Why this exists

A Google Business Profile can look complete and still fail the customer.

It can have a category, hours, a phone number, a website, and 38 reviews — and still give a ready customer no appointment action, no after-hours path, a broken handoff, or no way to tell whether anyone responds.

Traditional GBP checklists reward field completion. Completeness is easy to count and hard to defend as a business diagnosis. A contractor does not lose a job because the description is 12 characters short. A contractor loses a job when a homeowner cannot tell what the business does, cannot take the next step, does not trust the profile, or never hears back.

The question

How effectively does the observable Google Business Profile support a local customer's journey from discovery to meaningful action?

Not:

How completely was every Google field populated?

The separation that matters

PUBLIC PROFILE SCORE          REVENUE OPPORTUNITIES TO INVESTIGATE
observable, scored            owner/CRM data, never scored

A strong public profile with unknown internal response time should score well and produce owner-data questions. It should not be penalized for a CRM it cannot see.

A weak public profile with a great CRM should still be diagnosed as a weak public profile.

The family

This repository is one engine in a coordinated scorecard family:

LOCAL VISIBILITY SCORECARD   Can customers discover the business?
            ↓
GBP SCORECARD                Does the profile create trust and enable action?
            ↓
WEBSITE SCORECARD            Can the owned site convert the demand?
            ↓
RAD                          Remediation and operations

The engines remain separate. A website defect should not be scored twice. A local ranking should not be disguised as a profile defect. RAD consumes findings; it does not rewrite the public score with private operational data.

Published scorecards:

• Local Visibility Scorecard
• GBP Scorecard
• Website Scorecard

What makes this credible

• Every check has a canonical ID, an anchor set, and an evidence type.
• Every scored finding has evidence and a confidence state.
• N/A checks leave the denominator and weights are renormalized.
• Owner-data questions are separate and never reduce the public score.
• Five hard caps are deterministic, documented, and tested.
• The scorer, renderer, and verifier are file-based and reproducible.
• The public example is synthetic.
• Privacy is part of the release gate, not a note at the end.

What it refuses to do

• Invent a review quote, a field, a call metric, or a revenue outcome.
• Call a business "best-in-class" or claim a guaranteed lift.
• Treat a missing optional service as a universal failure.
• Treat public availability as publication consent.
• Let a model decide which service to sell.
• Mix public and authorized data silently.