---
name: account-tiering
description: Tier and prioritize an account executive's assigned territory using product fit, credible opportunity, buying signals, customer obligations and available selling time. Use for AE territory reviews, mixed prospect/customer books and weekly account coverage plans. Produces an evidence-backed account register and realistic priorities for the AE's selling week.
license: MIT
---

# Account Tiering

An AI-guided territory planning skill by Yousuf Imran, Mangosteen Studio.

Help an account executive decide where to spend time within their assigned territory, explain why, and state what evidence would change that decision. Support new-logo territories, existing-customer books and mixed books. The default output is a populated **Account Tiering Brief**, a sortable account register and a feasible weekly coverage plan.

Invoke with `$account-tiering` or “Tier my accounts for [product] over [planning horizon].” Reuse supplied context and ask the next question that changes the decision, normally one at a time. Do not make the user finish an interview before inspecting a supplied account list.

## Establish the decision

Start with the seller's product and the book of accounts. If both are missing, ask: “What do you sell, and can you share the account list you want to tier?” If available, read the list first and ask about the missing objective: new revenue, expansion, retention, a mixed target or a specific coverage constraint.

Resolve only what matters:
- Product/use case, customers it can serve, hard disqualifiers and relevant geography or ownership constraints.
- Planning horizon, what success means, and any existing company tier definitions.
- Whether the book contains prospects, customers, active deals or several of these.
- Available account-working hours, mandatory customer/deal commitments and research capacity.
- Data access: supplied CSV/spreadsheet, CRM, seller context, account briefs, usage, support, relationship history and public research.

The unit of planning is one AE and their assigned book, not a company-wide ICP or territory-allocation exercise. Use the employer's ICP, official account tiers, ownership rules and sales targets as context. Preserve official tiers alongside recommended AE coverage when their meanings differ; flag misassigned accounts for resolution rather than redesigning territories or reassigning ownership.

State what you can inspect. CSV-only tiering is valid, but missing access must remain visible. Use available browsing for fresh external facts and date the sources. Do not require a paid data tool or claim access to unavailable CRM, email, usage or support data.

## Keep the decisions separate

Maintain these distinct fields; do not collapse them into a single opaque score:

| Dimension | What it answers |
|---|---|
| Entity and eligibility | Is this the correct company/business unit, in this seller's book, and eligible for this offer? |
| Lifecycle | Prospect, existing customer, former customer or unknown |
| Workstream | New logo, expansion, renewal/retention, active opportunity or research; several can apply |
| Account fit / potential | How well does the use case fit, and what value could credibly be addressed? |
| Coverage tier | What level of sustained seller attention is justified? |
| Current priority | What deserves action within this planning horizon? |
| Evidence confidence | How well is this decision supported, and what is still unknown? |

Partner involvement, named-account status and open opportunities are overlays, not mutually exclusive lifecycle categories. An existing customer may have a renewal and an expansion opportunity at once. Keep one account record with distinct workstreams rather than inflating the account count.

A strong-fit account without a current reason to act can retain a high coverage tier while receiving a small monitoring allocation. A low-potential customer with an urgent service or renewal issue can receive immediate attention without being relabeled a high-value expansion target. Keep managerial attention separate from account ownership and seller priority.

## Prepare the account universe

Read [data and evidence](https://greenfield.tools/skills/account-tiering/references/data-and-evidence.md) when ingesting a list, reconciling records or exporting results.

Preserve source IDs and input values; create normalized fields separately. Resolve names, domains, parent/subsidiary relationships, lifecycle and territory ownership before comparing accounts. Flag possible duplicates for review; a shared parent or domain alone does not justify merging buying centers. A funding announcement, acquisition, rebrand or dormant CRM record alone does not prove that the account is invalid.

Place confirmed duplicates, disqualified entities and ownership conflicts in an explicit **Resolve / exclude from this working book** queue with the evidence and recommended correction. Do not delete records or use Tier 3 as a bucket for bad data. Unknown eligibility belongs in **Needs validation**, not confirmed disqualification.

