Score leads before you waste sends

Demand-gen / head of growth / RevOps / GTM engineer · Assemble in ~10 min; first scored batch of 30–50 leads in under 30 min once installed.

GTMEdit flowon X◆ new (unverified) 89

step 1ready

Define or load ICP

icp-definition-and-validation1 alts

step 2connector

Ingest raw lead list or post-engagers

⚡ bring your own tool

step 3partial

Scrape intent signals per account

niche-signal-discovery12 alts

step 4ready

Research each prospect

account-research

step 5ready

Score ICP-fit per lead

sales-icp1 alts

step 6ready

Build prioritized, tiered lead list

prospecting

step 7ready

MEDDPICC qualification check on top-tier leads

meddpicc-qualification

step 8connector

Export scored + qualified list to sequencing

⚡ bring your own tool

Did this recipe work for you?

What you get

▸Tiered lead list (hot / warm / deprioritize) with numeric ICP-fit scores and intent-signal evidence per lead
▸MEDDPICC qualification notes for the hot-tier shortlist, with explicit evidence-vs-assumption splits
▸Per-lead account briefs covering triggers, buying committee, and mapped pains
▸Sequencing-ready export with only the leads worth a sequence slot — protecting sender reputation

tools: Intent data provider optional — G2 Buyer Intent from ~$200/mo, Bombora from ~$500/mo; without them the recipe runs on public web only at $0 tool cost

tokens: ~$2–5 per 30-lead batch (account research + scoring + MEDDPICC on top 10)

≈ $5–15/mo in tokens at weekly cadence; add intent-provider cost if you connect one

What stays yours

· Review the tiered list and override any scores where you have relationship context the agent cannot see (e.g., warm intros, prior conversations)
· Validate MEDDPICC gap findings before routing hot leads to sequence — the agent flags gaps but the human decides whether to proceed or gather more info first
· Provide the raw lead list or connect the sourcing tool each run — the agent does not autonomously pull new leads

The problem

When I have a raw list or post-engagers, I want an agent to research and score each on ICP fit, intent strength, and recency before any email goes out, so I spend sequence slots and sender reputation only on the highest-probability leads.

Cybernetic layers

Memorybyo

Scored list and qualification notes should be saved to a Google Sheet or CRM field so re-runs don't re-score already-processed leads. No harvested memory artifact covers this yet — a sheet or local CSV is the pragmatic solution.

Background-schedulingpartial

The scoring run is batch-shaped and takes 20–40 min for 50 leads; it can be triggered manually or via a Claude Code cron (CronCreate). Automatic re-scoring on new list arrivals would need a file-watcher hook — not yet a library artifact.

Harnessready

Claude Code is the runtime. Drop each ready skill into ~/.claude/skills/ and invoke by name. The skills chain naturally when fed output from the prior step as context.

Click any step to see what fills it — and the ranked alternatives you can swap in.