Case 25
Map the Workflow Before Automating
Scenario
Your accounts payable team handles supplier invoices by email, spreadsheet and manual system entry. Leadership wants to know what could be automated.
Why this framework
Uses PIN: what it's for, the context and source of truth, and a narrow ask. Enough for most everyday work.
Example context
Process notes (accounts payable, Larkfield Facilities Ltd): Invoices arrive by email to a shared inbox (about 900 a month) and some by post (about 60). Clerk Amira opens the inbox each morning, saves PDFs to a shared drive, and keys supplier, date, amount and PO number into the accounting system, Sage 200. She matches to purchase orders by looking up the PO in a separate spreadsheet kept by procurement. Mismatches are emailed to the budget holder, who often replies after several days. Approved invoices are coded by Amira and reviewed by the finance supervisor, Colin, who approves payment runs every Thursday. Duplicate invoices are caught only by memory or by Colin's spot checks. Team description of a typical day: Amira spends about 4 hours on keying, 1 hour chasing approvers, 1 hour answering supplier queries by phone. Colin spends about 2 hours on Thursdays reviewing payment runs. Peak load at month end, when about 30% of invoices arrive. No scanner or OCR is currently used.
Prompt
Purpose: Map our current invoice handling process so we can see where automation might help.
Context: Use the attached process notes and the team's description of a typical day. Treat these as the current state; do not assume system features we have not described.
Narrow the ask: A numbered step list with who does each step, the system used, and pain points. Then mark each step as good to automate, automate with a person checking, or keep manual, with one line of reasoning each.
Materials (pasted here in place of the attachments):
Process notes (accounts payable, Larkfield Facilities Ltd): Invoices arrive by email to a shared inbox (about 900 a month) and some by post (about 60). Clerk Amira opens the inbox each morning, saves PDFs to a shared drive, and keys supplier, date, amount and PO number into the accounting system, Sage 200. She matches to purchase orders by looking up the PO in a separate spreadsheet kept by procurement. Mismatches are emailed to the budget holder, who often replies after several days. Approved invoices are coded by Amira and reviewed by the finance supervisor, Colin, who approves payment runs every Thursday. Duplicate invoices are caught only by memory or by Colin's spot checks.
Team description of a typical day: Amira spends about 4 hours on keying, 1 hour chasing approvers, 1 hour answering supplier queries by phone. Colin spends about 2 hours on Thursdays reviewing payment runs. Peak load at month end, when about 30% of invoices arrive. No scanner or OCR is currently used.Example output from ChatGPT
Larkfield Facilities Ltd - Invoice Handling Process Map
Scope: Current-state accounts payable process, based only on the supplied process notes and team's typical-day description.
Objective: Identify automation opportunities while retaining human oversight for exceptions, financial decisions and payment approvals.
-
Current-state process flow
-
Invoice received Shared email inbox / post 2–4. Save, key and match Amira - Shared drive, Sage 200, PO spreadsheet
-
PO details match? Yes Proceed to coding No Email budget holder; await reply 6–7. Code and review Amira / Colin
-
Approve Thursday payment run Colin - Sage 200 / payment process not fully described Supplier queries and duplicate checks occur alongside this flow. The exact routing of exceptions and payment execution are not described.
-
Numbered process steps and automation assessment
1 Receive invoices Amira (inbox); postal handler unspecified · Shared email inbox / post
Pain point: Two intake channels; approximately 60 paper invoices monthly.
Good to automate
Automate email intake, registration and tracking. Postal invoices still need manual handling until digitised.
2 Open inbox and save PDFs Amira · Email / shared drive
Pain point: Daily manual opening, saving and filing; risks of missed or repeated handling.
Good to automate
Automatically save attachments using consistent file names and an intake log.
3 Enter invoice data Amira · Sage 200
Pain point: Manual entry of supplier, date, amount and PO; part of four daily hours spent keying.
Automate with person checking
Use document extraction with validation, particularly for missing fields and unusual values.
4 Look up and match purchase orders Amira · Procurement PO spreadsheet / Sage 200
Pain point: Separate spreadsheet lookup; manual comparisons and possible data discrepancies.
Automate with person checking
Match by PO number and agreed rules; route incomplete or inconsistent records for review.
5 Resolve mismatches and chase responses Amira / budget holder · Email
Pain point: Budget holders can take several days; Amira spends about one hour daily chasing.
Automate with person checking
Automate routing, reminders and ageing alerts; budget holders retain exception decisions.
6 Code approved invoices Amira · Sage 200
Pain point: Manual coding adds effort and potential classification errors.
Automate with person checking
Suggest codes from approved rules or prior examples, then review uncertain classifications.
7 Review invoices and check duplicates Colin; Amira supports invoice processing · System for review not fully specified; Sage 200 is used for accounting
Pain point: Duplicates depend on memory or spot checks, creating a risk of missed duplicates.
Automate with person checking
Run duplicate-detection rules across invoices; Colin investigates flagged cases.
8 Review and approve weekly payment run Colin · Payment-run system not explicitly specified
Pain point: Approximately two hours of review each Thursday; timing creates a weekly approval cycle.
Keep manual
Retain Colin's payment approval authority; automate preparation and exception reporting around the decision.
9 Answer supplier queries Amira · Telephone; invoice status records not described
Pain point: Approximately one hour daily answering calls, potentially interrupting processing.
Automate with person checking
Offer validated status information for routine queries; keep disputes and sensitive cases with staff.
These are proposed classifications, not confirmation that Sage 200 or any current system supports the suggested functions.
- Current workload and bottlenecks
Invoice volume 960 Per month, approximately
Month-end concentration 288 Invoices (30% of monthly volume)
Amira's stated workload 6 hrs Per typical day across three activities
Colin's payment review 2 hrs Each Thursday
Amira's stated daily workload Keying 4h Chasing 1h Queries 1h
Based on the six hours explicitly described, not Amira's complete working day. Invoice keying accounts for 67% of that time. 4. Priority automation opportunities
| Priority | Opportunity | Expected benefit |
|---|---|---|
| 1 | Invoice capture, data extraction and validation | Reduce repetitive entry, the largest stated workload |
| 2 | PO matching and automated exception chasing | Reduce spreadsheet lookups and approval delays |
| 3 | Duplicate invoice detection | Strengthen controls before payment |
| 4 | Suggested accounting codes | Reduce coding effort and inconsistency |
| 5 | Supplier status lookup | Reduce routine phone interruptions |
These benefits are [Inference] based on the current process. No time savings, costs or error reductions have been measured.
- Checks before designing the future process
Confirm whether Sage 200 has an available, approved integration method for invoice data, PO matching and payment controls.
Establish how postal invoices will be digitised, given that no scanner or OCR is currently used.
Define matching tolerances, coding rules, exception ownership and approval authority.
Measure duplicate frequency, mismatch volumes, actual cycle times and supplier-query categories.
Confirm who performs invoice review, how payment files are generated and whether segregation-of-duties rules apply.
Overall assessment: The strongest initial opportunity is to automate invoice capture and routine processing while keeping people responsible for exceptions, coding validation and payment approval. This addresses the heaviest stated workload without assuming that financial judgement can be delegated safely to automation.