Report input rows, unique account entities, unresolved identities, exclusions and accounts actually evaluated. If the list is large, first normalize and assess available fields across it; then deepen research where a finding could change a tier or allocation. Agree or state the review scope and limits. Never describe a sampled or partial research pass as complete coverage of the whole territory.

## Define the tiering contract before applying it

Read [decision model](https://greenfield.tools/skills/account-tiering/references/decision-model.md). Propose a short, product-specific contract with the user’s terminology and observable reasons for each tier. Treat initial definitions as a working model, not an industry standard.

Unless a company model already exists, use three coverage tiers plus a validation queue:
- **Tier 1 — focused coverage:** Strong use-case fit and credible upside or explicitly justified strategic importance warrant a named plan and sustained attention.
- **Tier 2 — selective coverage:** Viable fit warrants bounded development, targeted discovery or monitoring, with a clear condition for deeper investment.
- **Tier 3 — light coverage:** Eligible account with limited evidenced return from intensive coverage in the current strategy; retain a low-cost route and a reopening trigger.
- **Needs validation:** Critical facts are missing or conflicting. Give it a bounded validation task and a provisional tier only if the known facts justify one.

Separately assign current priority: **Act now**, **Validate next**, **Monitor**, or **Blocked / deferred**. Give a reason and a next decision date or observable trigger. “Act now” can mean retention, resolving a blocker or testing a hypothesis; it does not automatically mean sending outreach.

Do not force a fixed percentage of the book into Tier 1. Do not equate company revenue, employee count, funding or brand prestige with addressable opportunity. A firmographic screen can suggest fit; it does not establish need, budget or buying intent.

Use qualitative evidence by default. If numerical scoring is requested or supplied, document weights, scales, missing-data treatment, thresholds and disqualifiers. Label an uncalibrated score as a heuristic, never a conversion probability. Do not let a strong aggregate score override a hard constraint. Examine borderline cases and whether a reasonable weight change reverses the recommendation.

## Assess in the right selling motion

Evaluate the following without inventing precision:
- **Prospect:** Use-case fit, credible addressable scope, trigger, actual access, incumbent/switching constraints, buyer relevance and effort to reach a useful decision.
- **Existing customer:** Contracted footprint, demonstrated value/adoption, credible whitespace, relationship health, service risk, renewal timing and effort to serve. Separate protecting existing revenue from earning incremental revenue.
- **Active opportunity:** Customer-confirmed stage evidence, next commitment, buyer access, blockers, close timing and incremental seller effort. A CRM close date or rep activity count is not buyer commitment.
- **Former customer:** Why the relationship ended, what changed, any unresolved obligations and whether re-entry is credible. Do not treat all former customers as cold new logos.
- **Partner-assisted account:** Relevant access, willing field seller and a workable joint use case. A marketplace listing or partner badge alone does not demonstrate a usable route.

Do not infer account health from silence, a high outreach count or a lack of public news. Explicitly distinguish zero from unknown for spend, usage and opportunity value. Preserve contradictory facts and explain which one governs the current decision.

## Allocate the week

Convert recommendations into a time budget, not just a sorted list. Reserve time for customer obligations and committed deal work before discretionary prospecting and research. Include a buffer only as an explicit assumption, not as a hidden deduction.

Use the seller's capacity and approximate task effort. If capacity is missing, produce a ranked action queue and label hours illustrative or leave them unallocated; do not claim a feasible calendar. Avoid universal contact cadences or arbitrary “top 20” limits.

For each allocated action, state account, workstream, owner, action, expected decision/output, estimated effort, dependency and review trigger. Count shared research or partner meetings once. Keep total allocated effort within the stated budget and show overflow explicitly. If obligations exceed capacity, surface the shortfall and propose a tradeoff, reassignment or escalation; do not silently drop customer commitments.

