Test and QA Plan

How the build is validated before go-live: three sequential test phases, edge cases, and production verification.

4 pagesPDF · Quality

FS-DOC-06Quality

Test and QA Plan

Accounts Receivable & Chasing

[YourCompany.com] · Finance Department · Prepared by FullSpec · [Today's Date]

This document defines the full quality assurance programme for the AR Chase Agent automation. It is written for the FullSpec build team and covers every test case required before the workflow goes live. Testing runs in three sequential phases: unit tests in sandbox, integration and edge-case tests in sandbox, and a final end-to-end production run. No phase begins until the previous phase has fully passed. The process owner joins Phase 3 only, to review and approve the production run before go-live is confirmed.

01Testing approach

Testing follows three phases in strict sequence. No phase begins until the previous phase has fully passed, all failures are resolved, and a pass record is logged. Phase 1 validates individual agent logic in isolation. Phase 2 confirms tool handoffs and system boundaries under normal and abnormal conditions. Phase 3 executes a complete end-to-end run in the production environment with real credentials and live data, observed by the process owner.

Phase 1 — Unit Testing

Individual agent logic and tool connections in isolation

Scope

AR Chase Agent internal logic: Xero poll parsing, reminder tier selection, Gmail send, Google Sheets read/write, Slack alert trigger, HubSpot note write. Each action tested independently with mock or sandbox data.

Environment

Sandbox only. Xero demo company, Gmail sandbox sending address, Google Sheets test tab, Slack test channel, HubSpot sandbox portal.

Credential rule

No production credentials permitted. All tokens and API keys must reference sandbox or test accounts exclusively. See Section 02 note.

Pass condition

All test cases T01 through T14 return the expected output with no errors. Any failure blocks progression to Phase 2.

Who runs it

FullSpec team only. Process owner not required for this phase.

Phase 2 — Integration and Edge-Case Testing

Agent handoffs, tool boundaries, and failure-mode handling

Scope

Full agent flow from Xero trigger through to HubSpot note write. All tool-to-tool handoffs, the 60-day escalation decision branch, and defined edge cases: duplicate invoices, missing data, tool unavailability, and human timeout scenarios.

Environment

Sandbox only. Integration tests use the same sandbox credentials as Phase 1. No production data is accessed at any point in this phase.

Credential rule

Sandbox credentials only. OAuth tokens must be scoped to test environments. Production OAuth tokens must not be loaded into the test workspace.

Pass condition

All integration test cases T15 through T22 pass and all edge-case tests E01 through E06 produce the defined safe-fail output. No unhandled exceptions permitted.

Who runs it

FullSpec team only. Process owner not required for this phase.

Phase 3 — End-to-End Production Test

Full live run with real credentials, observed by the process owner

Scope

One complete production run traced end to end: Xero poll, reminder history check, correct email sent via Gmail, Google Sheets row appended, Slack alert fired (where threshold met), Xero payment reconciled, and HubSpot note written. Error log verification also performed.

Environment

Production environment. Live Xero account, Gmail send-as address, live Google Sheets tracker, live Slack finance channel, live HubSpot portal. Internal test invoice and internal recipient address used.

Credential rule

Production OAuth tokens loaded. Xero connected app scoped to openid, profile, email, accounting.transactions, accounting.contacts. Gmail scoped to gmail.send and gmail.readonly. HubSpot scoped to crm.objects.contacts.write.

Pass condition

All production test cases T23 through T27 pass, including the final process owner review and sign-off (T27). Owner approval recorded in the FullSpec confirmation log.

Who runs it

FullSpec team executes cases T23 through T26. Process owner joins for T27 to review outputs and confirm approval.

Test and QA PlanPage 1 of 4

FS-DOC-06Quality

02Phase 1 — unit test cases

Sandbox credentials only throughout Phase 1. Do not load, reference, or test against any production Xero company, Gmail account, Google Sheets file, Slack workspace, or HubSpot portal. All test data must be synthetic. If a sandbox token expires during testing, refresh it within the sandbox environment before continuing. Do not substitute a production token as a workaround.

AR Chase Agent — unit test cases. The agent is the sole agent in this process. Test cases cover the happy path for each discrete action and the key failure modes for each tool connection.

