── SKILL
icp-analyst✓ APPROVED

Score whether an account is a real ICP fit.

Rolls firmographics, product usage, hiring intent, and tech stack into one 0-100 verdict, and flags where a scoring tool is wrong.

Inside: The weighted composite, the per-source breakdown, and channel tagging.

01 / HOW TO USE

Install it in one line, or paste it in.

1
In Claude Code (one command)
Copy the install line, paste it into your terminal, and restart Claude Code. The skill installs itself to ~/.claude/skills/ and runs automatically when it is relevant.
2
In Claude, ChatGPT, or a Project (no terminal)
Open the file, then upload it to your chat or paste its contents in. A skill is just a markdown file of instructions, so any capable AI can follow it.
New to skills? A skill is a plain-text file that teaches your AI a workflow. Point any capable assistant at it and it follows the steps, on your data.
02 / MAKE IT YOURS

Connect your context. Set it to your motion.

CONNECT YOUR CONTEXT · AND WHY IT HELPS
a CRM

it reads firmographics and any existing scores automatically across your book.

an account fit score or scoring vendor

it folds that score in as one input and tells you when to trust it.

a product-analytics tool

it adds a product-side fit signal from real usage.

an enrichment tool

it confirms employee count, hiring, and technographic stack.

SET IT TO YOUR MOTION

This was built for a B2B team scoring accounts against a defined profile. Set these to your stack:

Run any profile you like. The skill scores "does this account match the profile you defined," so point it at your criteria, not anyone else's.

Set thisWhat it isDefault / Example
FIT scoreyour account fit score or vendor scoreany account-scoring source
PROFILEwhat "good" looks likeyour ICP definition (sizeindustrystageregion)
SIGNAL sourceshiring, funding, product usagean enrichment toola product-analytics tool
DISPLACEMENT stackcompetitor tools worth a switch playthe tools you replace
WEIGHTShow much each input countssee The method (re-tune)
03 / THE FULL SKILL

Everything the skill does, in full.

── WHAT THIS DOES

Takes an account, or a whole list, and tells you whether it actually fits your ideal customer profile. It rolls every signal you have, firmographics, product usage, hiring intent, tech stack, into one 0-100 score, shows you which source drove it, and flags the accounts where a scoring tool is plain wrong. Then it tags each verdict with the channel it belongs to, so the play is obvious.

── THE METHOD
  1. 1
    Composite fit score (0-100)

    One headline number, built from weighted inputs. A starting split that you re-tune:

    • 35% account fit: employees, industry, funding stage, region.
    • 25% product-side signal: real usage or a product-engagement score, if connected.
    • 20% hiring and intent: recent hiring, funding, growth signals.
    • 15% tech-stack displacement: accounts running tools you replace are switch candidates.
    • 5% qualification completeness: a bonus if a rep has already done the homework.

    Penalties subtract from the total: too small or off-profile, an already-saturated customer, recently churned inside a cooling-off window, or already worked by another rep (avoid the double-tap).

  2. 2
    Source breakdown, never a single number

    Every score shows its parts. If a source is missing, the skill says so ("no product-analytics connected, score is from fit, hiring, and stack only") instead of hiding the gap.

  3. 3
    Override detection

    When the composite disagrees with a scoring tool, it says so out loud. An account a tool scored low while hiring 14 reps and running a stack you displace is a false negative the tool missed. An account a tool scored high with none of the real signals is a false positive. Both get flagged with the reason.

  4. 4
    Channel classification

    Every verdict is tagged Inbound (they came to you), Product (they self-served a signup), or Outbound (you go to them). The play and the outreach differ by channel, so the tag ships with the score.

  5. 5
    Lookalike search

    Anchor on your highest-value customers and find accounts matching their profile, employee count, industry, funding stage, and technographic stack.

── QUALITY GATES
  • No single-source fit claim. The breakdown by source ships with every score, and missing sources are declared.
  • Override visibility. When the composite disagrees with a scoring tool, it says so and names why.
  • Channel-tagged outputs. Every verdict carries Inbound, Product, or Outbound, because the play changes with the channel.
── OUTPUT (EXAMPLE)
EXAMPLE OUTPUT
ACME CORP · Fit Score 87 / 100 · STRONG FIT

Source breakdown:
  Account fit score:        78 / 100  (strong, approved)
  Product-side signal:      real usage detected
  Enrichment:               805 employees, 22 reps, 14 hiring
  Tech-stack displacement:  3 tools you replace
  Qualification:            complete, homework done
  Channel:                  Outbound + Product (hybrid)

Signals:
  Account fit: SaaS, 500+ employees, late stage, target region
  Hiring 14 reps is a top-of-funnel scream
  3-tool consolidation is a strong messaging hook
  A rep has already qualified it

Verdict: STRONG FIT. Pursue with full outbound. The hybrid channel means a
cold sequence into the usual entry role, plus a watch for any signup from
this domain that would trigger an inbound handoff.

Override: none. The fit score and the composite agree.
── WHERE THE NUMBERS COME FROM

The weights (35/25/20/15/5) and the penalties are a starting point, not a law. They suited one B2B motion. If product usage matters more to you than hiring, shift the weight. The scoring logic does not change, the weights are yours.

── EXAMPLE CHAT
EXAMPLE CHAT
You
Is acme.com a real fit? Score it against our profile, and tell me if the vendor score is right.
CLAUDE
Composite lands at 87, strong fit, and it overrides your vendor tool. Source breakdown: Account fit score: 78/100 (approved) Product-side signal: real usage detected Enrichment: 805 employees, 14 hiring Tech-stack: 3 tools you replace Channel: Outbound + Product (hybrid) The override is the finding: your vendor score said 78 and called it borderline, but 14 open sales reqs plus a 3-tool displacement opportunity are the kind of signal a firmographic-only score can't see. That's the false negative this skill exists to catch. One sharpener: connect a product-analytics tool so the usage signal comes from real data instead of "detected," which right now is the softest part of the read.
04 / GO FURTHER

Where an operator takes this next.

The verdict is step one. Here is where an operator takes it once the manual version proves out.

You built the override once, now it runs against every account, every night.

01
Score the whole book overnight

Point a scheduled Claude task at Salesforce and your enrichment tool nightly, and write the composite and the override flag back to the account record.

02
Catch only the disagreements

Post to Slack just the accounts where the composite and the vendor score diverge by more than 20 points, so a human only looks at the interesting ones.

03
Turn the strong fits into a list

Feed every STRONG FIT verdict into Clay or a sequencer so the outbound list builds itself from the accounts that actually clear the bar.

05 / PART OF A BIGGER PLAY

One skill is the on-ramp.

── GET NEW SKILLS IN YOUR INBOX

One operator-built skill a week. Free.