Peregrine field notes
The $136,000 question
The $136,000 AI productivity figure is a national opportunity estimate—not a promise. Here’s how Australian small businesses can turn it into measurable admin savings.
What does the $136,000 figure actually mean?
The number comes from the Tech Council of Australia’s Accelerating AI Impact for Small and Medium Enterprises report. It estimates that an average Australian SME moving from no AI use to broad adoption could unlock a potential annual productivity gain worth $136,000. The same report estimates that wider SME adoption could create up to $93 billion in additional annual economic value over the next five years. [2]
That is a useful way to show the scale of the opportunity. It is not a promise that every business will save $136,000, add $136,000 to profit or recover that amount in wages. The estimate describes potential productivity uplift across a broad adoption scenario. It also assumes a business can move beyond occasional experimentation and use AI across its operations.
That distinction matters. A business using an AI assistant to write the odd email is not in the same situation as a business that has redesigned repetitive work across lead handling, scheduling, customer service, finance and reporting.
The report says only around one in four Australian SMEs was using AI, and just 5% had adopted it broadly across the organisation. It also identifies limited understanding, skills gaps, uncertainty about return on investment, and technical readiness as barriers. [1]
Where can a small business find a realistic first gain?
Most owners do not need a grand “AI transformation” to test the value. They need a clear before-and-after measurement for one workflow.
| Workflow | What to measure before changing it | A practical first improvement |
|---|---|---|
| Lead intake | Time from enquiry to a complete record and first reply | Send web or email enquiries into one pipeline and trigger a prompt follow-up |
| Quoting | Time spent gathering details, preparing a quote and chasing a decision | Capture the required information once and create a consistent quote hand-off |
| Invoicing | Minutes spent re-keying job, customer and payment information | Pass approved job details into accounting and schedule reminders |
| Calendar coordination | Back-and-forth messages and double-booking corrections | Sync the agreed booking information between the relevant tools |
| Weekly reporting | Time spent collecting figures and checking spreadsheets | Pull the agreed measures into one review view with clear ownership |
The aim is not to claim that every minute becomes revenue. Some recovered time becomes capacity, some becomes faster service, and some reduces errors or owner stress. Those are valuable outcomes, but they should be measured separately rather than bundled into an inflated ROI number.
What could a conservative business case look like?
Use an example that a business can test, not a headline that cannot be audited.
Suppose a service business identifies five hours of repeatable admin each week and values that time at $60 per hour. Across 48 working weeks, the measurable capacity released is:
- 5 hours × $60 = $300 per week
- $300 × 48 weeks = $14,400 of annual capacity
That does not mean the business will bank $14,400 in profit. It is a starting estimate. The owner still needs to check whether the time is genuinely removed, redirected to billable work, used to improve response times, or simply absorbed by other tasks.
A stronger business case records three separate measures:
- Time returned: hours no longer spent on the manual step.
- Operational result: faster response, fewer errors, fewer overdue invoices or more completed jobs.
- Financial result: additional gross margin or avoided cost that can be tied to the change.
That approach keeps the $136,000 figure in its proper place: a national opportunity estimate, not a substitute for your own baseline. For the broader maths of automation payback, see our automation ROI guide for small business.
Why broad adoption is harder than adding another AI tool
The TCA research is about adoption across an organisation, not simply buying software. Broad adoption usually requires three things to work together:
- The process is clear enough to automate.
- The connected systems hold reliable information.
- The people doing the work trust the result and know what to do when something falls outside the rules.
This is why a done-for-you automation system should start with the workflow, not the tool. A reliable lead-to-follow-up process may combine a website form, CRM, email, calendar and accounting system. The useful outcome is not “we added AI”; it is that a lead is captured once, routed correctly and followed up without relying on someone’s memory.
The report’s practical message is similar: many SMEs need help understanding which applications fit their business and how to put them into practice safely. [2]
Common questions about AI productivity for small business
Is $136,000 the amount a small business should expect to save?
No. It is the report’s estimated potential annual productivity gain for an average SME moving from no AI use to broad adoption. Treat it as a directional benchmark, then build a smaller business case from measured hours, outcomes and financial impact.
What should an Australian small business automate first?
Start with a repetitive, high-friction workflow that has a clear owner and a measurable outcome. Lead capture and follow-up, quoting, invoicing and data movement between existing tools are sensible candidates. Choose the process where missed steps are already costing time or opportunities.
How can I tell whether an automation is working?
Record a baseline for two to four weeks, then compare the same measures after the change. Track time returned, response or completion speed, errors and a financial measure where one can be evidenced. Do not count an unavailable result as zero, and do not treat capacity released as revenue until it is actually converted.
Do I need an AI agent to get a productivity gain?
Not necessarily. A deterministic workflow is usually better for predictable steps. AI is useful where the input is messy or needs interpretation, provided there are clear checks and a human fallback. The right system may be a workflow, an AI component or a hybrid of both.
The useful question is smaller than the headline
The $136,000 question is attention-grabbing, but it is not the question an owner should use to approve a project. Ask instead: which repeatable task is costing us the most time, what would “better” look like, and how will we prove it?
That turns a national productivity estimate into a local experiment. Start with one workflow, connect the tools you already use, measure the result, and expand only when the evidence supports it. If you want help finding the best starting point, Book an Automation Audit or try the ROI calculator.
Matt Sullivan | Founder - Peregrine Automations