Run a 30–90 minute scripted interview plus two quick data pulls. Convert headcount, median minutes per transaction, volume, hourly cost, error/rework and delay into four cost components — direct labour, overhead, rework, delay/opportunity — and deliver one annual cost, a confidence band and a prioritized automation shortlist.
- A 30–90 minute scripted audit produces a conservative, finance‑grade annual cost you can take to procurement.
- Price the work across the whole team; the sum of payroll forces a decision faster than one person's timesheet.
- Include rework, delay/opportunity loss and single‑person risk to avoid inflated ROI claims.
- Turn the annual cost into simple funding rules: payback months, first‑year cash saving and a pilot scope.
- A one‑page deliverable with a confidence band and three prioritized options is enough to sign off a pilot.
Signs this is happening in your business
- Three or more people each spend multiple hours weekly on the same manual task and complain about time.
- A single person holds the exceptions and handles escalations or the manager’s urgent calls.
- Items sit in a queue with timestamps and routinely miss SLA targets.
- Work is rekeyed between systems and errors trigger customer callbacks or credit notes.
- Estimates of task time vary by person and nobody has pulled the activity report.
- Finance asks for a cost estimate and you can only offer headcount and a vague efficiency percentage.
- Pilots stall because stakeholders disagree on the definition of 'done' and expected savings.
If several of these are true at the same time, the problem is usually the process rather than the people running it.
What this is costing you a year
(Team hours per week x hourly cost of employees x 52) = annual payroll loss
- People doing this work4
- Hours each per week6
- Hourly cost of employees$45
- Weeks52
The annual payroll cost of four people each spending six hours per week on this manual process.
Find the number before you spend anything
Run the formula above on your three most repetitive processes. If any of them clears a year of a salary, that is the one to bring to a call.
Quick audit script: one 30–90 minute session that produces a number
Run a tightly scripted 30–90 minute session to produce a conservative annual cost, a confidence band and a prioritized automation shortlist you can fund. The session is interview plus two data pulls so you leave with the inputs to calculate direct labour, overhead, rework cost and delay/opportunity cost.
- Total duration: 30–90 minutes (simple processes 30, multi‑hand processes 60–90).
- Participants: process owner, two practitioners, optional finance or HR rep.
- Deliverable: one‑page cost summary, confidence band, prioritized automation shortlist.
What to ask and which reports to pull first
Collect five input categories: who does the work, median minutes per transaction, transaction frequency, hourly cost, and what exceptions look like. Pull two reports: activity counts and time‑stamped queue entries for the busiest month so you can cross‑check volumes and delay.
- Headcount actively doing the task (exclude supervision).
- Median minutes per transaction — time one live example together.
- Frequency: transactions per day/week/month for the last quarter.
- Hourly cost or salary band for each person doing the work.
- Overhead multiplier for payroll taxes and benefits.
- Error/rework rate and minutes per exception from logs or recent tickets.
- Visible delays: average time from handoff to completion and any penalties.
Who to invite and a minute‑by‑minute timetable
Invite the process owner, two people who do the work, and a finance or HR representative if available. Keep the session scripted, record answers, and time one transaction live to avoid gut estimates.
- 0–5 mins: state purpose, confirm roles, show one‑page audit template.
- 5–15 mins: process owner explains end‑to‑end and single‑person risks.
- 15–30 mins: time one sample transaction together and record minutes and decisions.
- 30–45 mins: practitioners confirm volume and exception rates; finance confirms salary bands.
- 45–60 mins: pull reports and cross‑check counts; reconcile discrepancies.
- 60–90 mins: capture qualitative impacts and agree next steps for the one‑page deliverable.
Exact inputs to collect and where to source them quickly
Collect concrete inputs only and record who provided each value and their confidence. Source headcount from HR, minutes by timing a live example, frequency from system reports, and salary from payroll. If a report is missing use the most recent manual log or a ticket export and mark confidence lower.
- People doing this work — HR headcount or team roster.
- Minutes per transaction — time a sample with the practitioner.
- Frequency — activity counts from the operational system for the last quarter.
- Hourly cost — payroll report or salary band divided by work hours.
- Overhead multiplier — finance confirmation or rule of thumb.
- Error/rework rate — tickets marked rework or recent log validation.
- Customer/finance impact — escalations, credit notes or delayed invoices tied to the process.
A simple conservative formula that produces the annual cost
Convert inputs into four components: direct labour, overhead, rework/exception cost and delay/opportunity cost. Sum them for the annual process cost and show a conservative remaining cost after automation; prefer median times, assume higher rework minutes and include single‑person redundancy.
- Direct labour = People × Minutes per transaction × Transactions per year × Hourly cost.
- Overhead = Direct labour × Overhead multiplier (subtract 1 if multiplier given as total).
