Section 2 · Qualification & Opportunity Scoring

2.1Hidden 0-100 opportunity score

Built in the POC

No.: 2.1
DMA's reference: Decision Book p.11
Phase: Phase 1
Autonomy: L4 · Autonomous
DMA's decision: Yes

★ Marked "non-negotiable, build first" in the book.

What DMA asked for · Decision Book p.11

Every prospect gets an internal 0–100 score from company size, spend, pain, authority, timing and fit. The prospect never sees it.

Why it matters to DMA: It is the routing backbone: priority lists, alerts and next steps all depend on it.

In our words, from DMA's Product Decision Book v3, page 11.

How we do it

The AI's job is to find the evidence: it reads the notes, deals and conversations and extracts facts, each with the exact quote it came from. Code, not the AI, turns those facts into the score, so the same facts always give the same score.
Two scoring versions: one for the dormant Nutshell records (weights size 15, spend 10, pain 10, authority 10, timing 25, fit 15, history 15) and one for website visitors (size 15, spend 15, pain 20, authority 15, timing 20, fit 15). Tiers: A from 70, B from 45, C below.
Safety gates can cap the tier whatever the score: too little evidence, no lawful basis to email, no reachable contact, a duplicate, not a buyer, or a small account with nothing live.
Every score is stored with its reasons, the points each reason earned, the evidence, and a plain "could be wrong if" line. DMA opens it with Why? on any list. The score, tier and reason are written to Nutshell's AI fields through the write gate.

The flow

It plays on its own while it's on screen; hover or use the controls to pause or step through.

Step 1 of 5

The record

Northside Dental Group: a former client, notes from 2022 and 2026, a lost renewal.

Nutshell
AI
Stored data
Rule in code
Routing

Where the data goes

The same six stages on every page. Nothing reaches Nutshell except through the write gate.

1Nutshell and chats
Source
Notes, deals, people, and website conversations.
2Sync
Copies Nutshell changes into the bridge database
Brings each record into the mirror; a change triggers a re-score.
3Bridge database
A copy of the CRM data, plus what the AI works out
Facts with quotes; the score, tier, reasons and counter-evidence.
4AI
Claude models, only through DMA's own gateway
Extracts facts only; it never sets the number.
5Write gate
The only way back into Nutshell: checked, approved when needed, sent once
Writes AI Score, AI Tier and AI Reason to Nutshell (allowed on its own, L3).
6Nutshell
The system of record
The team sees the AI fields on every record.

The tables behind it

Drawn from the POC's database catalogue: structure only, no data.

PK primary keyFK reference the database enforces (solid line)ID reference kept by id (dashed line)

Table	What it holds	Columns
ai.extracted_facts	Facts found in the notes, each with its quote and confidence.	15
ai.scores	One current score per record: tier, gate, reasons with points, and the could-be-wrong line.	20
ai.scoring_configs	DMA's scoring versions: weights and tier cut-offs, changed only with approval.	15

How the tables connect (3)

Column	Points to	Kept by
ai.extracted_facts.prompt_id	agent.prompts	The application (by id)
ai.scores.prompt_id	agent.prompts	The application (by id)
ai.scoring_configs.approval_id	ctrl.approvals	The application (by id)

Worked example

Synthetic demo data: every name, business and number is made up.

Northside Dental Group's score, from list to Nutshell

1 / 5

What you see

1.Ranked by score on Today

The top revivable accounts, highest opportunity first, each with a value and a one-line reason.

localhost:13301/

Today: accounts ranked by opportunity score.

What happens in the system

Bridge databaseai.scores
Today reads each account's current score and tier; Northside is 85, tier A.
Rule enginedormant scoring v1
The active scoring version, approved by a person; a new version needs a new approval.

Northside Dental Group in the prototype's demo data (synthetic account): the score as the system stored it.

Why Northside scores 85 (tier A)

Reason	Evidence	Points
Incumbent contract ending	"contract ends March" (note, June 2026): March 2027, upcoming	25
Former client, fixable churn	"unhappy with reporting"	13.5
Fits several services	SEO and local search	12
Multi-location	"all three offices"	12
High spend	"15k SEO" a month	10
Decision maker known	Ops Director on the account	8.5
Total	Tier A (70 and above). Next step: win-back email to Dana	85

What else is stored

Could be wrong if: They renew with their current agency before March 2027, or the reporting problem that drove them away hasn't been fixed.
Deal value: $144,000 a year
Seen by the prospect: Never

In the running POC

localhost:13301/

1 / 4Today: accounts ranked by opportunity score.

1What the top of the list is worth.
2How many dormant records were reviewed.
3Draft emails for the whole list.

Dashboard · Today. Captured from the running POC on September 25, 2026; synthetic data.

See it live

Opens the exact screen in the running POC, in a new tab.

Open	Where	What to try
Today	DMA AI dashboard	Open Why? on any row.
Accounts	DMA AI dashboard	Every account with its score and tier.

Status

Built in the POC

Built in the POC. This capability runs in the POC today, on synthetic data, end to end.