Terminal
Terminal/Skill
Labels each contact in a list as buyer, influencer or not a buyer for your product, using job titles and company context.
By Clay
Works with
npx skills add clay-run/clay-skill-creator --skill buyer-classificationThe insight: is_buyer is a product-relative verdict, not a title attribute — the same Controller is THE buyer for spend management and noise for DevTools. There is no universal buyer list to look up, so classification starts by deriving two requirements from the user's product: the FUNCTION that owns this purchase, and the SENIORITY that can decide it — both must hold. Then the cheap discipline: deterministic keyword gates run first (auditable, free, identical on re-run), judgment spends only on the residual, and company headcount decides what an ambiguous middle title means. The naive version asks an LLM "is this a buyer?" per row and gets confident, unexplainable, product-blind answers.
Nothing here ships with a value. Each one is the installer's, not the author's: ask for it, never substitute a plausible default, and where an answer does not exist say which step becomes unavailable rather than guessing. Where a default IS defensible it is named below, and using it means saying so in the output.
| Input | What the installer supplies | If it is missing |
|---|---|---|
| The product, and who owns buying it | one sentence each | no default — the classifier is derived from this |
| Function definition | which department owns or directs the purchase, with inclusion keywords and exclusion traps | ask. For a finance product, "Account Executive" is sales, not accounting |
| Seniority line | the deciding tier, and the sub-decider line below it | ask — where a title stops being a decision-maker is theirs, not ours |
| Persona enum | 5–8 labels their team actually uses | ask, then add Influencer, Non-Buyer Leader and Unclear |
| Output shape | a lean is-buyer flag, or full committee framing | lean by default. Build the committee machine only when something downstream genuinely consumes it |
If an answer sheet is present beside this skill, load it and ask only for what it does not cover. A partial sheet is normal; a value it is missing gets asked for on its own rather than restarting the interview. Say which values came from the sheet before using them — a sheet applied silently is a wrong field nobody catches. If there is no sheet, say nothing about sheets — the check is a file lookup, not a question, so run the interview as though the feature did not exist rather than reporting an absence. At delivery, offer to save the answers back (identifiers only — never a token or a password), private and never published — and phrase the offer so it explains itself: "want me to save your answers to a file, so the next person on your team doesn't have to answer these again?"
Run clay whoami; echo "exit_code=$?". If it fails or Clay tools are missing,
run the Clay plugin's setup skill, restart if it says to, and re-run this
skill. (Classification itself spends no credits; Clay is needed only if titles
are missing and the user wants enrichment first — that's a sibling's job.)
Before touching a row, establish with the user:
references/classification-mechanics.md):
Influencer, Non-Buyer Leader, and Unclear.Show the derived definition (keywords, exclusions, enum) and get approval — it IS the classifier; the user must be able to disagree with it in review.
Per contact (dedupe first; title required — no title → unclear, never a
guess):
via: judgment — never batch
them into vibes. Still stuck → Unclear with the reason.Every verdict carries: is_buyer · persona · confidence (gate-matched / judgment / unclear) · rationale (≤20 words quoting the title evidence).
The classified table + the funnel (contacts in, buyers, influencers,
non-buyers, unclear) + the definition used (keywords, exclusions, thresholds —
so the next run is reproducible) + a review sample: every judgment row and
5-10 gate rows for spot-checking. Flag Procurement/Legal/Security-type roles
separately when relevant: they classify cleanly but are process gatekeepers,
not buyers — the user decides whether they get outreach.
Unclear for empty/undecidable titles — NEVER a guess, NEVER a
silent drop; absence of title evidence is not evidence of non-buyer.Ask: "Tag our 400-contact conference list — we sell spend-management software. Who's a buyer?" Interview → FUNCTION: finance/accounting (keywords: finance, FP&A, controller, treasury, CFO…; exclusions: account executive/manager = sales); SENIORITY: director+ (with the manager-at-small-co modifier, threshold 200); enum: CFO · VP Finance · Controller · Finance Decision-Maker · Finance Influencer · Non-Finance Leader · Unclear. Approved. Classify: 396 unique; gates place 371 (94%): 44 buyers (11 CFOs, 9 VP Finance, 24 director-tier), 61 finance influencers (analysts, AP specialists, one "Chief of Staff to the CFO" caught by the relation rule), 259 non-buyers (a "Account Executive, Mid-Market" correctly SALES), 7 headcount-modified (finance managers at sub-200 companies → Finance Decision-Maker, threshold stated). Judgment places 18 long-tail titles (each with rationale); 7 land Unclear (blank, emoji-only, two untranslatable). Deliver: table + funnel + the definition block + review sample. 0 credits.