Framework 03 · Carbon

Horizon Scope 3.

Continuous, activity-based Scope 3 inference for listed companies. Normalised across reporting taxonomies, refreshed monthly, shipped with calibrated uncertainty intervals. Built for asset managers and sustainability-linked instruments that need numbers tough enough for audit.

Request a demo Read the method →

The state of disclosure

Most published Scope 3 looks like data. It is mostly narrative.

Listed-company Scope 3 disclosures today are typically twelve to eighteen months lagged, locked to spend-based emission factors from the previous reporting year, and silent on uncertainty. That is fine for compliance. It is useless for capital allocation, and increasingly dangerous for sustainability-linked instruments where compensation depends on the number being honest.

Lag

12–18 months behind

Disclosed Scope 3 reflects the operating reality of the year before last. For companies retooling product mix or shifting suppliers, the published number is a historical artefact.

Spend-based

Stale emission factors

Most disclosures multiply procurement spend by a year-old emission factor. Suppliers who have meaningfully decarbonised mid-cycle do not appear.

No uncertainty

Single-number outputs

Investor decks publish a single Scope 3 tonnage. The actual confidence interval is often wider than the reported number itself.

Activity-based reasoning

Where we have data. Where we don't.

Horizon Scope 3 uses activity-based reasoning wherever it can be sourced. Shipment-level customs records, machine-level energy mixes, verified supplier disclosures, satellite-derived production proxies, sub-national grid intensity at the locations the company actually operates from. Where the activity data is robust, the interval is narrow.

Where it isn't, we fall back to inference, with explicitly wider intervals and a documented prior. We are upfront about what we don't know. For thinly disclosed sectors, our intervals are wide — and the wideness is itself information for the investor.

What this delivers

Monthly refresh, not annual. A company that retools its product mix sees Scope 3 shift in the cycle it actually shifted.
Calibrated 90% intervals on every estimate, with the underlying activity data inventoried.
Methodology versioning. When we update emission factors, we re-run the back-series and publish the diff.
Counterfactual sensitivity. "What would the estimate be if scope-boundary X were drawn at Y?" — answered for every output.
Sector benchmarks. Every issuer scored against a same-sector peer distribution.

What it changes

For investors. For corporates. For the SLB market.

For ESG investors

Sustainability-linked bonds tied to honest, continuously updated Scope 3 stop being a soft target and start carrying real consequence. Engagement programmes can finally hold issuers to evolving facts rather than self-reported aspirations.

For corporate sustainability teams

Internal Scope 3 reporting moves from a year-end accountancy exercise to a continuous management tool. The same dashboard the IR team uses externally is the one the supply-chain team uses to track decarbonisation.

For audit and assurance

Methodology is versioned, dated, archived. Every published number is reproducible from the documented inputs. External assurance is straightforward; we publish the assurance evidence pack alongside the number.

Sample output · listed issuer (anonymised)

Scope 3 with calibrated intervals.

Category (GHG Protocol)	Latest estimate · MtCO₂e	90% interval	Method weight	Refresh
Cat 1 — Purchased goods & services	4.82	[4.31, 5.40]	Activity 72% / Spend 28%	Monthly
Cat 2 — Capital goods	0.64	[0.51, 0.81]	Activity 88%	Monthly
Cat 3 — Fuel & energy	0.27	[0.24, 0.31]	Activity 95%	Monthly
Cat 4 — Upstream transport	0.91	[0.78, 1.07]	Activity 81%	Monthly
Cat 11 — Use of sold products	12.40	[10.20, 14.95]	Activity 64% / Inference 36%	Monthly
Cat 12 — End-of-life	0.83	[0.61, 1.11]	Inference 100%	Quarterly
Total disclosed	19.87	[16.65, 23.65]	Activity 71% wt.	—

Illustrative output for a hypothetical listed industrial. "Method weight" indicates the proportion of the estimate derived from activity-based vs. spend-based or inference methods.

Request a demo