Skip to main content

​
What it is

signal_coverage_rate is the share of your active customers whose signal coverage score is high. The aim is 85% or more. It is the headline figure of the Signals module. A customer’s signal coverage score says how much of the signal Beacon expects for that customer it can actually see and trust: whether the tools that would show a payment slipping, usage dropping or a deal stalling are connected and up to date, whether the customer is matched in them, and whether the data behind each signal is reliable. A high score means Beacon would notice when something changes for that customer. A low score means it might not. So this figure answers a simple question: across your customers, how much of what happens can Beacon see? It says nothing about whether the news is good or bad. That is what the signals themselves are for. It is published for your company as a whole, on the company and period object.
This value is not calculated yet. The rule below is fixed, but nothing works the figure out today. It needs two things that are not calculated yet either: Customer matching’s confirmed matches and the data confidence worked out by Data integrity. Every door returns not_computed for it. Nothing is estimated in its place.

​
How it is calculated

Modelled. The share of your active customers whose signal coverage score is in the High tier: a score of 80 or more, with data confidence of 80 or more behind it. It is a ratio from 0 to 1, shown as a percentage. Each customer’s score is worked out in three steps.
  1. Each signal Beacon expects for the customer counts as covered or not. It is covered when its tool is connected and has synced recently, the customer is matched in that tool, and the data confidence of the field it reads is 60 or more. A covered signal counts in full when confidence is 80 or more, and at confidence divided by 100 between 60 and 79. Below 60 it counts as not covered.
  2. Each kind of signal gets a coverage from 0 to 1, using the weights of the signals inside that kind.
  3. The nine kinds are combined at fixed weights, and the total is put on a scale from 0 to 100.
A tenth kind, company growth, is measured once for the whole company and is not part of any customer’s score. A kind that is not switched on yet for your company counts as not covered. How old a signal is does not matter here. A quiet customer on connected, up-to-date and trusted tools is fully covered. Signals fade over time when Beacon weighs what they mean; that does not reduce coverage. Why it is modelled. The weights of the kinds are fixed. The weights of the individual signals inside each kind are not entirely fixed: once your company is fully set up, Beacon may adjust them by up to 5% on its own, as it learns which signals best predict what happens to your customers. Larger changes always need a person’s approval. So this figure can move slightly because Beacon tuned itself, not only because your data changed.

​
Where it comes from

No read interface or MCP tool serves this value yet, and the value itself is not calculated yet. This table publishes its shape ahead of both. Because it is modelled, it is written into a CRM as a band rather than an exact figure.
Where the choice is made. This value is not open by default on any door. The choice happens when a reader grant that reaches it is issued or widened, when you subscribe to events, or when you turn the CRM property on. Whoever makes it is shown who becomes able to see the value, and their yes is recorded.

​
How fresh it is

Once calculated, it moves when a tool connects or stops syncing, when a customer is matched, or when data confidence changes. Which reading stands for a period is not yet declared; until it is, read the value with its own computed_at. Both timestamps travel with it: the moment the value describes, and the moment Beacon last worked it out.

​
Currency and rounding

It carries no currency. It is a share of customers.

​
What changes it

Connecting a tool, or a tool falling behind on its syncs. A customer being matched, or a match being undone. Data confidence rising or falling on the fields signals read. A new customer arriving. Beacon adjusting the weights inside a kind of signal, within its 5% limit.

​
What it is not

  • Not good or bad news. A customer at high risk of leaving can be fully covered. Coverage says Beacon can see; the signals say what it sees.
  • Not signals_noticed. That is how many signals fired in the period.
  • Not data_integrity_rate. That is the share of records passing Beacon’s daily checks. Data confidence is one of the inputs here; the integrity rate is not.
  • Not a count of connected tools. Two companies with the same tools can differ widely, depending on how well their customers are matched and how reliable their data is.