, because the answer is rarely small, and because closing that gap asks nothing of a single counterparty.
A coverage check on a representative sample of your holdings answers all three: how many of those carry a figure the organisation published itself, the source document behind each, and what is left resting on a sector average. Send a sample to the coverage check. DitchCarbon gives you numbers you can defend within 2 weeks.
Last reviewed August 2026.
Our calculator is verified to ISO 14064-3, limited assurance, by UL Solutions, renewed annually.