- Rework cost = Transactions requiring rework × Extra minutes × Hourly cost.
- Delay/opportunity cost = Estimated lost revenue or credits due to delays (use finance’s lowest credible estimate).
If you price a process on one person's time you will always underfund the fix; price it across the team and the number forces action.
Direct labour — where the time is actually spent
Direct labour is the total paid minutes the team spends on transactions. Measure it by timing a representative transaction, multiply by transactions per year and by the hourly cost of everyone who handles the task to produce the base payroll cost you can challenge in finance.
Overhead — how to convert payroll into full cost
Apply the overhead multiplier finance uses for benefits, taxes and facilities. Multiply the direct labour figure by the multiplier to capture total payroll burden; if finance gives a total multiplier subtract 1 to calculate the extra cost over salary.
Rework and exceptions — the hidden extra minutes
Count transactions that require additional handling and time each exception. Multiply exception count by extra minutes and hourly cost. Use logs or recent tickets to validate the rate rather than relying on memory; rework often adds 20–40% to effort.
Identify your biggest operational bottleneck
Most owners already know which process is the problem. What they do not have is the annual figure attached to it. That is a 15-minute conversation, not a project.
Delay and opportunity — quantify customer and revenue impact
Translate delays into a dollar effect: missed revenue, credits or premium penalties. Use finance’s conservative estimate for lost revenue or the cost of missed service levels and add that to the annual cost so procurement sees the full impact.
Worked example across a team
Currency: $; inputs: People = 4, Hours each per week = 6, Hourly cost = $45, Weeks = 52. Multiply across the team so anyone in finance can rerun the math and confirm the figure.
- Inputs: People 4, Hours each per week 6, Hourly cost $45, Weeks 52.
- Calculation in the cost_math block produces $56,160 annual direct labour cost across the team.
Automation vendors sell reduction percentages; buy a signed first‑year cash saving and a pilot that proves exceptions are covered.
Convert annual cost into payback months, ROI and priority
Compute first‑year cash saving = current annual cost − expected annual cost after automation. Payback months = implementation cost divided by first‑year cash saving × 12. Add contingency to implementation estimates and use payback bands to prioritize funding decisions.
- First‑year cash saving = current annual cost − expected annual cost after automation.
- Payback months = Implementation cost / First‑year cash saving × 12.
- Priority bands: green ≤12 months, amber 12–24 months, red >24 months.
- Add 20–50% contingency to quoted implementation cost unless fixed‑price vendors used.
Prioritisation matrix and mapping to build, buy or hire
Score opportunities on annual saving, implementation complexity, confidence and handover effort. High saving with low complexity and high confidence maps to buy; high saving with high complexity maps to build; low saving maps to hire or outsource part‑time.
- High saving + low complexity + high confidence = Buy (subscribe and deploy).
- High saving + high complexity + moderate confidence = Build (custom, phased pilot).
- Low saving + any complexity = Hire (outsourced specialist or contractor).
- Reduce priority if a single person owns knowledge or seasonal volume skews numbers.
The one‑page deliverable you should expect
The deliverable is a single page with the annual cost, the conservative remaining cost after automation, a confidence band, the top three automation opportunities with payback months, recommended next step and a handover checklist for procurement and IT.
- One‑line annual cost and what it represents.
- Confidence band and which inputs caused low confidence.
- Top 3 automation opportunities with payback months and recommended path.
- Handover checklist: access to reports, named SME, sample data extract, acceptance criteria for pilot.
Why these audits fail and how to stop it
Audits fail when you accept optimistic memory estimates, ignore rework, or rely on one person’s knowledge. The predictable sequence is: estimate, vendor demo, enthusiasm, then stalled implementation when exceptions appear. Fix it by timing live transactions, requiring reports, costing rework and insisting on a pilot that proves exceptions are handled.
- Failure sequence: optimistic estimate → underestimate exceptions → vendor misses edge cases → stalled pilot.
- Force a live sample transaction and require the vendor to handle three real exception cases during the pilot.
- Reduce estimated savings and add contingency when confidence is low.
Next step: run a session tomorrow and get a signed number
Book a 60‑minute quick audit with the process owner, two practitioners and finance. Time one transaction, pull the two reports, fill the one‑page template and leave with a number procurement can sign. If you want me to run the session or deliver the one‑page estimate, see how I work at How to work together or review Services.
- Schedule 60 minutes and invite the four participants.
- Bring the one‑page template and a recorder.
- If you lack access to reports, have IT or the admin on standby for 15 minutes.
