Potential, planned or realised: which avoided-emissions figure are you looking at?

Short answer. Potential is what the solution could avoid if it takes its obtainable market. Planned is what the company's own business plan would avoid. Realised is what the units actually sold have avoided. The first two are forecasts; the third comes from real sales. Carbbin labels every figure with its tier and never adds tiers together.

Why the label matters

The same company can carry three very different numbers, and all three can be correct. The trouble starts when a forecast is read as a result, or when a market model is presented as the company's plan. In an LP report, that is the difference between a number that holds and one that has to be withdrawn.

What the standards say

Project Frame (Guidelines 2024) defines three assessment types, each with its own volumes (Section 2, pp. 10 to 11; Section 3, pp. 26 to 27):

• Potential impact multiplies net unit impact by the obtainable market (SOM) for the technology class, on a standard growth curve. It looks at what the solution could change if it scales over the next 15 to 30 years, and "copycat solutions are also a form of success" (p. 26).
• Planned impact uses the company's own business plan, which typically projects "no more than 5 to 7 years" (p. 27). It only applies to companies that have one.
• Realised impact is "backward-looking, not forward-looking" (p. 27). It uses historical volumes, and unit impact should increasingly rest on real data about how the solution is built, used and retired. Investors should start tracking as soon as actual data exists (p. 27).

Frame's glossary calls potential and planned "two distinct classes of forward-looking assessment". The reason for the labels is stated plainly: classifying assessment types tells readers how reliable each figure is, and "This is essential to avoid greenwashing" (Section 3, p. 26). The stage chart maps seed and Series A to potential, Series B or C to planned or potential, and pilots and commercial companies to planned or realised (p. 28). Investors should also compare business plans and actual sales against their original estimates (p. 28).

For external reporting of realised impact, Frame uses tonnes of CO2e per year, and any cumulative figure is clearly defined and kept apart from forecasts, which "can be misleading if taken out of context" (Section 5, p. 41).

ISO 14067 sets the rules for tracking a footprint over time. Assessments at different points in time use the same functional unit and the same method throughout: data systems, boundaries, allocation and characterisation factors. The interval and its basis are stated in the scope (6.4.7).

The WBCSD Guidance on Avoided Emissions (v2.0) offers two ways to count sold products, year over year or forward-looking over their lifetime, which it treats as theoretically equivalent given the same data (Step 1). The point is to pick one and count each unit once. It keeps the reference technology mix fixed for each year of sale.

How Carbbin does it

Carbbin works in three tiers, and each case carries its tier on the summary dashboard.

1. Potential, a forecast. Carbbin models adoption itself, top-down from the market segments, not from the company's sales plan. Every screening or due-diligence case starts here. On the dashboard: adoption modelled by Carbbin from market segments, not the company's own sales plan.
2. Planned, a forecast. When the volumes come from the company's own plan or committed volumes, the case is labelled planned.
3. Realised, plus a refined forecast. Each year, Carbbin counts the avoided emissions of every unit the company has sold and that is still in use, for the years elapsed only, not the unit's remaining life. It uses the same method as the forecast, and earlier years are not rewritten when later ones are updated. Then the real sales recalibrate the market adoption curve, and a new CPP refines the forecast for the years ahead. The forecast gets sharper each year instead of ageing. Founders only confirm sales.

Because every company and every year runs on the same method, a fund can add up its portfolio, per fund and year by year, realised and forecast, without mixing the two.

A few rules hold across all three:

• Tiers are never added together. Realised results and forecasts sit side by side, each labelled.
• Each unit is counted once. Where a report shows a full-lifetime figure and a figure inside an assessment window, both describe the same units in the same tier, and they are never summed.
• The figure is the solution's. Each case carries the line "this is the potential of the solution, not the share attributable to any investor". Attribution is the fund's decision.

Common mistakes

• "Realised" or "actual" for a seed-stage forecast. Nothing has been sold yet.
• "Planned" for a figure built on someone else's market model. Planned means the company's own plan.
• Adding potential and realised into one total. One is a forecast, the other a result.
• Re-running past years on today's baseline. History changes every year and nobody can reconcile last year's report.
• Counting a unit's lifetime at the sale and again year by year. Pick one convention and count once.
• Never comparing the forecast with what happened. That comparison is how the next forecast earns trust.

Sources

• Project Frame. Project Frame Methodology: Evaluating Greenhouse Gas Impact for Early-Stage Investments. December 2024. https://projectframe.how/s/2024-Project-Frame-Methodology.pdf
• Project Frame. Glossary: potential impact, planned impact, realized impact. https://projectframe.how/glossary
• ISO 14067:2018. Greenhouse gases: Carbon footprint of products. Requirements and guidelines for quantification. Clause 6.4.7.
• WBCSD. Guidance on Avoided Emissions, version 2.0. 2025.