Scenario

Input

Expected Output

Pass Criteria

T01

Xero poll detects one overdue invoice

Xero demo company: invoice INV-001, due date yesterday, balance $1,200, status AUTHORISED

Agent receives invoice payload: invoice ID, contact ID, amount due, due date, days overdue

Payload parsed without error; all five fields present and correctly typed

T02

Xero poll detects multiple overdue invoices in a single run

Xero demo company: 5 overdue invoices with varying ages (1, 7, 14, 30, 65 days)

Agent processes all 5 invoices sequentially; no invoice skipped

Exactly 5 execution branches initiated; no duplicate processing

T03

Xero poll returns no overdue invoices

Xero demo company: all invoices either paid or not yet due

Agent exits gracefully; no downstream actions triggered

Zero reminder emails sent; zero Sheets rows appended; no errors logged

T04

Google Sheets reminder history read — first reminder (count 0)

Google Sheets test tab: invoice INV-001, no existing rows

Agent determines reminder count = 0; selects Tier 1 (polite) template

Correct template ID returned; no read error

T05

Google Sheets reminder history read — second reminder (count 1)

Google Sheets test tab: invoice INV-002, one existing row with reminder_count = 1

Agent determines reminder count = 1; selects Tier 2 (firm) template

Correct template ID returned; read operation returns exactly one matching row

T06

Google Sheets reminder history read — third reminder (count 2)

Google Sheets test tab: invoice INV-003, two existing rows with reminder_count = 2

Agent determines reminder count = 2; selects Tier 3 (final notice) template

Correct template ID returned; escalation flag set to true

T07

Gmail sends Tier 1 reminder email

Invoice INV-001, contact email test@internal.com, amount $1,200, due date populated

Email delivered to sandbox inbox; subject line contains invoice number; body references amount and due date

Gmail API returns status 200; message ID recorded; no delivery error

T08

Gmail sends Tier 2 firm reminder email

Invoice INV-002, reminder count 1, days overdue 14

Firm-tone template sent; subject prefix 'Reminder:' present; correct invoice details in body

Correct template rendered; Gmail API 200 response; message ID logged

T09

Gmail sends Tier 3 final notice email

Invoice INV-003, reminder count 2, days overdue 30

Final notice template sent; subject prefix 'Final Notice:' present; escalation language in body

Correct template rendered; Gmail API 200 response; message ID logged

T10

Gmail send fails — authentication error

Gmail sandbox token revoked mid-run

Agent catches 401 error; logs failure to error log; does not retry; alerts via Slack error channel

Error entry written to log with timestamp; no unhandled exception; Slack error alert fired

T11

Google Sheets append — new reminder row written

Post-send: invoice INV-001, reminder count 1, sent timestamp, email subject

New row appended to test Sheets tab: client name, invoice number, reminder number, date sent, email subject, status = 'sent'

Row count increments by exactly 1; all 6 fields populated correctly

T12

Slack escalation alert fires when invoice is 60+ days overdue

Invoice INV-004, days overdue = 62, reminder count = 3

Formatted Slack message posted to test channel: contact name, invoice number, amount, days overdue, reminder count

Slack API returns 200; message visible in test channel; all 5 data fields present in message

T13

Slack escalation does NOT fire when invoice is under 60 days and fewer than 3 reminders

Invoice INV-005, days overdue = 20, reminder count = 1

No Slack message posted; agent continues to Sheets log step only

Zero Slack messages in test channel for this run; no escalation flag set

T14

HubSpot note written to contact record

HubSpot sandbox contact ID matching Xero contact; note text: reminder stage, date, invoice reference

Note created on HubSpot contact timeline; note body contains invoice number, reminder stage, and date

HubSpot API returns 201; note ID returned; all three fields present in note body

Test and QA PlanPage 2 of 4

FS-DOC-06Quality

03Phase 2 — integration and edge case tests

Phase 2 tests the full agent flow as a connected system, validating handoffs between each tool and confirming that the escalation decision branch routes correctly. Test IDs continue the sequence from Phase 1. A separate edge-case table follows, covering conditions outside the happy path that the agent must handle safely.

Scenario

Input

Expected Output

Pass Criteria

T15

Full happy-path run: Tier 1 reminder, no escalation

Xero: INV-010, 3 days overdue, balance $800. Sheets: no prior rows. Days overdue = 3, reminders = 0.