Where businesses waste the most time
- Re-keying customer dataMultiple staff type the same fields into two systems, creating errors and callbacksCopy incoming form fields into the CRM and show a single confirm screen while keeping existing systemsA person still checks: A person checks flagged mismatches and confirms the record
- Chasing approvalsApprovals sit in inboxes and require follow‑ups, causing days of delayCreate an approval queue that nudges approvers and records timestamps while leaving original systems intactA person still checks: Approver signs off unusual or high‑value items
- Manual invoice matchingFinance spends time matching PO numbers and chasing suppliers for missing dataAuto‑match invoices to purchase orders and flag non‑matches for review while keeping the finance systemA person still checks: Finance resolves flagged mismatches and authorises payments
How this compares with what you already run
| Operational area | The manual way | The automated way | Annual business impact |
|---|---|---|---|
| Customer onboarding | Staff copy form data into CRM and send emails | Records created automatically; starter email sent | Faster revenue recognition |
| Invoice processing | Finance retypes supplier invoices and chases fields | Invoice fields read and matches flagged | Fewer late payments |
| Support triage | Team reads each ticket and assigns priority | System sorts routine tickets; agents handle escalations | Lower response times |
| Order entry | Orders rekeyed from email into ERP | Order details passed into ERP with verification | Shorter lead times |
| Contract renewals | Manual reminders and ad hoc tracking | Renewal dates tracked automatically with prefills | Fewer late renewals |
What the end-to-end workflow looks like
- Task arrivalA new item appears in email or a queue and is visible to the team.Automated routing creates the task in the shared queue
- Initial triageAn agent scans the task and decides if it’s routine or an exception.Routine items are pre‑classified; exceptions are flagged
- Data entryRelevant fields are populated into the working system.Most fields are copied automatically from the source
- First pass validationThe system or agent checks the record for basic errors.Automatic checks run; discrepancies sent to a human
- Exception handlingA person resolves mismatches or customer queries.Stays human
- Final approvalA named approver signs off the completed item.Approver receives a summarized task to accept or reject
- Completion and loggingThe task is closed and timestamps are recorded for reporting.Completion logged automatically and metrics updated
The metrics that tell you it worked
| Metric | What it tells you |
|---|---|
| Team hours per week on the process | Sum of hours all team members spend weekly; shows where labour is concentrated and tracks change after automation |
| Transactions per week | Count of items handled weekly; baseline for capacity and projection |
| Median minutes per transaction | Typical handling time measured by timing sample transactions |
| First‑pass success rate | Percent of transactions completed without rework; a proxy for quality and rework cost |
| Time in queue (hours) | Average time from arrival to completion; measures delay and customer impact |
| Escalations per month | Number of items escalated to senior staff; indicates edge cases automation must handle |
| Cost per transaction | Annual process cost divided by annual transactions; useful for vendor comparisons |
When this is the wrong move
- When transaction volumes are under a few hundred a year and automation costs exceed conservative first‑year saving.
- When the process is a one‑off or strategic knowledge work that cannot be reduced to predictable steps.
- When data quality is so poor you must first invest in cleaning; automation on bad data multiplies failures.
- When legal or regulatory rules require manual review of every item with no sampling allowed.
- When seasonal spikes make it impossible to capture a representative period during the audit.
The mistakes that cost the most
- Timing by memoryPeople guess minutes. You accept the number, buy a vendor, then discover actual times are double and exceptions dominate, causing pilot failure.
- Ignoring reworkYou calculate savings using only direct labour. Rework consumes substantial effort; when ignored the ROI and payback months are wrong and finance rejects the project.
- Single‑person knowledgeOne person knows the exceptions. You automate without knowledge transfer and the automated flow breaks when they are unavailable.
- Automating the broken processYou automate a flawed sequence instead of simplifying steps first. Automation then moves errors faster and preserves complexity.
Frequently asked questions
How long does a quick audit take and who should attend?
A quick audit takes 30–90 minutes depending on complexity. Invite the process owner, two practitioners who do the work, and a finance or HR representative to confirm salary and overhead inputs.
What if I don’t have system reports to confirm volume?
Time three sample transactions and use manual logs for the busiest month, or have the admin export a CSV. Mark the confidence lower and add contingency to estimates.
How conservative should my post‑automation saving estimate be?
Assume labour will be reduced but not eliminated: plan for 50–80% direct labour removal depending on exceptions. Always include rework reductions and run a pilot to validate the assumption.
When should I build rather than buy?
Choose build when the process has high complexity, unique data handling needs or long‑term IP value. Choose buy when a product maps closely to your process and offers quick deployment with fixed pricing.
What does the one‑page deliverable give finance?
It gives a single annual cost number with visible inputs, a confidence band, top automation candidates with payback months, and a recommended next step ready for pilot funding approval.
Stop paying salary for robotic work
Bring your messiest process to a 15-minute audit and we will map where the hours actually go, live on screen. You leave with the number whether or not you work with me.