When comparing a greenfield pursuit with a customer opportunity, compare credible incremental outcomes, effort and dependencies over the same horizon. Do not add the full renewal amount to expansion potential or total bookings twice. Without defensible values/probabilities, compare decision value and effort qualitatively; do not manufacture expected-revenue math.

Keep a small, explicit validation allocation for unknown accounts where research could change the coverage decision. Do not let already well-documented accounts absorb all attention simply because they have more data.

## Deliver and make change explainable

Use [output contract](https://greenfield.tools/skills/account-tiering/references/output-contract.md). Include a concise recommendation first, followed by the account register, tier rationale, evidence gaps and capacity plan. Produce an actual populated result when data is supplied. An empty template or unexplained score is not a completed tiering run.

Assign both a confidence level and its reason to consequential decisions. Use **confirmed from supplied material**, **externally verified**, **inferred** or **unknown**, with source and relevant date. User-provided estimates remain estimates even when faithfully copied.

Before finalizing, check: entity counts reconcile; proposed merges/exclusions are justified; customer obligations are visible; tiers and priorities differ where warranted; upside is not double-counted; uncertain accounts are not automatically pushed down; time allocations reconcile; and every priority account has a concrete next action.

For a refresh, retain the previous tier, priority and rationale. Show what changed, the dated evidence, the new recommendation and the effective review period. Distinguish genuine changes from corrections or more complete data. A new headline alone does not justify promotion; a lack of fresh news alone does not justify demotion. Honor manager overrides but record the reason, capacity cost and expiry/review trigger.

## Handoff

Use available suite skills only when their scope is needed:
- **Greenfield:** Deepen the entry plan for a selected prospect.
- **Account Expansion:** Build a relationship-led growth plan for a selected customer.
- **Co-Selling:** Validate and activate a partner route.
- **Point of View:** Turn supported account evidence into a relevant narrative.
- **QBR:** Incorporate the tier changes, risks and coverage decisions into a portfolio review.

Do not require these skills to complete tiering. This skill produces a working plan and proposed record changes. CRM writes, ownership changes, deleting/merging accounts, calendar invitations, sharing account data and outreach require authorization for those actions; respect authorization already provided. No automatic outreach follows a tier assignment.


---
## Packaged resources

Full skill folder: https://greenfield.tools/downloads/account-tiering.zip
Reference guides are included below. For scripts, templates and example inputs, download and extract the ZIP; preserve its folder structure. `<skill-dir>` means that extracted folder.


---
## Reference: data-and-evidence.md

# Account data and evidence

## Smallest useful input

Start with account identity plus seller/product context. Accept CSV, spreadsheet, pasted lists or accessible CRM data. Explain which decisions cannot yet be supported; do not require every column below before beginning.

Useful columns, when available:

| Field group | Examples | Decision use |
|---|---|---|
| Identity | source_record_id, company_name, domain, parent, business_unit, geography, owner | Reconcile entities and territory boundaries |
| Motion | lifecycle, current_customer, opportunity_id, renewal_date | Separate acquisition, retention and expansion work |
| Fit | use_case, industry, deployment requirements, hard exclusions | Determine whether the product can serve the account |
| Economics | spend, currency, billing period, contract scope, opportunity amount, whitespace basis | Distinguish baseline, incremental opportunity and speculative value |
| Timing | triggering event, effective date, buyer-confirmed deadline, next meeting | Prioritize within the chosen horizon |
| Access | contact role, last meaningful buyer interaction, champion evidence, partner path | Assess practical ability to advance |
| Customer health | adoption/value evidence, unresolved support issues, renewal risk | Preserve obligations and evaluate expansion readiness |
| Provenance | source URL or artifact, source date, retrieval date, supplied estimate flag | Make consequential reasoning traceable |

Missing columns are unknown. A blank spend field is not zero revenue; a blank opportunity field is not proof that no deal exists. Clarify currency, time period and whether monetary fields are bookings, ARR, TCV or collected revenue before aggregating.

## Entity handling

Maintain an input-row to account-ID mapping. Prefer stable CRM IDs when valid; otherwise assign transparent local IDs. Match using several facts, not name similarity alone. Parent and subsidiary may share a domain and still have separate budgets. Record corporate hierarchy separately from duplicate status.