Tier 1 email sent via Gmail; Sheets row appended (reminder 1); no Slack alert; HubSpot note written

All 4 downstream actions complete in sequence; no step skipped; no error logged

T16

Full happy-path run: Tier 3 final notice, escalation triggered

Xero: INV-011, 35 days overdue, balance $3,500. Sheets: 2 prior reminder rows.

Tier 3 email sent; Sheets row appended (reminder 3); Slack escalation posted to test channel; HubSpot note written

All 5 actions complete; Slack message contains correct contact and amount; escalation flag confirmed in Sheets row

T17

Xero to Google Sheets handoff: reminder count correctly read from Sheets before email selection

Xero: INV-012, 10 days overdue. Sheets: 1 existing row for INV-012.

Agent reads Sheets row, determines count = 1, selects Tier 2 template; Tier 2 email sent

Sheets read completes before Gmail send executes; correct template confirmed by inspecting sent message subject

T18

Gmail to Google Sheets handoff: Sheets row is written only after Gmail confirms delivery

Invoice INV-013, Tier 1 send. Simulate delayed Gmail response (2-second mock delay).

Sheets append executes only after Gmail API returns 200; row timestamp is after send timestamp

Sheets row created; row timestamp within 5 seconds of Gmail send timestamp; no premature write

T19

Escalation branch: Slack fires and finance channel receives correctly formatted alert

Invoice INV-014, 61 days overdue, 3 reminders. Slack test channel configured.

Slack message contains: client name, invoice number, $amount, days overdue, reminder count. Finance channel tag present.

Message format matches specification; all 5 data fields rendered; channel is correct test channel ID

T20

Non-escalation branch: Slack step skipped, flow continues to HubSpot

Invoice INV-015, 10 days overdue, 1 reminder. Escalation condition not met.

Slack step bypassed; HubSpot note written; no Slack message in test channel

Zero new messages in test Slack channel; HubSpot note present; no branch error logged

T21

Google Sheets to HubSpot handoff: HubSpot note references same invoice logged in Sheets

Invoice INV-016, Tier 2 send. Sheets row appended with invoice number INV-016.

HubSpot note body contains 'INV-016' and matches the reminder stage recorded in Sheets

Invoice reference consistent across Sheets row and HubSpot note; no mismatch

T22

Run with multiple concurrent invoices: all processed without data collision

Xero: 4 overdue invoices simultaneously in sandbox — INV-020 through INV-023, mixed reminder counts

Each invoice processed independently; 4 separate Gmail sends; 4 Sheets rows appended; correct tier applied to each

No invoice data written to another invoice's Sheets row; 4 HubSpot notes created; no cross-contamination

Edge-case test table. These cases simulate conditions outside the normal operating path. The agent must handle each case with a defined safe-fail outcome and must not crash, skip silently, or write incorrect data.

Scenario

Input

Expected Safe-Fail Output

Pass Criteria

E01

Duplicate invoice returned by Xero poll in the same run

Xero poll returns INV-030 twice in a single response (simulated API quirk)

Agent detects duplicate by invoice ID; processes INV-030 only once; second instance discarded

Exactly one Gmail send, one Sheets row, one HubSpot note for INV-030; no duplicate

E02

Missing contact email on Xero invoice

Invoice INV-031 is overdue but linked Xero contact has no email address recorded

Agent logs 'missing contact email' error for INV-031; skips Gmail send; appends error row to Sheets; continues processing next invoice

Error row written to Sheets with status = 'error: no email'; Gmail not called; subsequent invoices still processed

E03

Google Sheets service unavailable during reminder history read

Mock Sheets API to return 503 on read request for INV-032

Agent retries Sheets read twice with 30-second backoff; on third failure, logs error and skips this invoice; moves to next

Two retry attempts logged with timestamps; error entry written after third failure; no Gmail send for INV-032

E04

HubSpot contact record not found for Xero contact

Invoice INV-033 linked to Xero contact whose email does not match any HubSpot contact

Agent completes Gmail send and Sheets log successfully; HubSpot step logs 'contact not found' warning; does not crash

Gmail and Sheets steps succeed; HubSpot returns 404; warning logged; workflow run marked 'partial success'

E05

Slack API unavailable at time of escalation alert

Invoice INV-034, 65 days overdue. Mock Slack API to return 500.

