ApexCloud Retention Operations
This skill covers the end-to-end workflow for building structured retention and
revenue-operations reports from the ApexCloud Retention Operations API. The
tasks are JSON-in/JSON-out: you read an answer template, fetch across several
REST endpoints, compute derived fields with deterministic precision, and return
exactly the completed template.
Quickstart
Read the answer template — it lives at input/payloads/answer_template.json
or wherever the prompt points you. Every slot that is null, 0, 0.0, "",
[], or false is unfilled and needs your computed value.
Identify the data you need. Walk the template keys and map each one to the
endpoints in references/api.md. The prompt usually tells
you which account IDs, date ranges, months, and quarters to use.
Fetch all API data. Parallelize fetches for independent resources (account
profiles, metrics, tickets, NPS, billing snapshots, A/R aging, opportunities,
HR summaries, events, churn exports). Use the base URL given in the prompt
(often <TASK_ENV_BASE_URL> — substitute it before calling).
Compute derived fields. Follow the rules in
references/conventions.md for ARR sourcing,
overdue-balance aging-bucket sums, ticket counting, NPS resolution, trend
classification, risk scoring, and link-status matching.
Select controlled vocabulary values. Every enum string the template
requires is defined in references/enums.md. Policy
codes are also listed there; choose the code that reflects the data sources
and approach you actually used.
Output only the completed JSON object. No markdown fences, no commentary —
just the raw JSON document.
Workflow Pattern
Every ApexCloud task follows the same rhythm:
Phase 1 — Parse the template
Do not skip this. The answer template tells you exactly what the task expects.
For each field, note its type and what endpoint(s) will supply its value. Build a
mental (or scratch) table: field → endpoint → computation. Ambiguous fields
(e.g. current_arr vs revenue vs recognized_revenue) are disambiguated by
reading references/conventions.md.
Phase 2 — Gather all data
Fetch everything you will need before you start computing. The API is read-only
and idempotent; there is no reason to sequence calls that do not depend on each
other.
Common endpoint families (full shape catalog in references/api.md):
| Endpoint |
Provides |
/api/accounts |
All accounts with ARR, tenure, segment, region, renewal date, CSM, aliases |
/api/accounts/{id} |
Single account profile |
/api/accounts/{id}/metrics |
Monthly recognized revenue, SLA%, NPS, usage %, seat count |
/api/accounts/{id}/tickets |
Support tickets with SLA flags, spam/duplicate markers |
/api/accounts/{id}/nps |
Individual NPS survey responses |
/api/billing/snapshots |
Quarterly billing ARR snapshots |
/api/finance/ar-aging |
A/R aging buckets (1-30, 31-60, 61-90, 90+ days) |
/api/opportunities |
CRM pipeline (open/closed, product lines) |
/api/hr/summary |
Headcount, unpaid claims by quarter and region |
/api/events/performance |
Event orders and revenue |
/exports/churn/train.csv |
Churn-model training rows |
/exports/churn/validation.csv |
Churn-model validation rows |
/exports/churn/candidates.csv |
Candidate accounts for churn ranking |
Phase 3 — Compute and fill
Work field by field. Use the conventions reference for every derived value
(ARR source, overdue balance, ticket hygiene, NPS, trends, risk scoring,
account matching). Select controlled enums deliberately — do not guess.
For risk scoring (risk_score, risk_level): build a composite from the
available signals rather than using a single axis. The conventions reference
provides a scoring rubric derived from the train evidence.
For ranking / ordering: the prompt will tell you whether to return a
top-N (e.g. top 5 risk accounts) or all accounts in board order.
Phase 4 — Validate and emit
Before returning:
- Currency values are to 2 decimal places.
- Percentage values are to 1 decimal place (except churn probabilities, which
are to 3 decimal places).
- Counts are integers.
- Every enum value is one of the allowed strings in
references/enums.md.
- Policy codes are selected from the set the template provides; default to the
middle option unless a specific data source or approach warrants a different
code.
- No extra keys exist beyond what the template defines.
When outputs differ from per-account API values
Some API endpoints return data at a different grain than the template expects:
Account-profile ARR vs billing-snapshot ARR: For retention tasks,
prefer the billing snapshot that matches the assessment date. The account
profile's billing_arr_current is a convenience field; the snapshots are
the authoritative quarter-close source. When the task includes
model_checks.uses_billing_arr_source, set it to true.
Metrics NPS vs NPS-endpoint NPS: The metrics endpoint returns a single
nps_score per month (the survey-aggregated value for that month, or null
when no survey completed). The per-account NPS endpoint returns individual
survey responses. For monthly QBR-style reporting, use the metrics endpoint.
For "latest NPS," use the most recent non-null value from the NPS endpoint
unless the task explicitly calls for metrics-endpoint values.
Tickets: clean_ticket_count means non-spam, non-duplicate tickets.
Always filter out is_spam: true and is_duplicate: true before counting.
Reference files
- references/api.md — Full endpoint catalog with field shapes.
- references/enums.md — Every controlled string value organized by category.
- references/conventions.md — Precision rules, computation recipes, risk-scoring rubric, and matching logic.
