skills/score-accounts/SKILL.md

---
name: score-accounts
description: Score and rank a list of accounts (mixed ZoomInfo company IDs, names, or domains) by ICP fit + buying intent + recent triggers. Returns per-account composite score (0–100), tier (A/B/C), explainable component breakdown (fit / intent / trigger / engagement), a specific "why now" sentence per account, and the working weight set as a saveable search filter set. Resolves name/domain inputs via search_companies with explicit confirmation for ambiguous matches. Iteratively refinable — adjust weights, swap axes, retier, or drill into a specific account. Use for account-based selling, ABM list prioritization, territory planning, sales prospecting prioritization, signal-based selling, buyer intent ranking, B2B prospecting. Triggers on phrases like "score these accounts", "prioritize this list", "rank by ICP fit and intent", "which accounts should I work first", "build a tiered account list".
---

# Score Accounts

Rank a list of accounts by ICP fit + intent + trigger signals. Calls `get_gtm_context(detailed: true)` unconditionally, resolves mixed-identifier inputs explicitly surfacing ambiguity, scores each account on four axes, and presents both the ranking and the weight set as iteratively-refinable artifacts.

## The bar

1. **Resolution accuracy 100%** — every input auto-resolved / verified / ambiguous / failed. Nothing silently picked.
2. **Every score explainable** — composite is a transparent weighted sum, never an opaque number.
3. **"Why now" cites a specific signal** — not the composite restated.
4. **Every tier comes with a recommended action.**
5. **Weights and axes are exposed and overridable.**

Sellers reject black-box scores. Transparency + per-account "why now" are what make this skill trusted.

## Scope

Scores **company-level accounts**, not contacts. Persona-aware ranking is a chain target via `personalize-email` after tier-A is produced.

## Input

- **Accounts (required)** — list of ZI IDs / company names / domains / mixed CSV.
- **Use case (default `prospecting`)** — `prospecting`, `abm`, `territory_planning`, `pipeline_acceleration`. Affects tier thresholds + recommended actions.
- **Weight overrides (optional)** — `{fit, intent, trigger, engagement}` summing to 100.
- **Tier thresholds (optional)** — `{A, B}`. C is the remainder.
- **ICP override (optional)** — natural-language refinement on top of `get_gtm_context.icp`.
- **Intent topics (optional)** — explicit list overriding GTM-derived defaults.

## Four-axis framework

| Axis | Question | Source |
|---|---|---|
| **Fit** | Does this match our ICP? | `enrich_companies` vs `get_gtm_context.icp` |
| **Intent** | Are they actively researching topics we sell into? | `enrich_company_signals` (intent), matched to GTM priorities |
| **Trigger** | Fresh event creating a window? | `enrich_company_signals` (news + scoops), last 90d by signal date |
| **Engagement** | Already interacting with us? | `account_research` narrative for known accounts. If absent, weight redistributed. |

Each axis 0–100 independently. Composite is the weighted sum — never collapsed to an opaque number.

## Default weights

```
fit:        45%
intent:     25%
trigger:    25%
engagement:  5%   (redistributed if unavailable)
```

User overrides accepted. Weights are exposed in every output. Cache per-axis scores; recompute only the composite when weights change.

## Tier thresholds

| Tier | Composite | Recommended action |
|---|---|---|
| **A** | ≥ 75 | Route to AE for 1:1 outreach within 24h. Chain to `personalize-email`. |
| **B** | 50–74 | SDR sequence; ABM retargeting; nurture-to-meeting. |
| **C** | < 50 | Watchlist; monitor for tier-promotion signals. |

Use-case adjustments: `abm` → A=80/B=55 · `territory_planning` keeps defaults · `pipeline_acceleration` → A=65/B=40.

## Workflow

### 1. Pull GTM context (always)
`get_gtm_context(detailed: true)`. Capture ICP, personas, competitors, offerings, strategic priorities. ICP = fit-axis target; strategic priorities → intent-topic curation.

### 2. Honor input data first
Use user-supplied weights / thresholds / ICP refinements / intent topics. Fall back to GTM defaults only for missing fields.

### 3. Resolve identifiers (four-bucket routing)

- **Auto-resolved** — top match dwarfs alternatives. Score without confirmation.
- **Verified** — clear top match BUT plausible alternatives exist. Score; surface verification note.
- **Ambiguous** — no dominant match. Pause scoring; surface candidates.
- **Failed** — no match. List separately.

