How to calculate Copilot ROI for an Australian business

Every Australian business case for Microsoft 365 Copilot comes down to a simple ratio: the fully-loaded cost of the licences versus the fully-loaded value of the hours Copilot gives back. The arithmetic isn't complicated. What's complicated is agreeing on what goes into each side.

We've written and defended a few dozen Copilot business cases for Australian organisations over the past two years. The ones that survive contact with the CFO share four traits: conservative hours-saved assumptions, Australian salary bands not US ones, on-costs that include superannuation and leave, and a sensitivity analysis that shows the CFO where the case breaks. Here's how the maths actually works.

The cost side — what Copilot actually costs in Australia

Microsoft 365 Copilot is listed at USD $30 per user per month on an annual commitment. For Australian businesses buying via the standard CSP channel that converts to AUD $45 per seat per month, give or take depending on currency and Microsoft's periodic regional price adjustments. Enterprise Agreement customers sometimes negotiate lower rates; charities get education SKU pricing on different terms. For most AU mid-market organisations, AUD $45 is the right planning number.

Three implications from the cost side that catch CFOs by surprise. First, it's an annual commitment — you're not licensing seats monthly, you're licensing them for the year. Second, the seat cost is on top of your existing M365 E3/E5 or Business Premium licence, not in place of it. Third, Copilot is priced per seat assigned, not per seat actively used. You pay for what you buy, whether adoption lands or not.

The value side — hours saved, done realistically

Microsoft's own research on early Copilot adopters cites time savings in the range of 10 to 15 percent on knowledge work. That's the headline number you'll see quoted back at you by vendors. It's directionally right and it's not what we use for an AU business case.

What we use for planning is a per-role hours-saved estimate that reflects what Copilot actually does well and where it struggles. Copilot is excellent at drafting, summarising, restructuring and information retrieval. It's mediocre at specialised domain reasoning and currently unreliable at anything requiring multi-step coordination across applications. That means the hours-saved profile differs sharply by role.

• Leadership and executives: 3–5 hours per week. Mostly briefings, draft correspondence, meeting summarisation, rapid document review.
• Managers: 2.5–4 hours per week. One-pagers for the exec team, operational reports, team communications, performance review drafting.
• Knowledge workers (analysts, project managers, operations): 2–3 hours per week. Document drafting, SharePoint search, data summarisation.
• Sales and client-facing: 1.5–3 hours per week. Call summaries, proposal drafts, CRM updates — higher if you have Copilot for Sales.
• Specialists (engineering, legal, clinical): 1–2 hours per week. The domain reasoning Copilot can't yet do is a large portion of specialist work.
• Back-office (accounts, payroll, admin): 1–2 hours per week. Copilot is less useful in highly structured workflows — unless paired with custom agents.

Use the lower end of each band for the business case. The upper end is what you'll hit after twelve months of deliberate adoption, strong change management, and custom agents for the functional areas where generic Copilot doesn't fit. Most Australian rollouts we've seen are at the lower end at six months.

On-costs and the fully-loaded rate

A Copilot business case done at base salary is wrong in the CFO's favour. The fully-loaded cost of an Australian employee-hour includes superannuation (currently 12 percent), payroll tax (varies by state, typically 4–6.85 percent above the threshold), workers' compensation, leave loading, public holidays, training and the amortised cost of the tech stack that employee uses. A reasonable all-in multiplier for most AU mid-market businesses is 1.3 to 1.4 times base salary.

If your business keeps more detailed cost-to-company figures, use those. If not, 1.3× is defensible. It's also conservative — most genuine fully-loaded costs come out closer to 1.4×. Under-estimating the value side is always safer than over-estimating it.

The adoption curve you should plan for

Licence-assigned does not equal actively-used. The adoption curve we see across Australian mid-market rollouts looks like this: 30 to 40 percent active use at 60 days post-rollout, 50 to 60 percent at 120 days, 70 to 80 percent at twelve months if change management lands well. Plan the business case at 70 percent steady-state for most roles, 90 percent for leadership (exec teams adopt fast), and 60 percent for back-office (adopts slowest).

If you're presenting the case to the board, present two scenarios: the planning case (conservative adoption, lower-end hours-saved) and the stretch case (mature adoption, upper-end hours saved). Both should show positive ROI. If the planning case doesn't, you don't have a business case yet — you have a wish list.

The CFO pushback, and how to answer it

A Copilot business case will get three questions at a competent Australian audit committee or CFO review. Prepare the answers before you go in.

"How do we know the hours saved are real?"

Hours saved are self-reported until you measure them. The honest answer: the first 90 days of the rollout are a measurement exercise, not a productivity one. Sample 20 users per role, time-track the specific tasks Copilot touches, and compare to a non-Copilot control group. The business case in months 4–12 is based on observed data, not survey responses. Copilot licence economics work even at 70 percent of the projected hours-saved figure, so there's headroom for reality to disappoint the model.

"What if adoption is lower than we projected?"

Show the sensitivity analysis. Re-run the calculator at 50 percent adoption instead of 70 percent. If the ROI is still positive at that rate, the case holds. If it isn't, the case was over-fit. The right answer is usually to scope Copilot to the roles where adoption is most likely — leadership, managers, sales — rather than every seat in the organisation.

"What's the downside risk?"

Three real downsides deserve specific answers. Data leakage via Copilot surfacing oversharing — mitigated by the Essential Eight plus oversharing audit work. Licence burn on inactive seats — mitigated by a 6-month adoption review with the right to de-assign. Cultural pushback from staff who feel surveilled — mitigated by a clear policy on what Copilot logs and who reads them. None of these are hypothetical; all of them are manageable.

The one-page business case

A defensible Copilot business case for an Australian board fits on one page. It contains: the seat count and total licence cost in AUD; the per-role hours saved and dollar value in AUD; the fully-loaded multiplier with a footnote on super and payroll tax; a base case and a stretch case; a sensitivity analysis at 50 percent of projected adoption; and a clear risk statement covering oversharing, adoption and cultural pushback.

The calculator on this site outputs exactly that — with AU defaults in every field and a PDF that's formatted for a board pack. Use it as the arithmetic. The judgement calls on adoption and hours saved are your own, and worth more time than the spreadsheet.