Efficiency, and whether you are running the plan you chose

Two questions sit above everything else in the finance model, and they are different questions. Are we efficient? — how much growth you get for what you spend. Are we on plan? — whether the business you are actually running is the one you said you were going to run.

​

The four numbers a board asks for

Every input here is owned somewhere else, and that is structural rather than incidental. Sales and marketing spend comes from the cost classification, net burn from the cash model, margin from the margin work, recurring revenue from the revenue model. None of it is re-derived here. Efficiency is a read across those figures, not a second opinion about them — which is why the chain in Cost and margin has to hold before any of these means much. Alongside the four, Beacon reports a composite efficiency score with a band, built from five sub-scores, so a board conversation can start at one number and open into the five underneath it rather than jumping between four unrelated ratios.

​

Are you running the plan you chose?

You commit to a growth plan — aggressive, balanced, disciplined, or capital preservation. Each one carries targets. The question this answers is whether the business is behaving like the plan you picked, and it is a more useful question than any single metric, because a company drifts off a plan gradually and in several places at once. Beacon scores six dimensions against the targets your plan carries, and weights them: Each dimension scores 0–100 by how far actual sits from target relative to the tolerance you allowed — so a dimension inside its band scores well and one outside it degrades in proportion, rather than flipping between pass and fail. The six roll into a single growth coherence score: And there is a floor that matters more than the bands. If any single dimension falls below 30, the overall score is capped at 65 however good the other five are. One badly broken dimension cannot hide behind five healthy ones. Runway carries an additional hard penalty: a projected breach of your cash floor caps the capital dimension outright, regardless of how the rest of the business looks. This evaluates; it never recalculates. Not one of the six numbers is re-derived here. If your margin figure is wrong, the fix is in the margin work, not here — and the score will move when it does. A verdict is advisory until it has two quarters behind it. The first reads are informational, and Beacon says so rather than presenting a first-cycle verdict as an established one.

​

The one place Beacon reasons rather than calculates

This is worth understanding whether or not you ever look at a growth score, because it is the clearest statement of a boundary that governs the whole system. Almost everything Beacon produces is calculated — a stated formula over stated inputs, run in the data layer, reproducible by anyone with the same inputs. Beacon’s reasoning layer does not write those figures. It explains them, drafts around them and answers questions about them, and it does not touch them. The growth coherence score is the single deliberate exception: a judgement across six dimensions that resists a formula, so it is reasoned. Because it is the exception, it is fenced harder than anything that is calculated:

• It writes only its own rows. It cannot overwrite a calculated figure, a dimension score, a target, or anything locked. If it disagrees with the margin figure, the margin figure stands.
• It is invalid without a full audit trail. A growth score is only valid if it carries a snapshot of every input it saw, the pinned version of the instructions it ran under, and the clock reference it ran against. Missing any of the three, the row does not count. You can reconstruct exactly what it looked at and what it was asked.
• It recommends and never acts. It changes no parameter, moves no bound, and authorises nothing. See Proposals, not actions.
• It produces no money figure at all. Every value here is a dimensionless score. No currency amount anywhere in Beacon is reasoned into existence.

If you are evaluating Beacon and want one thing to check, check this: every number with a currency symbol on it is calculated from a stated formula, and the reasoning layer cannot write one.

​

What none of this does

• It is not a forecast. A coherence score says where you are against a plan, not what happens next. See One forecast, not two.
• It is not a decision. No parameter, bound or plan is changed on the strength of a score.
• It is not a per-person or per-team read. Everything here is company-level.

​

Related