The current state
Open the latest annual report of any large listed industrial. Find the Scope 3 inventory. Read carefully. What you will see, in the great majority of cases, is a single tonnage figure for each Scope 3 category, computed by multiplying procurement spend or production volume against an emission factor drawn from a public database, and lagged by twelve to eighteen months relative to the operating reality. There will be no uncertainty interval. There will be no statement about the marginal sensitivity to scope-boundary choices. There will, in the better disclosures, be a methodology note that explains the calculation; in the worst, there will not even be that.
This is a perfectly defensible compliance exercise. It is also, considered as an instrument of capital allocation, close to useless. Sustainability-linked bonds whose coupons depend on Scope 3 trajectories are pricing off a number that may be a year and a half out of date. ESG funds executing engagement programmes are arguing about a tonnage whose true uncertainty interval is often wider than the reported number itself. Internal sustainability teams are setting decarbonisation targets against a baseline that nobody internally believes.
The problem is not the framework. The GHG Protocol, with its fifteen categories and its scope-boundary conventions, is a reasonable framework. The problem is that the reporting practice has been allowed to optimise for compliance rather than truth.
Activity-based reasoning
Where activity data exists, use it. This sounds tautological; in practice it is not. There is a vast amount of activity data available about every listed company's supply chain that does not currently make it into the Scope 3 inventory. Shipment-level customs data is one source: tonnage, origin, destination, mode and time, against named consignors and consignees. Sub-national grid intensity is another: the emissions intensity of grid power in the specific city in which a fabrication plant operates is in many cases public and updated quarterly. Verified supplier disclosures — those that come with assurance — are a third. Machine-level energy mixes for heavy industry are a fourth, increasingly available through industrial-IoT integrations and regulatory disclosures.
Horizon Scope 3 stitches these sources together for every listed issuer in our coverage universe. Where the activity data is robust — typically Categories 2, 3, 4 and 6 of the GHG Protocol — the resulting estimate ships with narrow intervals. Where the activity data is partial, we use it where it exists and fall back to inference with widened intervals. Where it is absent, we infer from sector peer distributions with explicitly wide intervals and a documented prior.
The single most important property of this approach is honesty about what we do not know. For thinly disclosed sectors and remote tiers of the supply chain, our intervals are wide — sometimes spanning a factor of two. That wideness is information. It tells the investor exactly where to push the issuer for better disclosure, and exactly where the published single-number estimate from the issuer was overconfident.
Where we have shipment-level customs data, machine-level energy mixes, and verified supplier disclosures, we use them. Where we don't, we infer with calibrated uncertainty — and we are upfront about the fact that the wideness is the message.
Continuous, not annual
Annual Scope 3 reporting is a relic of the audit cycle. There is no operational reason why Scope 3 must refresh annually; the underlying activity data is generated continuously. Horizon Scope 3 refreshes monthly. A company that quietly retools its product mix in Q2 sees its Category 11 estimate shift in the Q2 cycle. A company whose largest supplier brings a clean-energy PPA online in March sees the upstream emissions intensity adjust in the next refresh.
The implication for stewardship is significant. Engagement programmes can hold issuers accountable to evolving facts in something close to real time, rather than waiting for the next annual report. Sustainability-linked instruments tied to honestly-updated Scope 3 stop being a soft target — where the issuer can quietly redraw scope boundaries between disclosures — and start carrying real consequence.
What this changes for investors
Three things change concretely. First, sustainability-linked bonds become useful instruments. The KPI is observable, verifiable, and updated monthly; the step-up clauses fire on real evidence rather than self-reported numbers. Second, ESG equity allocation becomes defensible against the now-common challenge that "the underlying numbers are not investment-grade." With calibrated intervals, dated methodology versioning, and documented assurance, the numbers are investment-grade. Third, divestment and exclusion decisions are easier to justify — and harder to attack — when they rest on continuously-updated, uncertainty-aware estimates rather than two-year-old spreadsheets.
The limits
Honest Scope 3 has limits and we are upfront about them. We cannot estimate Category 11 (use of sold products) precisely for issuers whose product mix is poorly disclosed. We cannot tightly bound Category 1 emissions for issuers whose Tier 2 supply chain is opaque. We cannot resolve the Cat 7 (employee commuting) inventory below a coarse demographic prior. For each of these we publish the interval, document the limit, and tell the user where additional disclosure from the issuer would meaningfully narrow it. The wideness of the interval is, again, the message.