1---2name: apexcloud-retention-ops-313description: Use the ApexCloud Retention Operations API to build renewal risk queues, QBR metrics packets, receivables and pipeline reviews, churn model validation readouts, and high-touch retention action boards. Use this skill whenever the task mentions ApexCloud, retention operations, renewal risk, customer success QBR, A/R aging, churn predictions, NPS analysis, support health, billing snapshots, expansion pipeline, or a task-environment base URL pointing to a retention operations API — even if the user does not name ApexCloud explicitly. Also use it when you see enumerated answer template fields like risk_level, primary_action, reason_codes, policy_codes, or controlled source enums typical of CS ops tooling.4metadata:5 source: https://skillmd.com/skills/prism-shadow/apexcloud-retention-ops-316---78# ApexCloud Retention Operations910This skill covers the end-to-end workflow for building structured retention and11revenue-operations reports from the ApexCloud Retention Operations API. The12tasks are JSON-in/JSON-out: you read an answer template, fetch across several13REST endpoints, compute derived fields with deterministic precision, and return14exactly the completed template.1516## Quickstart17181. **Read the answer template** — it lives at `input/payloads/answer_template.json`19 or wherever the prompt points you. Every slot that is `null`, `0`, `0.0`, `""`,20 `[]`, or `false` is unfilled and needs your computed value.21222. **Identify the data you need.** Walk the template keys and map each one to the23 endpoints in [references/api.md](references/api.md). The prompt usually tells24 you which account IDs, date ranges, months, and quarters to use.25263. **Fetch all API data.** Parallelize fetches for independent resources (account27 profiles, metrics, tickets, NPS, billing snapshots, A/R aging, opportunities,28 HR summaries, events, churn exports). Use the base URL given in the prompt29 (often `<TASK_ENV_BASE_URL>` — substitute it before calling).30314. **Compute derived fields.** Follow the rules in32 [references/conventions.md](references/conventions.md) for ARR sourcing,33 overdue-balance aging-bucket sums, ticket counting, NPS resolution, trend34 classification, risk scoring, and link-status matching.35365. **Select controlled vocabulary values.** Every enum string the template37 requires is defined in [references/enums.md](references/enums.md). Policy38 codes are also listed there; choose the code that reflects the data sources39 and approach you actually used.40416. **Output only the completed JSON object.** No markdown fences, no commentary —42 just the raw JSON document.4344## Workflow Pattern4546Every ApexCloud task follows the same rhythm:4748### Phase 1 — Parse the template4950Do not skip this. The answer template tells you exactly what the task expects.51For each field, note its type and what endpoint(s) will supply its value. Build a52mental (or scratch) table: field → endpoint → computation. Ambiguous fields53(e.g. `current_arr` vs `revenue` vs `recognized_revenue`) are disambiguated by54reading `references/conventions.md`.5556### Phase 2 — Gather all data5758Fetch everything you will need before you start computing. The API is read-only59and idempotent; there is no reason to sequence calls that do not depend on each60other.6162Common endpoint families (full shape catalog in `references/api.md`):6364| Endpoint | Provides |65|---|---|66| `/api/accounts` | All accounts with ARR, tenure, segment, region, renewal date, CSM, aliases |67| `/api/accounts/{id}` | Single account profile |68| `/api/accounts/{id}/metrics` | Monthly recognized revenue, SLA%, NPS, usage %, seat count |69| `/api/accounts/{id}/tickets` | Support tickets with SLA flags, spam/duplicate markers |70| `/api/accounts/{id}/nps` | Individual NPS survey responses |71| `/api/billing/snapshots` | Quarterly billing ARR snapshots |72| `/api/finance/ar-aging` | A/R aging buckets (1-30, 31-60, 61-90, 90+ days) |73| `/api/opportunities` | CRM pipeline (open/closed, product lines) |74| `/api/hr/summary` | Headcount, unpaid claims by quarter and region |75| `/api/events/performance` | Event orders and revenue |76| `/exports/churn/train.csv` | Churn-model training rows |77| `/exports/churn/validation.csv` | Churn-model validation rows |78| `/exports/churn/candidates.csv` | Candidate accounts for churn ranking |7980### Phase 3 — Compute and fill8182Work field by field. Use the conventions reference for every derived value83(ARR source, overdue balance, ticket hygiene, NPS, trends, risk scoring,84account matching). Select controlled enums deliberately — do not guess.8586For **risk scoring** (risk_score, risk_level): build a composite from the87available signals rather than using a single axis. The conventions reference88provides a scoring rubric derived from the train evidence.8990For **ranking / ordering**: the prompt will tell you whether to return a91top-N (e.g. top 5 risk accounts) or all accounts in board order.9293### Phase 4 — Validate and emit9495Before returning:96- Currency values are to 2 decimal places.97- Percentage values are to 1 decimal place (except churn probabilities, which98 are to 3 decimal places).99- Counts are integers.100- Every enum value is one of the allowed strings in `references/enums.md`.101- Policy codes are selected from the set the template provides; default to the102 middle option unless a specific data source or approach warrants a different103 code.104- No extra keys exist beyond what the template defines.105106## When outputs differ from per-account API values107108Some API endpoints return data at a different grain than the template expects:109110- **Account-profile ARR vs billing-snapshot ARR**: For retention tasks,111 prefer the billing snapshot that matches the assessment date. The account112 profile's `billing_arr_current` is a convenience field; the snapshots are113 the authoritative quarter-close source. When the task includes114 `model_checks.uses_billing_arr_source`, set it to `true`.115116- **Metrics NPS vs NPS-endpoint NPS**: The metrics endpoint returns a single117 `nps_score` per month (the survey-aggregated value for that month, or `null`118 when no survey completed). The per-account NPS endpoint returns individual119 survey responses. For monthly QBR-style reporting, use the metrics endpoint.120 For "latest NPS," use the most recent non-null value from the NPS endpoint121 unless the task explicitly calls for metrics-endpoint values.122123- **Tickets**: `clean_ticket_count` means non-spam, non-duplicate tickets.124 Always filter out `is_spam: true` and `is_duplicate: true` before counting.125126## Reference files127128- [references/api.md](references/api.md) — Full endpoint catalog with field shapes.129- [references/enums.md](references/enums.md) — Every controlled string value organized by category.130- [references/conventions.md](references/conventions.md) — Precision rules, computation recipes, risk-scoring rubric, and matching logic.