The plan, the budget, and why you came out somewhere else

A plan is only useful if you can say afterwards why you did not hit it. That needs three things Beacon keeps separate on purpose: one plan that is actually committed to rather than one of several floating around, a bounded set of alternatives so you know what you were exposed to, and an honest split of the difference into what you did and what happened to you.

​

One plan, and two signatures

Beacon holds exactly one committed plan per cycle — the scenario you chose, recorded with the reason you chose it. Everything downstream measures against that, so there is never an argument about which version of the plan a variance is against. A plan is not locked without both a CFO and a CEO signature. That is not a workflow nicety; a record arriving with one signature is rejected outright. The reasoning behind a lock is recorded alongside it, so a year later you can see not just what was committed to but why.

​

Five scenarios — and deliberately not a simulation

Around the committed plan, Beacon builds a fixed set of five: the planning case itself, an upside, a downside, a stress case, and a case where you deliberately change your growth plan. Custom scenarios exist but need both signatures to request. These are not Monte Carlo runs, and that is a deliberate refusal rather than a limitation. Each assumption is varied inside a band derived from how wrong that assumption has actually been in your business historically — narrow, medium or wide, set from your own record rather than a coefficient someone picked. Run the same inputs twice and you get the same five scenarios. A thousand simulated paths would look more sophisticated and would tell a CFO less, because nobody can defend the distribution the paths were drawn from. Each scenario carries a coherence score out of 100 — how internally consistent it is, given that revenue, cash, runway and margin all have to hang together. A scenario cannot become your committed plan below 60. A plan that is not internally coherent is not a plan.

​

Which assumptions actually matter

Alongside the scenarios, Beacon reports which levers move the outcome and by how much, across revenue, cash, runway and margin — and narrows it to the top five. Most planning conversations are about six things that do not matter and one that does; this exists to make the one obvious. The bands those levers move inside are set from your history and are not adjustable by the system to make a scenario more palatable.

​

The budget, and moving money inside it

Beacon holds an annual envelope, the allocation of it by line, and every reallocation made during the year, with a running net movement figure so quarterly shuffling does not disappear into the noise. Over the top sits a budget coherence score with a band — whether the budget as it now stands still agrees with the plan it was built against, with your growth plan, and with the bounds you set. A budget drifts by many small defensible moves; the score is what makes the drift visible before the year-end does.

​

Variance, split honestly

When a period closes, Beacon reports the difference against plan in absolute and percentage terms — and then does the two things that make variance useful rather than merely accurate. It separates what you did from what the currency did. If you plan in one currency and bill in several, part of any variance is exchange rates and nothing to do with performance. Beacon splits the two: the part at your fixed planning rate, and the part caused by rates moving. Without that split, a good quarter in a bad currency looks like a bad quarter. See Currency. It attributes the remainder to a cause. The operational difference is assigned across four cause categories with weights rather than to a single headline reason, because it is rarely one thing. Each attribution carries a status, so an unexplained variance stays visibly unexplained rather than being assigned to the nearest plausible cause. And the attribution itself carries a confidence score with a band, plus its own movement quarter on quarter. A confident number with an unreliable explanation attached is worse than no explanation.

​

The bounds you set, and what happens at the edge

You can declare bounds — across six dimensions including a cash floor and burn ceilings. Beacon reports where you are against each as a distance to the bound and a severity, so the answer is never a binary that only tells you once it is too late. Three things about how it behaves there:

• Nothing is blocked. These are your bounds and Beacon reports against them; it does not veto your spending.
• Beacon will not move a bound. Changing one is your decision, recorded. The system never adjusts its own limits, and no automated action overrides a pause that is in force.
• Not enough data is its own state. Beacon distinguishes within bounds from cannot yet tell, rather than reporting an untested bound as satisfied.

​

What none of this does

• No second forecast. Scenarios sit on top of the one committed forecast; none of them quietly becomes a rival version of the truth. See One forecast, not two.
• No autonomous parameter change. Bands, weights, thresholds and bounds are proposed with the evidence behind them and changed only by a person.
• No plan approved on Beacon’s say-so. Beacon assembles, scores and explains. Committing is a signed human act, twice over.

​

Related