Methodology · Signals and products
How the Signals and their impact is calculated
Method version signals-v1 · Last updated
In short
A signal is a meaningful change in a tracked website, score or product since the previous measurement, written as what changed, why it matters and the move worth making. Changes below fixed thresholds are treated as noise; above them the change is rated high, medium or low impact by its kind.
Formula
Score jumps: Site health ≥ 5 points, Website score ≥ 10 points, AI content ≥ 10 points Sitemap change: ≥ 10% and ≥ 50 URLs · New section: ≥ 5 URLs · Structured data type on ≥ 3 pages Product: any change of the key figure (high impact) or of a fact (medium impact)
How it is calculated, step by step
- Compare every new measurement with the previous one of the same brand, module and method.
- Keep changes above the thresholds; ignore differences caused by a method change.
- Rate the impact and write the signal, insight and move.
What goes into it
- Two consecutive measurements
- Of the same brand and module, made with the same method version.
- Thresholds
- Fixed per kind of change (see formula).
Data sources
- Actavero's own measurements: website, site health, AI readiness, products.
How often it is updated
Change detection runs daily on new measurements; signals are e-mailed weekly (digest), instantly (high impact, Scale) or monthly (recap, Scale).
How to read it
- High impact means a change that typically affects how buyers compare you, such as a competitor's new product section or a price change.
Limitations
- Only what Actavero measures can become a signal.
Frequently asked questions
Why didn't a small change become a signal?
Changes below the thresholds are normal variation; turning them into signals would bury the important ones.
Related
- Product comparison — Claude reads your product and finds each competitor's equivalent on its own website; after your approval the products are compared fact by fact and re-checked monthly.
- Site health score — 45 technical and content checks on a crawl of up to 200 pages, grouped into six categories; each category loses points by how often its checks fail, weighted by severity.
