When Should a Small Business Automate? A Break-Even Worksheet

Quick answer: Automate a stable, repeated task when the time saved still justifies setup, checking, exceptions and software costs. Calculate with ordinary weeks, not the busiest day you can remember. A workflow that saves five minutes and creates an hour of detective work has not quite retired you.

Start here: See the break-even example · Copy the decision card

For: small teams deciding whether to connect forms, spreadsheets, CRMs or routine reports. This is an operating-cost model, not an investment forecast. All amounts are hypothetical US dollars; no tool prices or measured savings are claimed.

Start with one task and a clear finish line

“Automate sales” is too broad to test. “Copy an approved enquiry into the CRM and assign its owner” gives you an input, a destination and an outcome to inspect.

Observe the manual version first. Sample an ordinary week and a quiet week, and record the number of items, time for a normal item, and cases requiring judgment. Write down which inputs arrive incomplete. Averages alone can hide a small set of costly exceptions; keep their count and recovery time visible.

• Good pilot candidate: repeated steps, consistent input, a clear destination and a result someone can verify.
• Simplify first: unclear ownership, inconsistent spreadsheet columns or several competing definitions of “complete.”
• Keep a human gate: actions where a wrong result could expose sensitive information, commit money or send a consequential message.

Sometimes a better template, a shared queue or a scheduled batch removes most of the work. Compare those alternatives too. Automation is one possible solution, not the admission ticket to being organised.

Count four costs people often leave out

Setup: design, permissions, building, testing, documentation and training. Include your own time at a consistent planning rate even if nobody sends an invoice.

Routine supervision: checking outputs and reviewing the queue. If every record still needs a minute of inspection, count that minute.

Exceptions: finding and fixing failed, duplicated or incomplete items. Measure the rate and recovery time rather than assuming failures cost nothing.

Recurring cash cost: the incremental software, usage charges and supporting services needed for this workflow. Do not count a shared subscription in full against every workflow, or treat a necessary upgrade as free.

For a planning comparison, use:

• Manual hours per month = monthly items × manual minutes per item ÷ 60.
• Automated-process labour hours = routine review hours + exception count × recovery minutes ÷ 60 + maintenance hours.
• Monthly net capacity value = (manual hours − automated-process labour hours) × planning hourly rate − incremental recurring cash cost.
• Simple payback in months = setup value ÷ positive monthly net capacity value.

Do not calculate a meaningful payback when the monthly net value is zero or negative. This simplified model ignores financing and tax effects and assumes the workload and costs stay broadly stable.

Worked example: useful at 300 items, weak at 100

Illustrative assumptions: each manual item takes four minutes. The automated version still needs one minute of review. Ten percent of items need five additional minutes of recovery. Maintenance takes one hour per month. Use a $30 hourly planning rate, $40 monthly incremental software cost and $600 of one-time setup effort.

1. Manual workload: 300 × 4 ÷ 60 = 20 hours per month.
2. Review: 300 × 1 ÷ 60 = 5 hours.
3. Exceptions: 300 × 10% × 5 ÷ 60 = 2.5 hours.
4. Total ongoing labour: 5 + 2.5 + 1 = 8.5 hours.
5. Net capacity value: (20 − 8.5) × $30 − $40 = $305 per month.
6. Simple payback: $600 ÷ $305 ≈ 2.0 months.

On a small screen, swipe the table sideways.

The table keeps every other assumption unchanged; actual usage charges and exception rates may rise with volume. Calculations use unrounded hours. The hourly planning rate values time for comparison—it is not evidence that payroll will fall.

Under these assumptions, each item creates $1.25 of capacity value before the fixed $70 monthly burden of maintenance and software. Net value becomes positive above 56 items per month. Recovering the $600 setup within six months needs 136 items per month. These thresholds belong to this example, not to automation generally.

If work arrives in bursts, use two cases

A monthly average can conceal the day that overloads a person. Run the calculation once for a normal month and once for a peak month, using the same review and exception assumptions. If automation only pays during the peak, compare it with temporary batching or a clearer hand-off before accepting a year-round subscription.

Illustrative decision: a team processes 100 items in most months and 300 for two seasonal months. The table gives $55 and $305 of monthly capacity value respectively. Across ten ordinary and two peak months that is $1,160 of annual planning value before the $600 setup effort. That is a possible case for a pilot, but it leaves less room for slower review or higher software usage than a constant 300-item workload.

Time saved is not automatically cash saved

If the team remains on the same payroll, the immediate result may be spare capacity rather than a lower cash bill. Name what that capacity will be used for: faster replies, more billable work, fewer late evenings or avoiding an additional hire as volume grows.

Keep a separate cash view. In the example, software costs $40 per month. The $600 setup is a valued effort assumption; if it is instead a contractor invoice, it is also a cash outlay. Do not label the $305 monthly planning value as profit without evidence of reduced spending or additional contribution.

Non-financial benefits can still justify a pilot: clearer records, more consistent hand-offs or a shorter response delay. Write the benefit and how you will observe it separately, so the arithmetic does not have to pretend to prove everything.

Check usage and recovery costs in the actual tool

One business item does not necessarily equal one billable unit. Make’s credits documentation distinguishes operations from billing credits and explains that usage can depend on features and connection type. Inspect representative runs and the required plan rather than multiplying a headline price by your item count.

Recovery is also a designed process. Make’s Retry error-handler documentation describes saving failed work and retrying it manually or automatically according to settings. That capability does not decide whether repeating your business action is safe.

Before repeating a partially completed job, inspect the destination for effects already created. Use the recovery and duplicate-prevention checklist to define the safe path. The repair time belongs in the cost model.

Run a bounded pilot and choose a stop rule

Start with a low-risk sample and retain the manual route. Prevent the manual and automated versions from both creating the same customer-facing result. Compare the output, not just whether the automation platform reports success.

Measure normal cases, an exception, a repeated event and a staff hand-off. Have the backup owner find a failed item, decide whether it already changed the destination, and recover it without the builder explaining each click. If review takes longer than expected, recalculate. If the process cannot recover safely, stop the pilot regardless of its apparent payback.

Set your own acceptance limit before building. For example: “We will proceed only if mandatory safety tests pass, observed work supports our six-month payback target and a named person can operate recovery.” Six months is an illustrative management choice, not a universal rule.

Copy the decision card

• Task and completion evidence: input, destination and verified result.
• Baseline: items per month, manual minutes and variation.
• Ongoing burden: review time, exception rate, recovery time, maintenance and software cost.
• Setup: effort value, actual cash outlay and responsible owner.
• Benefits: capacity value, cash effect and non-financial benefit recorded separately.
• Decision: automate, simplify first or remain manual; review date and stop rule.

Next: use the SaaS trial scorecard to compare tools and the subscription audit to assess the wider stack.

Official documentation checked September 29, 2026. The model, thresholds and examples are editorial calculations using hypothetical inputs, not vendor claims or measured business results.