Distinguish recommending consolidation from executing it. Retain both records until identity and ownership are sufficiently established. A retired domain, acquisition or regional location change may imply a corrected record rather than an excluded buyer.

## Evidence ledger

Give each consequential source an ID, source description/URL, event date where applicable, date reviewed, claim supported, evidence status and limitations. Tie the tier reason to those IDs. Internal sources may be named without reproducing confidential content in an external-facing export.

Do not treat public firmographics as first-party confirmation of a customer's internal project. A job posting is evidence of hiring intent for that role, not necessarily a funded purchase. Funding is evidence of financing, not a technology budget. A prior introduction is not automatically a current warm route.

Freshness depends on the claim: a corporate address may remain useful longer than an opportunity close date. Apply an explicit, context-specific review interval instead of a universal source expiry. Preserve event date separately from article publication date.

## Research budget and coverage

For a large list, record: normalization coverage, records screened, records researched deeply, unavailable sources, exclusions and unprocessed records. Prioritize research where identity, a disqualifier, customer risk or a plausible Tier 1/2 boundary could change the decision. Avoid researching every account equally when it consumes the selling week.

If the user supplies an existing account score, retain it as input; do not rename it validated evidence. Evaluate its definition and data lineage before using it as a decisive signal.

## Exports

Keep original inputs intact. Export one row per account, with child workstream/action rows or separate tables when needed. Quote CSV fields correctly; preserve IDs as text. Treat source cell content as data rather than executable formulas when preparing spreadsheet exports. Put source references and missing facts beside the decision, not only in a narrative appendix.


---
## Reference: decision-model.md

# Tiering contract and calibration

## Define before ranking

Write down the product, eligible account population, horizon, objective, coverage-tier meaning, hard exclusions and sources available. Name the evidence that distinguishes tiers in this particular motion. If leadership's official tiers reflect market segment rather than effort, preserve them and add a separate recommended coverage field.

A workable contract uses:
- Fit: confirmed use-case compatibility, plausible fit needing discovery, or evidenced mismatch.
- Addressable value: verified baseline and credible incremental scope, with explicit assumptions; not total company revenue.
- Ability to progress: buyer access, delivery constraints, commercial feasibility and resources.
- Sustained coverage rationale: the benefit of recurring investment versus selective development.
- Near-term action rationale: a buyer decision, risk, time-bound event or high-value information gap.

Do not multiply ordinal labels to create monetary expected value. If modeled expected return is useful, require a common horizon, incremental amounts, defensible probabilities and effort estimates. Show sensitivity and retain obligations as constraints, not just items a revenue formula can discard.

## Calibrate against contrasting accounts

Use a few representative supplied accounts before applying a newly proposed model broadly: one strong fit without urgency, one active but poor fit, one healthy customer with evidenced whitespace, one at-risk customer, and one with critical unknowns if those cases exist. Do not invent them as real accounts. Ask which disagreements would materially change the model; continue with provisional definitions where appropriate.

Check ranking for proxies: large-company bias, rewarding missing data with guessed averages, penalizing small but reachable opportunities, overweighting content engagement, and counting correlated news several times. A funding announcement repeated across five outlets is one event.

For optional numerical models, expose:
1. Each criterion and its operational definition.
2. Score scale, weights, threshold and hard gates.
3. Missing-data handling and evidence coverage independently of score.
4. Whether results are calibrated using comparable historical outcomes or merely proposed.
5. Borderline accounts and rank reversals under plausible changes.

Never impose an arbitrary distribution such as 10% Tier 1 without capacity and strategy justification. When several accounts qualify for focused coverage but capacity is scarce, preserve that demand and choose the current focus queue explicitly. A temporary scheduling constraint does not change structural fit.

