Clay logo, go to homepage

Search Marketplace

Browse all Skills
Score & qualifySkill

Score Accounts With Jev

Builds a Clay workflow that scores and tiers your target accounts by combining plain numeric rules with Jev's judgment calls on qualitative fit.

Shy Rahnama
Shy Rahnama

Partner at Redline Growth

About this Skill

When to use it

Use this Skill when you want to score and tier your target accounts against your ICP using Jev, TypeSafe's decision model, inside Clay. It builds a scoring rubric from what it already knows about your business, previews it on your real accounts, and turns it into a callable Clay workflow that returns a score, a tier and the reasons behind it. Avoid using it for scoring individual contacts, finding or enriching companies, routing or assigning leads, pushing scores into a CRM, or training a model on past deals.

What it delivers

You get a published Clay workflow that scores any account from 0 to 100, assigns it a tier, and names the criteria and reasons behind that score. It supports deciding which accounts to prioritize while flagging accounts with too little data instead of guessing at a grade.

SKILL.md preview

Sign in to install and view the complete SKILL.md

Illustrative preview. Sign in for the actual file.

How someone would prompt this

Can you set up account scoring in Clay using Jev, pulling companies from our target accounts table and telling me why each one lands in its tier?

Inputs & outputs

What someone provides before the Skill runs, and the reviewable artifacts it returns.

Inputs

  • 01Account source
    • Choose where accounts come from: a Clay table, a saved Audiences audience, or a webhook from another system
    • this decides which fields are available to score.
  • 02Business & fit context
    • Describe what you sell and who you sell to, if it is not already found in your notes or memory
    • name the company types that are best-fit, ok, and poor-fit.
  • 03Scoring criteria
    • Set numeric cutoffs like headcount or region, only for fields your data actually has
    • name disqualifying competitors or segments, or say none
    • flag signals worth points
    • confirm or adjust the tier cutoffs.
  • 04Jev API access
    • Provide a TypeSafe or OpenRouter API key, saved locally or entered through a secure prompt
    • reused automatically if one is already found.
  • 05Clay connection setupPoint to an existing Clay HTTP API connection holding that same key, or create one in Clay's UI when asked.

Outputs

  • 01Scoring rubric cardYou review a plain-language summary of every rule and Jev question the score uses, complete with points and cutoffs, and correct it before the agent saves it.
  • 02Ten-account previewYou see ten real accounts scored with the exact rubric code the workflow will run, then confirm the scores match your judgment before the agent builds anything.
  • 03Delivery summary & call guideA final summary of the published workflow's rubric, connection and cost, plus the three ways to send it accounts: from a Clay table, an audience, or a webhook.

Representative output

What comes back on every account

FieldExample
Score98
TierA
Whya distributor (+25) · runs its own delivery fleet (+24) · 200–1,999 staff (+15) · explicit fleet expansion (+15)
Needs reviewany answer Jev was less than 60% sure of
Statusscored · disqualified · insufficient_data · not_scored

A reason counts for an account as (+25) and against it as (2 of 25), so a low score always names what pulled it down.

How the score is built

Every criterion is decided either by plain code or by Jev — never both.

CriterionDecided byPoints
Company sizeCode50–199 +10 · 200–1,999 +15 · 2,000+ +5
RegionCodeUS or Canada +10
CompetitorCodenamed domains disqualify
Runs its own fleetJevyes +25
SegmentJevdistributor +25 · field service +15 · retailer +8 · logistics provider disqualifies
Growth signalJev0 to +15

90 points possible. A ≥ 70 · B ≥ 45 · C ≥ 25 · D below that. Under 50% coverage an account comes back insufficient_data — never a low tier.

Jev is TypeSafe's decision model. It answers the judgment questions — *is this a distributor? does it run its own fleet?* — and returns a probability for every option rather than a single label, so an unsure answer moves the score less than a confident one. All three questions go in one request per account. Code does every number: the points, the tiers, the size bands. About $0.03 per 1,000 accounts.

