How the MSP Score is built

The MSP Score runs from 0 to 100 and is the weighted sum of five pillars. Every input is observed from public sources or supplied by the company itself. Nothing is modeled, estimated or bought.

What the score is

What goes into each pillar

Reputation, 30 of 100

Google rating and Google review count.

Presence, 25 of 100

Website maturity, PageSpeed, and DNS posture.

Credibility, 20 of 100

Certifications, partner designations, and years operating.

Scale, 15 of 100

Headcount and number of locations.

Momentum, 10 of 100

Distinct change types detected in the last 180 days, 25 points per distinct type, saturating at 4 or more.

The momentum window

Momentum counts the number of distinct change types detected at a company in the last 180 days. Each distinct type is worth 25 points and the pillar saturates at four or more types, so one type scores 25, two score 50, three score 75, and four or more score 100. Repeat changes of the same type do not add points.

The window is a hard 180 days with no decay curve inside it: a change counts in full on day 1 and on day 179, and drops out entirely on day 181. That is why a company's momentum can fall without anything at the company getting worse.

The 25 change types we detect

A change is only recorded when it clears its noise floor, measured against the same source that reported the earlier value.

Data confidence

Data confidence is the share of the six scoring inputs we have actually observed for a company, expressed out of 6.0. It says how much of the record is filled in, not how good the company is.

Since 2026-09-17 the divisor includes change detection, so confidence figures measured before that date are not comparable with today's. A lower number after that date is a correction, not a regression.

The rank denominator

Every national rank is printed against national_total, the count of ranked providers at the moment the page is served. It is read live on every page, never stored in the text of a page and never carried over from an older run.

Crawl and detection cadence

Site re-crawl every 7 days since Sept 16, 2026, monthly before that; same-source field-change detection since Sept 17, 2026.

Formula change log

• 2026-09-17

  momentum changed from a raw signal count to a count of distinct signal types, 25 points each. data_confidence divisor changed from 5 to 6 to include momentum.

Why coverage matters

81.2% of the firms we track show no detected change in the last 180 days. That is not missing data. It is a measured finding about a market where most firms do not move in a way anyone outside them can see: no new service listed, no certification added, no rating shift, no site rebuild, no hiring visible from outside.

So a momentum score of zero is an answer, not a gap. It says we looked at the whole detectable surface of that company for six months and nothing changed. When a firm does move, it moves against a mostly static field, which is exactly what makes the pillar worth 10 points.

See the published aggregates behind these statements on published statistics, and where each field comes from on data sources.