The CII Rating Is Decided in March, Not December

There is a particular conversation that recurs on fleet calls in November. Someone opens the CII dashboard, the annual rating has already drifted past the point where any remaining voyage can move it, and everyone in the room knew it was slipping for months. What nobody had was a number they trusted enough to act on back in March, when there were still voyages left to change.

The argument here is simple and uncomfortable for anyone who has bought a reporting tool expecting it to solve this. A CII rating is not decided by the regulation. It is decided by when in the calendar year the operator starts steering against a forecast they can defend. Start in March and the levers are real. Start in October and you are reading a result, not influencing one.

This is not a software problem. It is a forecasting-practice problem, and it is fixable without buying anything new.

What a forecast has to do before it is worth anything

• •Tell you where the vessel lands by December if nothing changes. That is the baseline you steer against, and most fleets do not have one they believe.
• •Tell you which voyages are pulling the number down, and by how much. Not "ballast legs are expensive" but "this specific ballast leg cost you 0.4 grams of AER."
• •Tell you what each operational lever is worth in rating terms. A 0.5-knot reduction on ballast legs for the rest of the year is worth this many AER points. Hull cleaning in August is worth that many. Without those figures, the technical manager is guessing at which sacrifice buys the rating back.

Most fleets have none of this. What they have is a rolling consumption figure, a regulatory reference line that tightens under them, and the hope that the two do not cross before the year closes.

Why most forecasts cannot be trusted

The problem is rarely missing data. It is contaminated data. Noon reports that round fuel consumption. ROB entries that do not reconcile across voyages. Weather adjustments applied one way on one sister vessel and another way on the next. Each error is small. Together they widen the forecast confidence interval until it is too broad to support any decision, which is the same as having no forecast at all.

The fix is unglamorous, and that is the catch

Here is the tension nobody selling a dashboard wants to dwell on. The work that makes a forecast trustworthy is the dull, repetitive discipline that competes with everything else on a superintendent's desk. It is not a feature you switch on. It is a monthly habit someone has to own.

• •Validate at the point of entry. If the reporting tool does not flag an impossible fuel-curve deviation the moment the crew submits, that error compounds forward for ten months before anyone notices.
• •Reconcile against bunker records monthly, not annually. One quarterly reconciliation is worth more than any post-hoc correction exercise run in panic at year-end.
• •Model against the voyage pattern committed in the charter book, not a theoretical annual profile. If Q3 is heavily ballast-weighted, the forecast has to show that drag before it arrives, not after.

A forecast is not a spreadsheet. It is a weekly conversation between operations and technical, where someone says "we are drifting half a point on AER this month, here are the three voyages driving it, here is what we would have to change." That sentence is the entire deliverable. The number on the screen only matters if someone is in the room to act on it.

The operators running CII well in 2026 are not running better methodology. The regulation did not change. Their operational cadence did. They stopped treating the rating as a year-end reporting exercise and started treating it as a quarterly steering mechanism. If your Q4 conversation is going to read the same as last year's, something upstream has to move in Q2, and it has to be someone's job to move it.