Agent logs Slack delivery failure; escalation error recorded in Sheets row status field; Gmail and HubSpot steps still complete

Slack failure entry in error log; Sheets row status = 'escalation failed'; Gmail and HubSpot steps confirmed complete

E06

Invoice tagged as disputed or on exception list in Google Sheets

Invoice INV-035 has status = 'disputed' in the Sheets tracker exception column

Agent reads disputed status; skips all reminder and escalation actions for INV-035; logs 'skipped: disputed' in Sheets

No Gmail send; no Slack alert; no HubSpot note for INV-035; Sheets row updated with skip reason

Test and QA PlanPage 3 of 4

FS-DOC-06Quality

04Phase 3 — end-to-end production test

Important before running Phase 3. Use an internal test email address as the recipient for all production test emails — do not send to any real client address. Create a dedicated test invoice in the live Xero account (marked clearly as a test record) and void or delete it immediately after the run is confirmed. Confirm with the process owner that the live Slack finance channel will receive a test alert and that team members are aware before the run begins. Do not proceed if production OAuth tokens for any tool are not yet authorised and scoped correctly.

Phase 3 uses the live production environment, real API credentials, and a controlled test invoice to verify the automation behaves correctly from end to end before the first real daily run. The FullSpec team executes cases T23 through T26. The process owner joins for T27, reviews all outputs, and records approval. That approval is the go-live gate.

Scenario

Production Input

Expected Output

Pass Criteria

T23

Production Xero poll detects test invoice and triggers run

Live Xero account: test invoice INV-TEST-001, due date set to yesterday, balance $100, contact email = internal test address

Automation platform triggers a new run for INV-TEST-001 within the scheduled poll window (daily, within 10 minutes of poll time)

Run log shows trigger fired for correct invoice ID; no other live invoices accidentally reprocessed

T24

Full production send: Tier 1 email received at internal test address via Gmail

INV-TEST-001, 1 day overdue, 0 prior reminders in live Sheets tracker

Tier 1 polite reminder email arrives at internal test inbox; subject contains 'INV-TEST-001'; body contains amount $100 and due date; from address is the authorised Gmail send-as address

Email received within 2 minutes of trigger; all merge fields correct; no raw variable placeholders visible; Gmail API message ID logged

T25

Live Google Sheets tracker row appended and HubSpot note created

Post-send for INV-TEST-001

Live Sheets tracker: new row with client name, INV-TEST-001, reminder number 1, today's date, email subject, status = 'sent'. HubSpot live contact record: note added referencing INV-TEST-001, reminder stage 1, and today's date.

Sheets row present and all 6 fields populated; HubSpot note visible on contact timeline within 3 minutes of email send; no duplicate rows

T26

Error log verification: run completes with no errors recorded

Full production run for INV-TEST-001 as above

Automation platform run log shows status = 'success'; zero error entries; all steps show green completion status

Run log inspected by FullSpec team; no warnings, retries, or failures recorded; execution time within expected range (under 90 seconds for single invoice)

T27

Process owner review and approval — final go-live gate

Process owner reviews: test email received, Sheets row, HubSpot note, run log, and Slack test alert (if escalation threshold test was included)

Process owner confirms all outputs are correct, templates are acceptable, and the automation is ready for live operation. Approval recorded in FullSpec confirmation log with name, date, and sign-off statement.

Owner approval confirmed and logged in FullSpec system. Test invoice INV-TEST-001 voided in Xero. Go-live authorised.

Once T27 is approved, the FullSpec team will void the test invoice in Xero, confirm the daily poll schedule is active, and monitor the first two live daily runs with the bookkeeper present as described in the Launch Plan. Any issues surfaced in those first two live runs are treated as post-launch defects and resolved by FullSpec at no additional cost within the standard support window. Contact the FullSpec team at support@gofullspec.com for any questions or issues during testing or post-launch monitoring.

Test and QA PlanPage 4 of 4

Test and QA Plan

Test and QA Plan

01Testing approach

02Phase 1 — unit test cases

03Phase 2 — integration and edge case tests

04Phase 3 — end-to-end production test

More documents for this process