Most “ROI of AI” content is vibes. Here is what 12 months of actual placement data looks like for an Australian VA agency.

Numbers are blended across 30+ active placements ranging $30k-$300k client monthly revenue. Median client size: $80k/month e-commerce or service business.

The headline numbers

Per-VA, per-month, across the full portfolio:

Metric	Non-AI VA (2024)	AI-augmented VA (2026)	Change
Output velocity (tasks/hr)	4.2	6.3	+50%
Customer-facing copy quality (5pt)	3.4	3.9	+0.5 pt
Time per support ticket (min)	9.2	4.8	-48%
Time per Klaviyo email draft (min)	38	12	-68%
Time per CSV reshape (min)	22	4	-82%
Time per inbox triage cycle (min)	18	7	-61%
90-day retention rate	73%	88%	+15 pp
Client NPS	41	58	+17

We measure these consistently across all placements. The output-velocity number is the headline: a VA running AI properly produces roughly 50 per cent more work in the same hours.

The tool cost vs the productivity gain

Per VA, per month:

Cost	$ AUD
Claude Pro	20
ChatGPT Plus	20
1Password Teams seat	12
Loom Starter seat	9
Total stack	61

Productivity gain per VA per week at a $15/hr blended rate:

Task category	Hours saved	$ value
Customer support	2.5	38
Content drafts (copy, emails, social)	1.8	27
Data wrangling	0.9	14
Admin (inbox, calendar, summaries)	1.2	18
Total weekly	6.4 hrs	97

Monthly value of saved time: roughly $420 AUD per VA.

Tool cost: $61.

ROI: ~6.9x the tool cost. The stack pays itself back in roughly 4 working days each month.

Where the gains come from

Three buckets account for 85 per cent of the gain.

1. Customer support speed (40-50 per cent of the time saving)

Reply drafts that used to take 9 minutes now take 4-5 minutes. The VA reads the customer email, runs a structured prompt, edits the AI draft for tone (usually 2-3 sentences), ships.

The quality lift is real too: support replies sound more measured because the AI defaults to calm professional tone, and the VA’s edits add specific human warmth on top.

2. Content draft acceleration (25-30 per cent)

Klaviyo flow emails, product descriptions, blog drafts, social captions. AI produces a 70-80 per cent useful first draft in 2-3 minutes; VA spends 5-10 minutes editing.

Pre-AI, the VA was either staring at a blank page (slow start) or repurposing US template content (wrong voice).

3. Data and admin productivity (10-15 per cent)

CSV merges, inbox summaries, meeting note extraction, weekly reports. The 8x speed-up on CSV work is the standout — what used to be a 22-minute VLOOKUP exercise is a 4-minute prompt-and-verify.

Inbox triage: VA scans 60 emails, AI tags each, VA processes the batch. 18 minutes becomes 7 minutes.

Where AI doesn’t move the needle

These tasks are unchanged or slightly slower with AI:

Phone calls (no AI involvement on the call itself, briefing prep maybe 1 min faster)
Sensitive customer complaints (we deliberately route around AI, see Why we don’t replace VAs with AI)
First-time client onboarding tasks (novel context AI can’t carry)
Bookkeeping and ASIC/TPB compliance (regulated work, human-only)

Roughly 30-40 per cent of an average VA’s week falls in this category. The 50 per cent productivity gain on the other 60-70 per cent gives the overall ~33-35 per cent gain across the full role.

The training cost we don’t show in the table

To get to the numbers above, every VA goes through the 5-day AI training in week 1 (see Training a new VA on AI in their first week).

That’s 3-4 hours of VA time plus 1-2 hours of team-lead supervision. At blended cost: roughly $80-120 per placement.

Recoverable cost: the productivity gains start in week 2, so the training pays itself back inside the first month.

The 60-day retention insight

The biggest commercial outcome from AI augmentation isn’t the productivity gain. It’s the retention lift.

90-day retention went from 73 per cent in 2024 (non-AI placements) to 88 per cent in 2026 (AI-augmented). That’s a 15-point retention lift.

Why: replacement is expensive. Each replaced placement costs roughly $400 in recruiting + training + lost productivity during transition. A 15-point retention lift across 30 placements means 4-5 fewer replacements per year, or ~$1,800-2,000 in saved replacement cost per VA over their average tenure.

Adding that into the ROI calculation, the tool stack returns closer to 8-10x its cost when retention is factored in.

The placements where AI didn’t work

Honest disclosure: roughly 12 per cent of placements never adopt AI properly. We track this and we replace the VA if needed.

The pattern: the VA either fundamentally doesn’t trust AI output and refuses to ship anything without re-doing it manually, or fundamentally over-trusts AI and ships hallucinations. Both are training failures. Both need either re-training or replacement.

In 2026, we caught these earlier (typically by week 3 instead of month 2 like in 2024) because our day-5 competency check makes the failure mode visible faster.

What this means for your placement

If you are evaluating a VA agency in 2026:

Ask whether they train on AI in week 1. If not, walk away.
Ask which AI tools they pay for per VA. If “none, we expect you to provide them”, walk away.
Ask for their training curriculum. If it’s verbal, walk away.
Ask what percentage of placements have a documented 30-day check-in for AI adoption specifically. If the answer is vague, walk away.

We publish all of this so you can ask the same questions of any agency you’re talking to, including us. The numbers above are the bar.

What’s next

For the stack details, see The AI-augmented VA stack.

For the training plan, see Training a new VA on AI in their first week.

For pricing context, see What does a VA actually cost in Australia in 2026.

If you want a placement that comes with the training and tool stack already running, book a discovery call.