Rethinking the ROI of AI in human services

Ask most organisations how they plan to measure the return on investment of AI, and the answer comes back quickly: money. Will it reduce costs? Will it increase revenue? Will it pay for itself within the financial year?

Those are fair questions. Every provider in disability and aged care is operating under real financial pressure, and no one can afford to invest in technology that doesn't deliver. But when ROI is measured through a purely financial lens, we see only a fraction of the value AI can create. Worse, we risk making poor decisions: walking away from tools that are quietly improving care, or backing tools that look efficient on paper but make work harder for the people doing it.

It's time for a broader view.

AI is an addition, not a replacement

There's an uncomfortable truth worth naming up front. For most organisations right now, AI isn't replacing an existing system. It's being added on top of an already costly tech stack.

Providers are already paying for client management systems, rostering and payroll platforms, incident management tools, finance software, learning management systems and more. AI solutions are not currently switching many of these off. Instead, they sit alongside them, connecting, summarising, automating and assisting.

That changes the ROI conversation. The question isn't “Is this cheaper than what we had?” because, in direct terms, it's an extra cost. The real question is “What value does this add on top of what we're already paying for?” And sometimes, “Does it help us get more out of the systems we already own?”

That won't always be the case. As AI matures, it will increasingly take on the work of the systems it currently sits beside, and system replacement will become part of the ROI equation. For now, though, the case has to be made on the value AI adds, not the systems it retires.

Answering that honestly means looking well beyond the finance line.

Decide what to measure before you begin

You can't understand impact without measuring outcomes. And you can't measure change without knowing where you started.

That's why the question of what to measure belongs at the very start of the AI conversation, not after go-live. Agreeing early on the outcomes that matter, and capturing a baseline for each, makes it possible to measure honestly after implementation.

Leave it too late and the baseline is gone. You're left with anecdotes and assumptions, and the return, however real, becomes hard to prove.

Four domains of return

In human services, the value of AI shows up in at least four places. Each deserves its own measures.

1. Financial

Financial return still matters, and it belongs in the picture. This includes the obvious measures, such as time saved on administration, reduced overtime and on-call costs, faster billing and claiming, and fewer errors that lead to rework or lost revenue. It also includes less obvious ones: lower recruitment and onboarding costs when staff stay longer, or reduced spend on agency staff.

The key is to treat financial return as one domain among several, not the only one that counts.

2. Workforce

The human services workforce is stretched. Frontline workers, coordinators, clinicians and team leaders spend significant time on documentation, reporting and administration that takes them away from the people they support.

When AI gives time back, the return shows up in workforce measures: reduced administrative load, improved job satisfaction, lower burnout, better retention and stronger capability. Asking staff directly through short pulse surveys before and after implementation, and tracking turnover and sick leave over time, can make this visible. Our article on workforce measures sets out what a single departure costs and how to measure it.

In a sector where every resignation carries a real cost to continuity of care, workforce return is not a soft benefit. It's fundamental.

3. Clients and families

Ultimately, human services exist for the people they support. AI's value should be felt by clients and families, even if they never see the technology itself.

That might look like more time with workers who are present rather than distracted by paperwork, more consistent support across shifts and teams, faster responses to questions and concerns, and communication that is clearer and more accessible. Measures could include client and family satisfaction, complaint trends, goal progress and feedback on responsiveness.

If an AI solution saves money but makes the experience of support worse, it hasn't delivered a return. If it improves that experience, that value deserves to be counted.

4. Quality and safety

High-quality, safe support depends on good information: accurate notes, timely incident reporting, clear handovers and early recognition of risk. These are exactly the areas where AI can help.

Return in this domain might include improved documentation quality and completeness, earlier identification of emerging risks, more consistent application of practice standards, stronger audit readiness and fewer reportable incidents. For organisations operating under quality and safeguards frameworks, these outcomes carry both ethical and regulatory weight.

The ripple effect: why ROI must be enterprise-wide

These four domains don't sit in silos, and neither does the impact of AI.

When AI is introduced in one part of the business, its effects ripple into others. Reduce documentation time for support workers, and you may see more complete notes, which leads to smoother audits, earlier risk detection, fewer incidents and lower turnover in teams that never touched the tool directly. Improve intake processes, and finance, rostering and clinical teams all benefit downstream.

This is why measuring ROI service by service, or tool by tool, consistently undervalues AI. If you only look at the team where a solution was implemented, you'll miss much of the return. A meaningful assessment looks across the whole enterprise and asks where value has been created, not just where the cost sits.

It also means ROI conversations need to involve more than the finance team and the project sponsor. Operations, quality, people and culture, and frontline leaders all hold part of the picture.

Getting started

A broader ROI lens doesn't need to be complicated. A few practical steps can make a big difference:

• Set baselines before you start. Capture current measures across all four domains so you have something to compare against.
• Choose a small number of measures per domain. Two or three meaningful indicators each are better than a sprawling dashboard no one reads.
• Look beyond the pilot team. Track indicators in adjacent teams and functions to capture the ripple effect.
• Combine numbers with stories. Staff and client experiences often reveal value that metrics miss.
• Give it time. Some returns, such as retention and quality improvements, take months to emerge. Judge too early and you'll only see the cost.

A fuller picture of value

AI in human services will rarely justify itself on cost reduction alone, especially while it sits on top of an existing tech stack. But that doesn't mean it isn't delivering a return. It means we've been measuring too narrowly.

When we look across financial, workforce, client and family, and quality and safety outcomes, and across the whole enterprise rather than a single service, a much richer picture emerges. It's one that reflects what this sector actually values: sustainable organisations, supported workers, and safe, high-quality care for the people at the heart of it all.