This page describes the module as it runs today. The readings behind each of the six parts — read from the modules that already run for you — and what is holding the score back are live. The score, the score for each part and the level it sets — what Beacon may do by itself — are designed and not calculated yet, and so are the alerts and the confidence statement for the board. How each reading becomes a part score is not yet written down. Until there is a score, a person confirms every action on a forecast. The rules the live page follows are set out below.
What it does
A forecast that looks precise can rest on stale data, a pipeline behaving oddly, renewals going wrong or billing that does not add up. Forecast confidence is built to check all of that every morning, give one score from 0 to 100, and use the score to decide how much Beacon may do by itself — act, draft for a person, only suggest, or only watch. The score travels with every forecast that goes to the board. It says how far to trust a forecast; it never changes one. Today it does the part that needs no score. Each time its page opens, it:- Reads what the other modules have already worked out for each of the six parts — Data integrity’s morning checks, Pipeline velocity’s latest score, the open deals Deal likelihood read, Customer fit’s completeness, Usage intelligence’s usage, Churn risk’s subscriptions and renewal dates, Customer lifecycle, Expansion readiness, your revenue from billing and the signals noticed in the last 30 days.
- Shows each part’s readings and what it is still missing.
- Says what is holding the score back — the steps you can take now, each with a button to where it is done, and the steps that are Beacon’s.
- Saves your setup answers.
What it produces
Its headline is two figures. Neither is calculated yet.
Both are described in the Forecast confidence article. Neither is served through the Data API yet.
Behind them it also produces:
How it is worked out
What the page shows today
The page works nothing out. It shows figures other modules have already written, and the same data always gives the same page.- Read when the page opens. Each reading is the latest one the module behind it has stored. The page says when the readings were taken.
- Nothing is added, divided or rounded. The only changes are the ones every Beacon page makes: Customer fit’s completeness is shown as a whole percent, and money is shortened, such as $2.83M. Enterprise revenue is shown as two figures — Enterprise and the total — not as a share, because a share would be a division no module has made.
- A part with readings reads Readings in · not scored. A part whose module has not run yet reads Waits for its module. Intent strength reads Partial without a marketing tool, and financial integrity reads Counts 50 until accounting until your accounting tool is connected.
- Data quality uses Data integrity’s own readings, so the two pages never disagree.
- What’s holding the score back is a fixed list. Each line appears only while the fact behind it is still missing, and disappears when it is in. Yours: tell Beacon where your renewal dates are (Churn risk’s setup), fix the open data problems that need you (Data integrity), answer one setup question, and connect your accounting tool. Beacon’s: work out the data trust figure, finish how each reading is scored, read your renewal dates in, and finish the Churn risk, Deal health and Deal likelihood scores. The page never says how much a step would lift the score, because no such figure exists yet.
- The page reads Not measured yet until your CRM and billing are connected, Coming if its readings cannot be had, and Running with gaps while it has readings and no score. It reads Live only once a score has been worked out.
How the score will be worked out — designed, not calculated
The score is the one figure in Beacon that is reasoned rather than calculated: Beacon’s reasoning layer combines the six parts, and the parts come from the modules’ own figures. Every score is kept with its inputs, so it can be checked later.
From readings to a score. Each reading becomes a score from 0 to 100 and the readings in a part are combined. Each part is multiplied by its weight and added up. A part whose modules are not on yet counts a neutral 50, so the score is not dragged down before everything is connected — and the board statement says which parts were counted that way.
What the score allows.
The level rises at most one step every 4 hours, so after a fall Beacon waits before acting by itself again.
Bands. High 80 to 100 · Medium-High 70 to 79 · Medium 60 to 69 · Medium-Low 50 to 59 · Low 40 to 49 · Critical under 40.
How high it can go. While only the Shared foundation runs, the highest possible score is 62.5. With Sales it is 75; with Customer and marketing modules, 95; with Finance, 100.
Warning signs. Pipeline more than 25% slower than usual · stages too bunched · deal chances more than 12% from usual · over 20% of renewing revenue at high risk · usage down more than 30% in your top quarter of customers · unusual revenue moves over 3% of the quarter’s run-rate.