Routing by input type:
- **Numeric ZI ID** → auto-resolved.
- **Domain** (`.com` / `.io` / `.co` / `.ai`) → `search_companies(companyWebsite)`. Single match → auto-resolved. Multiple → ambiguous.
- **Name** → `search_companies(companyName)`.
  - Top match's size/revenue dwarfs alternatives → auto-resolved.
  - Clear top but 3+ plausible alternatives → verified with note.
  - No dominant match → ambiguous.
- **No match** → failed.

**Surface rule.** Never silently pick a winner. Present top 5 with attributes; ask user to confirm. Use GTM context as soft tiebreaker for `verified` (e.g., a B2B SaaS context defaults an ambiguous name to the SaaS-industry candidate over an unrelated-industry candidate, with a flag).

**Domain-confirmation gate (mandatory for high-collision names).** When `search_companies(companyName=X)` returns >100 matches AND no strong GTM tiebreaker exists, require domain confirmation. Surface top match's domain and ask. Never silently auto-pick — cost of getting it wrong is scoring the wrong company entirely.

**Duplicate-record detection (mandatory).** If top candidates share the same domain root (e.g., `acmeco.com` and `acmecoinc.com`) AND ≤20% revenue diff AND same metro/country → flag suspected duplicate. Surface both records and offer to union. For signal-heavy workflows, scoring both and unioning is the right default — signals may be split across records.

Resolution path must hit 100% accuracy. Score auto-resolved + verified immediately; pause ambiguous; list failed separately.

### 3.5. Relationship-context pre-flight (mandatory)

Tag each resolved account against GTM context. Tag visible on the row before the tier letter — sellers see relationship status BEFORE running the play.

- **`competitor`** ⚔️ — in `get_gtm_context.competitors`. Don't exclude from ranking (competitive intel matters) but make it impossible to miss visually.
- **`customer`** 🤝 — in `get_gtm_context.customers` / `proof_bank`. Shift recommended action to expansion / renewal.
- **`partner`** 🔗 — in `get_gtm_context.partners` / `integration_partners`. Shift to co-sell / integration angle.
- **`prospect`** — default; no tag.

In the row label: `⚔️ [Account] (B 62)`. Skill never silently produces "pursue this competitor" rankings.

### 4. Define the intent relevance set
From `get_gtm_context.strategicPriorities`, offerings, and competitor categories (or a user-supplied list), derive 5–10 themes you sell into. `enrich_company_signals` returns each company's active intent topics directly — there is no topic lookup or pre-query step — so these themes are the **match set** used in scoring (step 6): a returned topic counts toward intent only if it maps to one of them. Keep the themes; the matching happens per account during scoring.

### 5. Fetch data per account (parallel, batched ≤10; chunked for large lists)

Both calls below batch multiple accounts per request, so fetch a chunk of accounts together rather than one-by-one:
- `enrich_companies(zoominfoCompanyIds: [chunk], fields: industries, employeeCount, revenue, country, metroArea, businessModel, employeeCountByDepartment, foundedYear)` — up to 25 per call.
- `enrich_company_signals(zoominfoCompanyIds: [chunk], signalTypes: ["INTENT", "NEWS", "SCOOP"])` — up to 10 per call. Returns each account's recent intent topics (each with `signalScore` and `audienceStrength`), news (with `category`), and scoops (with `scoopType`), plus a `date` on every signal. Do **not** pre-filter on score, topic, category, or date — that is applied during scoring (step 6).

**Hard batch limit: ≤10 accounts per `enrich_company_signals` call (≤25 per `enrich_companies` call).**

**Batch + context-window discipline.** For lists >25 accounts, process in **chunks of ~25 accounts** end-to-end (resolve → fetch → score → compose row → write chunk → discard raw payloads) before moving to the next chunk. Don't accumulate full raw enrichment payloads for hundreds of accounts in working context — once per-axis scores + the winning trigger event + the winning intent topic are captured per account, drop the rest. For >100-account lists, summarize completed chunks into running totals (tier distribution, top-A list, multi-product anomalies, duplicate-suspected flags, missing-axes counts) and discard the per-account breakdowns from context. Output is built incrementally chunk-by-chunk so a long list doesn't blow context.

Skip `account_research` here; fire selectively in §7.5 for tier-A.

### 6. Score each axis

**Fit (0–100)** — compare `enrich_companies` to `get_gtm_context.icp`:

| Dimension | Max | Banded scoring |
|---|---|---|
| Industry / sub-industry | 25 | Primary = 25 · secondary = 15 · adjacent = 8 · none = 0 |
| Employee count band | 20 | In band = 20 · one off = 12 · two off = 4 · outside = 0 |
| Revenue band | 20 | Same banding |
| Geography | 15 | ICP country = 15 · in continent = 8 · outside = 0 |
| Business model | 10 | B2B/B2C match = 10 · mixed = 5 · mismatch = 0 |
| Technographic (optional) | 10 | Uses named tech-stack vendor = 10 · else 0. Verify via `search_contacts` + `techAttributeTagList` if needed. |

Cache per account; reuse across weight changes.