Ten accounts, scored

AccountScoreTierWhy
Northwind Supply98Aa distributor (+25) · runs its own fleet (+24) · 200–1,999 staff (+15) · fleet expansion (+15)
Adventure Works HVAC58Bfield service (+15) · 50–199 staff (+10) · North America (+10) · against: no own fleet (8 of 25) — review: runs its own fleet, unsure (32% yes)
Fabrikam Freight51disqualifieda logistics provider — Jev ruled it out
RouteWise22disqualifieda competitor — a rule; Jev was never called
Tailspin Tools11insufficient_data50–199 staff (+10) · no data for Region, Own fleet, Segment, Growth signal

Across all ten: A 2 · B 3 · C 1 · D 1 · disqualified 2 · insufficient_data 1 — total cost $0.0002.

Once it's live

WorkflowJev lead score: accounts · Route-planning fit (published)
Rubric3 rules, 3 Jev questions · tiers A/B/C/D
Jev keyin your Clay connection Jev (TypeSafe)
Smoke testNorthwind Supply — local 98 A, Clay 98 A, answers match ✓

Sending records in

The Skill builds one published workflow. Three things can call it:

A Clay tablebind it at node 1 Intake; the score lands back on each row
A Clay audienceaccounts are scored as they join
Anything that can POSTsend the fields to the workflow's webhook URL

Map your columns onto the rubric's inputs — in this example company_name, domain, description, industry, employee_count, country, recent_news. Your column names don't need to match.

How the Skill runs

Follow the steps in order. Each row includes the full instruction.

01
Explain cost and check setup

The agent explains what it will build, what Jev calls cost, and what data leaves your systems, then checks Clay access and any saved rubric or key before continuing.

Step
02
Pick account source

You choose whether accounts come from a Clay table, a saved audience, or a webhook, and the agent reads the available fields for that source.

Step
03
Build scoring brief from context

The agent drafts what you sell, your best-fit segments, and your disqualifiers from your notes and past conversations, then asks you only about anything missing.

Step
04
Draft and correct the rubric

The agent writes the full scoring rubric of rules and Jev questions and asks you to confirm or correct the points, cutoffs and criteria before saving it.

Step
05
Set up the Jev key

You confirm or provide a TypeSafe or OpenRouter API key; the agent saves it locally through a secure prompt and tests it with a small request.

Step
06
Preview on real accounts

The agent scores ten of your real accounts with the saved rubric and stops to ask if the scores look right before it builds any workflow.

Step
07
Connect the key in Clay

You confirm an existing Clay connection holding the same key, or create one following the agent's steps, since Clay's tools cannot create connections automatically.

Step
08
Build and publish, with approval

The agent shows exactly what it will write to your workspace and its cost, waits for your go-ahead, then builds and publishes the scoring workflow.

Step
09
Verify with one real account

The agent sends one preview account through the published workflow and compares it to the local score, stopping before live use if the results do not match.

Step
10
Deliver results and usage guide

The agent hands you the final rubric, preview numbers, and instructions for sending accounts to the workflow from a table, an audience, or a webhook.

Step

Connections required

Jev API (TypeSafe)Required connection
Clay CLI / agent pluginRequired connection
Clay Public APIRequired connection

Install Clay and run it

Sign in once, install the Skill, then run the prompt.

  1. 1Install ClayUse the connector or official plugin in your coding-agent environment.
  2. 2Sign up or log inAuthorize Clay access and return when the connection is ready.
  3. 3Install the SkillRun the one-line install command from the Add this Skill card.
  4. 4Run the SkillPaste the example prompt and review the result.

Version history

Version 1From submitted SKILL.mdCurrent

Related Skills

Score Contacts With JevBuilds a callable Clay workflow that scores your contacts on role, seniority, and fit using Jev's judgment plus plain code for the rest, returning a 0 to 100 score and tier for each person.View Skill →Account Tiering WorkflowThe Account Tiering Workflow automates the scoring and tiering of accounts as they enter a specified audience in Clay, enriching their data and writing the results back to the account record without human intervention.View Skill →Check Employment With JevBuilds a live Clay workflow that checks whether a contact still works at a company, and in what capacity, by judging their LinkedIn work history with Jev.View Skill →

Related searches

Score & qualify skillsJev API (TypeSafe) skillsClay agent plugin skillsClay API skills