Fixed settings. The levels at 80, 60 and 40 change only with the account owner’s approval, and Beacon never adjusts them. The six weights are the CFO’s: a change of 3 points or less may go through, at most once every 60 days. Pipeline stability weighs ×1.1 for Enterprise when Enterprise is over 40% of revenue; renewal stability weighs ×1.1 when renewals are over a quarter of yearly revenue. Beacon never changes its own thresholds.
When it runs. Every morning at 07:00 UTC, after the night’s runs of the modules it reads, and again whenever a part moves 3 points. Every run is kept for good and never edited. It holds no money value; it scores trust.
None of this is calculated today. How each reading turns into a part score is not yet written down, Data integrity’s trust figure is not worked out, and the Churn risk, Deal health and Deal likelihood scores do not exist yet. Until then the page shows the real readings and says Not yet calculated for the score.
What you see on its page
In Beacon, open The Brain and choose Forecast confidence, under What it projects. The banner reads Not yet for the score, how many of the six parts are scored (none yet) and which have readings, and that a person confirms every action on a forecast until there is a score. The page has six tabs:- Overview — the six parts, each with its weight, the forecast it protects, today’s readings and what is missing; what’s holding the score back; the last 90 days, waiting for the first score; what it means for each forecast; who it is for; its data sources; what it reads from and feeds; and where it shows up — Beacon, Slack, Reports & cadences and Claude.
- Dashboard — the module’s five views: the score now and over 90 days, the six parts, what Beacon may do by itself for each group of forecasts and alerts (today, a person confirms everything), the score week by week, and the confidence statement for the board. The parts table shows today’s state; the rest say what they wait for.
- Signals & alerts — what the module watches and what it has seen so far, and its seven alert rules with what each waits for.
- Scorecard — the six parts and their weights, how readings become a score, what each level allows, the bands, the goals, how high the score can go at each stage, the warning signs, the fixed settings and who changes them, and when it runs.
- Inputs — every reading behind the six parts, the module it comes from and what it reads today: Read, Partly, Not yet or Later. Then the daily log, empty until the first score.
- Setup — the module’s questions, the answer Beacon reads from your billing, the question it shares with Segments, and the company questions it uses.
What it needs
Forecast confidence reads no tool directly. Every reading arrives through another Beacon module; these are the tools behind them.
See Connect your tools for how each connection is made.
Reads from
Which modules read it
Every one of these reads the score or a part score, which are not calculated, so none receives anything from this module yet. The Modules page in The Brain shows which modules are running for you.
Alerts
Seven alert rules are described. None is about a single deal or customer — those come from the modules that read them. These are about whether the forecasts can be trusted.
No alert is sent yet. Every one waits for the score or the part scores.
Setup questions
- Who owns forecast confidence, and who is told? The owner — usually the CFO — is responsible for the score, the levels and the statement to the board; the account owner approves level changes; the RevOps lead watches the parts daily. Picked from your Beacon users. Not available on the page yet.
- In a usual quarter, are renewals more than a quarter of your yearly revenue? If yes, renewal stability weighs a little more. Answers: yes; no; not sure — check once renewal dates are known.
- How often does a forecast go to your board? Each time, Beacon attaches the score, the six parts and a plain statement of how far the forecast can be trusted. Answers: monthly; quarterly; not yet.
What is not calculated yet
- Forecast confidence — the score from 0 to 100 and its band.
- The score for each of the six parts.
- What Beacon may do by itself — until there is a score, a person confirms every action on a forecast.
- The score over time, the weekly history and the confidence statement for the board.
- Alerts — none is sent yet.
- Data integrity’s trust figure, and the Churn risk, Deal health and Deal likelihood scores this module waits for.
Related
Modules
The ten module families and which modules are available today.
Forecast confidence
The score from 0 to 100, its six parts, the bands and what it governs.
Data integrity
The morning checks behind data quality — the part every forecast depends on.
How confident a number is
How Beacon says how much of a figure it can see, and how sure it is.
Churn risk
Subscriptions, renewal dates and the risk score renewal stability waits for.
Deal likelihood
Your open deals, and the chances pipeline stability will read.