**Intent (0–100)** — from the `enrich_company_signals` intent topics, keep those that map to the relevance set (step 4) with `signalScore` ≥ 60 in roughly the last 30 days (use each signal's `date`). Score `max(signalScore × audienceStrengthFactor)` over the survivors. A=1.0 · B=0.85 · C=0.7 · D=0.55 · E=0.4 (from `audienceStrength`). Cap 100. Record the winning topic for "why now." If no relevant intent survives → 0 with "no relevant intent activity" flag.

**Trigger (0–100)** — from the `enrich_company_signals` news and scoop signals, kept to the last 90 days by each signal's `date` (drop older). Map each signal's news `category` or scoop `scoopType` to the weight below:

```
event_score = signal_type_weight × recency_factor
```

| Signal type | Weight |
|---|---|
| M&A, Funding, New CEO/C-suite hire | 95 |
| Product launch, Hiring surge, Earnings beat/miss | 75 |
| Partnership, New facility | 55 |
| Pain-point scoop, Other PERSON moves | 45 |
| Generic press release | 25 |

Recency: 0-14d=1.0 · 14-30d=0.7 · 30-60d=0.4 · 60-90d=0.2 · >90d=0.

Account trigger = `max(event_score)` capped at 100. Record winning event for "why now."

**Engagement (0–100)** — if `account_research` returns rich CRM context: active deal/renewal/champion = 80–100 · past meeting/known stakeholder = 40–70 · no history = null. If null, redistribute weight and surface gap.

### 7. Compute composite + assign tier

```
composite = round((fit × w_fit + intent × w_intent + trigger × w_trigger + engagement × w_engagement) / 100)
```

Assign per thresholds. Default A≥75 / B 50–74 / C<50 (use-case overrides apply).

### 7.5. Auto-pull `account_research` on tier-A rows (mandatory)

Tier A = "route to AE in 24h." Engagement-axis gap on tier-A is the highest-cost gap to close.

For each tier-A account (and ONLY tier-A — cost control): `account_research(zoominfoCompanyId, query="Open opportunities, active deal stages, named champion or blocker, last activity date, renewal timing")`. Parse for:
- **Open deal status** — stage, value, next step.
- **Renewal date** — surface prominently if within 90 days.
- **Named champion / blocker** — source-tag `[from account_research]`.
- **Last activity** — flag if >60 days old.

Append inline beneath the why-now:

```
| 1 | [Account] | 🤝 A | 84 | ... | [Trigger event] X days ago — [pain-bridge]
                                     ↳ Engagement: open deal $XXXk, champion [Name], last activity Xd ago [from account_research]
```

If no CRM history → annotate "no engagement signal — cold open."

For tier-B/C: skip — cost-to-value doesn't justify.

### 8. Compose "why now" per account

One sentence anchored on the strongest signal:

- **Trigger + in-tier fit** → cite event + date. "Closed [counterparty] acquisition 20 days ago."
- **High intent** → cite topic + score + recency. "Spiked on '[topic]' (score 92, audience A) over 14 days."
- **Strong fit, no fresh signal** → "Perfect-fit ICP — no fresh trigger; pursue on fit alone."
- **Engagement-driven** → "Active deal in flight; renewal due in 47 days."
- **Strong trigger BUT C-tier (fit mismatch)** → be explicit about routing: "Do not pursue — strong trigger (new CEO 10 days ago) but ICP mismatch ([reason]) keeps this low priority." Don't bury the trigger; surface BOTH signal and recommendation.
- **Low signal across all axes** → "Low signal — monitor only."

Never restate the composite as the why-now. Always cite the underlying axis driver.

### 9. Self-check before output

- ☑ Composite shown with component breakdown (fit / intent / trigger / engagement).
- ☑ "Why now" cites a specific signal, not the composite.
- ☑ Tier has a recommended next action.
- ☑ Weights + axes used exposed.
- ☑ Every input bucketed (resolved / ambiguous / failed) — none silently dropped.
- ☑ Ambiguous surfaced, not silently picked.
- ☑ Stale signals (>90d) contribute 0; not padded.
- ☑ Missing axes flagged + weights redistributed transparently.
- ☑ Iteration options offered.