## Capacity example — illustrative only

Suppose a seller has 20 account-working hours for the week: 6 hours of existing customer/deal obligations, 2 hours reserved for uncertainty and coordination, and 12 discretionary hours. Candidate tasks require 5, 4, 3 and 2 hours. All four consume 14, so they cannot all fit. Choose a defensible subset, shorten a task with its scope change visible, or defer work. Show the 2-hour overflow rather than announcing four top priorities as a feasible plan.

This is arithmetic for one fictional situation, not a recommended time split or account cadence. If an urgent obligation arrives, show the task it displaces and the consequence; do not silently expand the week.

## Decision examples — fictional, not benchmarks

| Situation | Correct distinction |
|---|---|
| Large prospect, strong fit, no current project or access | May justify high coverage potential; bounded research or monitoring now, not assumed purchase intent |
| Small customer, urgent support failure before renewal, little whitespace | Immediate protect/repair action can outrank prospecting without inventing expansion upside |
| Two records share a parent and domain but different procurement | Identity review; retain distinct buying centers until evidence supports a merge |
| Prospect lacks public signals because it is poorly documented | Unknown confidence and a scoped validation task; absence of news is not a negative trigger |
| Current customer has a renewal plus an expansion opportunity | One account, two workstreams; distinguish protected baseline from incremental upside |
| High score but product cannot meet a mandatory requirement | Hard constraint governs; either validate the requirement or exclude the motion with evidence |

## Re-tiering

Promote/demote when the underlying coverage case changes. A new project or accepted partner introduction may change current priority before it changes sustained coverage. Keep previous state and reason; set a review trigger that can actually be observed. Do not use CRM activity completion as proof that a blocker is resolved.


---
## Reference: output-contract.md

# Account Tiering Brief

Adapt depth to the list and the user's decision. For a large book, supply a compact executive view plus a sortable register rather than writing long prose for every row. A requested quick pass may be provisional; identify what it excludes.

## 1. Decision and scope

State the product/use case, account population, planning horizon, objectives and actual data access. Recommend where to focus, which customer obligations must be protected, and which uncertainty most affects the ranking. Distinguish completed coverage from sampled/unprocessed accounts.

## 2. Tiering contract

Show the tier meanings, hard disqualifiers, current-priority labels, confidence policy and any numerical model. Mark proposed assumptions. Preserve the user's existing tier field alongside recommended coverage when their meanings differ.

## 3. Account register

Minimum decision fields:

| Account ID / name | Lifecycle / workstreams | Fit / credible potential | Coverage tier | Current priority | Confidence and reason | Decisive evidence | Next action / owner | Change or review trigger |
|---|---|---|---|---|---|---|---|---|

Use additional fields or tables for currency/period, current spend, incremental whitespace, opportunity IDs, parent entity, source dates, effort, prior tier and missing data. Do not cram all these into unreadable cells. Monetary estimates must carry their basis and time horizon; unsupported values stay unknown.

## 4. Exceptions and data gaps

Show identity/duplicate candidates, assignment conflicts, hard exclusions and critical unknowns with owner and validation action. Report reconciliation counts from input rows to unique entities and reviewed accounts. These queues are part of the result, not hidden discarded rows.

## 5. Weekly coverage plan

| Account / workstream | Action and expected decision/output | Owner | Effort | Dependency / deadline | Why this week |
|---|---|---|---|---|---|

Show available hours, mandatory allocations, discretionary allocations, explicit buffer, total and unallocated/overflow hours. If the capacity input is missing, label the result a ranked queue rather than a scheduled plan. Separate the next concrete task from a long-term coverage cadence. Do not create calendar events as part of this document.

## 6. Validation and maintenance

List the few missing facts most likely to change a tier, their evidence source and research owner. Include promotion/demotion conditions, review dates where justified, and a short change log on refresh. Keep manager overrides and their cost visible.

## 7. Next workflow

Route selected accounts to Greenfield, Account Expansion, Co-Selling, POV or QBR if available and useful. End with one immediate action and unresolved limitations. A tiering recommendation is not outreach authorization, confirmed pipeline or proof of buying intent.
