opinion

AI Is Helping. So Where Are the Time Savings?

AI time savings depend on the work required to complete a task, including checking and correction after an answer arrives. This article explains how to measure that effort, distinguish released capacity from cash savings and include implementation and ongoing costs in the assessment.

By Lambros Photios

01

Where Is the Time Still Going?

The useful unit of measurement is a completed task with an acceptable result. Faster searches or first drafts can reduce one step while leaving reading, verification, hand-offs and corrections in place. A before-and-after comparison needs to include those steps wherever they happen, including work outside the AI interface.

Quality determines whether the apparent time saving is usable. If an answer needs additional checking or a task has to be repeated, that effort belongs in the new workflow's total. The comparison also needs similar kinds of work on both sides; a straightforward request and an unusual exception may require very different amounts of attention.

Elapsed time and staff effort answer different questions. A task can wait in an approval queue without occupying someone throughout that period. Recording both measures helps distinguish faster turnaround from time that people could use for other work.

02

What Would a Useful Answer Let Someone Do Next?

A clear outcome gives the development team something to build and the business something to test. For a document assistant, a useful answer may need to combine sources, identify the current instruction and show enough evidence for someone to check it. Finding a relevant file establishes only part of that requirement.

The people doing the work can identify the checks that matter. Their input should explain what information counts as current, who resolves conflicts and what someone needs before acting on an answer. Discovery translates those decisions into a proposed workflow and identifies what still needs investigation.

The solution may involve existing tools, a process change or additional software. Our business process automation guide covers selecting a process to investigate. Once the candidate is defined, the assessment here concerns the effort, costs and usable benefits of the proposed change.

03

What Could Make the Next Step Impractical?

Usage and information volume can change the cost of an otherwise workable design. The assessment needs assumptions about how often the system runs, how much material it processes and how frequently that material changes. A small demonstration does not establish the cost of operating at the intended scale.

Implementation work extends beyond producing the first answer. Preparing information, connecting systems, resolving access rules and testing outputs all require effort. Staff training and the transition from the current process also belong in the estimate. The custom software costs guide explains how these scope decisions affect a proposal.

Ongoing costs continue after launch. Depending on the arrangement, these can include licences, model usage, hosting, support, monitoring and maintaining integrations. Human checking is part of the operating workload. Count that effort consistently so it is neither omitted nor charged twice in the calculation.

04

What Did This Mean for a Document Assistant?

A high-volume manufacturing company in Australia wanted to move from finding documents to receiving useful answers from their contents.

In its existing workflow, the company used Copilot, supported by its managed service provider, to help find files in SharePoint. It had seen enough value to ask whether the experience could go further.

The volume of files, historical versions and access rules shaped the proposed solution. Reading the whole collection for every question was not a feasible cost model for this use case.

The approved design combined keyword and meaning-based retrieval to identify relevant material, with permission checks before answer composition, document-version checks and citations. It also included document cleanup, human review where needed and customer acceptance testing.

The business contributed its use cases, organisational structure and access rules, with its internal IT function and managed service provider involved. We translated those inputs into a technical plan and estimated initial ingestion and ongoing costs. The first stage covered a selected part of the environment.

The outcome at that point was an approved plan and authorisation to build. The time savings would need to be established through implementation and use.

05

How Much Capacity Could the Change Release?

Released capacity is the difference in staff effort across the same volume of completed work. The following arithmetic is hypothetical, not a client result, market benchmark or Adaca quote. It assumes the quality of completed tasks remains acceptable and that the measured effort includes checking, exceptions and correction.

Suppose 600 tasks each month take an average of 12 minutes before the change and eight minutes afterwards. The monthly reduction is 600 × (12 − 8) ÷ 60 = 40 hours. If maintaining the workflow requires another six staff hours each month, the net capacity released is 34 hours.

The outcome is sensitive to checking effort. If the new average is ten minutes instead of eight, the same calculation releases only 14 hours after maintenance. Measuring the completed task during a trial gives the business evidence for choosing between those assumptions.

Released hours need an identified use. They might allow the team to clear a backlog, handle additional work or spend more time on tasks that previously received less attention. These are potential operating benefits to verify. Multiplying the hours by a salary rate does not establish that payroll expenditure will fall.

06

When Does Released Time Become a Cash Saving?

A cash saving requires a cost that can actually be reduced or avoided. Examples to investigate include paid overtime, external processing fees or an additional purchase that would otherwise be necessary. The assessment should identify the expense, the decision that changes it and the period in which that change takes effect.

Unchanged staffing costs remain unchanged in the cash calculation. If the team uses the released capacity for other work while salaries stay the same, the benefit belongs in the capacity assessment. A forecast of avoided future hiring also needs its own evidence about workload, timing and the work the existing team can absorb.

A cash-based automation ROI calculation compares attributable cash benefits with incremental costs over an explicit period. In a separate hypothetical example, assume a verified reduction in external processing expenditure of AUD $1,000 per month, incremental running costs of AUD $400 per month and a one-off implementation cost of AUD $6,000. These are arithmetic inputs only, not price guidance or a predicted result.

Under those assumptions, the recurring net cash benefit is AUD $600 per month. Over 12 months, the AUD $12,000 gross benefit less AUD $4,800 running costs and AUD $6,000 implementation cost leaves AUD $1,200. Dividing that net benefit by total incremental costs of AUD $10,800 gives an illustrative first-year ROI of about 11.1%. The simple payback period is ten months from the start of the recurring benefit.

Delayed adoption or lower usage changes that result. The example assumes the full monthly benefit begins immediately and remains constant; it omits tax, financing and discounting. A project assessment should state its timing and cost boundaries, including any transition period, rather than treating this arithmetic as a forecast.

07

What Should the First Measurement Plan Contain?

The measurement plan should establish the baseline before the change begins. It needs a count of completed tasks, staff effort, exceptions and an agreed quality check. Recording the source and measurement period makes the comparison reviewable and helps identify a change in task mix.

The proposed benefits should have separate owners and evidence. The process owner can verify released capacity and how it is used, while the person responsible for the budget can establish whether expenditure changed. Implementation and support estimates should show which costs are already included and which remain uncertain.

A trial should test the assumptions most likely to change the decision. Those may concern checking time, the proportion of work the system can handle or the cost at realistic usage. The result can then support a decision to expand, revise or stop the proposed approach.

The next review should compare the completed work with the original baseline. That provides a basis for deciding whether the change has released usable capacity, reduced expenditure or needs further work.