### 10. Present + offer iteration

1. Accept ranking; save filter+weight set.
2. **Adjust weights** — re-rank without recomputing axes.
3. **Tighten / loosen tier thresholds.**
4. **Refilter** — remove tier C / specific industries.
5. **Swap ICP** — different ICP definition.
6. **Drill into one account** — chain to `personalize-email`.
7. **Add accounts** — extend list and re-score.

Re-execute step 5 only when account list changes. For weight / threshold / ICP changes → recompute from cached axis scores.

Terminate when user accepts, saves, or hands off.

## Anti-patterns — fail-fast checklist

1. **Black-box composite** — single number without component breakdown.
2. **"Why now" = composite restated.**
3. **Silent identifier resolution** on ambiguous names.
4. **Fixed weights not exposed.**
5. **Tier without action.**
6. **Stale signal padding** — events >90d contributing.
7. **Generic "why now"** — "good fit" applies to every account.
8. **Ignoring missing axes** — pretending engagement exists when null.
9. **Auto-accepting ambiguous matches.**
10. **No iteration affordance.**

## Fallback rules

- **`get_gtm_context` empty** → use user-supplied ICP override; surface gap.
- **No intent returned, or none matching the relevance set** → intent score = 0 (real signal, not a gap).
- **No news or scoops returned** → trigger = 0; flag.
- **Engagement unavailable** → weight = 0; redistribute proportionally.
- **All axes thin** → tier C "monitor only"; honest.
- **Resolution failure** → list separately; never silently drop.

Never block ranking on a single missing axis. Never invent data.

## Output Format

### TL;DR — Account Scoring · N accounts · Pass [M]

*Use case: [restate]. Weights · Thresholds A≥[X] · B[Y–Z].*

**Resolution:** [R resolved · A ambiguous · F failed]. [If A>0: "User confirmation required."]
**Tier distribution:** A: x · B: y · C: z.

**Top 3:**
1. [Account] (tier · composite) — [why now]
2. ...

---

### Resolution Summary

| Input | Resolved To | ZI ID | Confidence | Status |
|---|---|---|---|---|

`Status` legend: ✅ Auto-resolved · 🔍 Verified · ⚠️ Ambiguous · ❌ Failed.

**Ambiguous matches — please confirm:** [list top 5 candidates per ambiguous input with attributes].

### Ranked Accounts

*Sorted by composite descending. Engagement column `–` when redistributed.*

| # | Account | Tag | Tier | Composite | Fit | Intent | Trigger | Eng | Why now | ZI ID |

(Tier-A rows also carry an "↳ Engagement: ..." sub-line from §7.5.)

### Weights & Axes Used

```
fit:        [%]
intent:     [%]
trigger:    [%]
engagement: [%]   (redistributed if axis unavailable)
```

**Axes missing this run:** [list, or "none"].

### Recommended Actions per Tier

- **Tier A** — Route to AE for 1:1 outreach within 24h. Chain to `personalize-email`.
- **Tier B** — SDR sequence; ABM retargeting; cadence with the why-now as opener.
- **Tier C** — Monitor; re-score weekly.

### Iteration Options

1. Accept ranking; save filter+weight set.
2. Adjust weights.
3. Tighten thresholds.
4. Refilter.
5. Swap ICP.
6. Drill into one account.
7. Add accounts.

### Caveats (when relevant)

- **Ambiguous pending** — N accounts not yet scored.
- **Failed resolutions** — N inputs had no match.
- **Engagement axis unavailable** — surface per-account (`no CRM signal — consider cross-check`) for each tier-A row.
- **Signal depth** — `enrich_company_signals` returns the most recent signals per type (server-capped), so for very active accounts the intent/trigger axes reflect the most recent window rather than an exhaustive history.
- **Intent thin** — <3 topics resolved; intent directional.
- **Stale-signal cliff** — N accounts' best trigger >60d old.
- **Edge-of-recency** — N trigger events 80–90d.
- **GTM-context gap** — `icp` sparse; fit-axis precision reduced.

### Final Filter + Weight Set (on accept)

```json
{
  "icp": { /* GTM ICP or user override */ },
  "weights": {"fit": 45, "intent": 25, "trigger": 25, "engagement": 5},
  "tier_thresholds": {"A": 75, "B": 50},
  "intent_topics": ["..."],
  "use_case": "prospecting",
  "_meta": {"account_count": ..., "tier_distribution": {...}, "axes_missing": [...], "pass_count": ...}
}
```

### Chain Targets

- `personalize-email` per tier-A contact → grounded in the same "why now" signal.
- `build-list` to extend the universe.
- `find-similar` on a tier-A seed.
- `tam-sizer` with this filter set to confirm universe size.