Score and Segment

Outcome: every repaired record placed in a fit-by-recency quadrant, with a decided action per quadrant including an archive rule.

Surface: Sync GTM app
Level: Intermediate
Uses: Formula columns
Credits: 0
Prerequisite: Lesson 04's repaired records

Two axes

Fit — should we sell to them?

Computed from the now-repaired firmographics: size, geography, industry, stack. Free formula.

Recency — is there any relationship?

Last activity, last opportunity, engagement. Also free — it comes from the CRM.

Both were unreliable before lesson 04. Scoring a CRM before repairing it produces a tiering built on the gaps rather than on the data.

The four quadrants

	High recency	Low recency
High fit	Work now. Active and qualified.	Reactivate. Good accounts that went quiet.
Low fit	Review. Why are we engaged with a poor-fit account?	Archive. Neither qualified nor engaged.

Each quadrant gets a different treatment, and naming them is what turns a score into an action.

The reactivate quadrant is usually the most valuable and the most neglected. These are accounts that fit and that you have already spent money acquiring — and after a repair pass, many of them have new contacts, new headcount, or a new decision maker. A reactivation campaign against a repaired list outperforms cold outbound by a wide margin.

What to do with each

Work now

Route to the owner with the repaired data. Flag anything that changed materially — a contact who moved, a company that grew a band, a newly verified email.

Reactivate

The strongest play in this course. For each:

Check whether the original contact is still there — if not, that is a job-change lead
Check whether anything changed at the company: funding, hiring, a new exec
Build a reactivation message around the change, not around “checking in”

Review

Low fit, high recency usually means one of three things: the ICP definition is wrong, someone is working an account they should not be, or the record is misclassified. All three are worth knowing.

Scoring on repaired data


fit_score
  employee_band in target        +3
  country in target list         +2
  industry matches ICP           +2
  stack signal present           +3
  industry_source = "inferred"   -1

recency_score
  activity in last 90 days       +5
  activity in last 12 months     +3
  open opportunity ever          +2
  closed-won ever                +3
  no activity in 24 months       -3

Note the penalty for inferred values from lesson 04. A record qualified on machine-inferred data is a weaker qualification than one qualified on observed data, and the score should say so.

The archive rule

Write it explicitly, because nobody will volunteer to archive anything.


archive_if:
      fit_score < threshold
  AND no activity in 24 months
  AND no open opportunity
  AND never closed-won
  AND not manually flagged as strategic

Five conditions. Anything meeting all five is not going to be worked, and every enrichment run that includes it is waste.

Do this now

Build both score columns

Free formulas over repaired fields.

Set thresholds

Fit from your ICP, recency from your definition of active.

Assign each record a quadrant

Count each quadrant

The distribution is itself a finding about your database.

Route work-now records to owners

With the changes flagged.

Build the reactivate list

And check each for a change worth referencing.

Review the low-fit, high-recency records

Find out which of the three causes applies.

Apply the archive rule

And report how many records it caught.

Check your work

Scoring ran on repaired data, not raw
Every record has a quadrant
Inferred values are penalized in the fit score
The archive rule is written with all five conditions
The reactivate list has been checked for changes worth referencing

Where this breaks

Archiving on fit score alone will remove accounts that fit perfectly and were simply mis-enriched — a company whose industry came back wrong, or whose headcount was undercounted. That is why the rule requires five conditions rather than one. Before applying it at scale, read fifty records it caught and confirm you agree with every one.

Further automation

Sync the quadrant and both scores back to the CRM so account owners see them on the record. A tiering that lives only in a table changes nobody’s behaviour; one that appears next to the account name changes what gets worked.

Next lesson

06 — Push updates back, the step where staged changes finally reach the CRM.

Reference for this lesson: Tables, CRM integrations, TAM Sourcing, GTM Engineering — scoring and gating.