How a Brisbane Shopify brand cut customer support response from 36 hours to under 4

The most common pattern we see in Australian e-commerce is a founder who’s great at product, stuck doing all the support. It happens because the support is the work that feels most personal, most “I can’t trust anyone else with this,” most “if I’m not doing it the brand voice breaks.” It’s the work that scales worst as the brand grows.

This is one of those founders. We’ll call the brand a Brisbane Shopify skincare brand, which is enough industry detail to anchor the numbers — the specifics are anonymised.

The business

Solo founder, six years in, $2.4M ARR. Direct-to-consumer skincare brand on Shopify. Average order value $48. Around 320 orders a week peak, 180 at base. Two retail stockists. About 8% of customers contact support per month — that’s 60-90 tickets per week through Gorgias, 30-40 Instagram DMs, and roughly 100 reviews + UGC mentions per month that need response.

The support breakdown was unbalanced:

• Gorgias tickets (email + chat): 12-14 hrs/wk
• Instagram + Facebook DMs: 4-5 hrs/wk
• Returns + refunds processing: 2-3 hrs/wk
• Review monitoring + responding: 1-2 hrs/wk
• Wholesale enquiry follow-up: 1 hr/wk

That’s 20-25 hours a week of support, all on the founder. Plus all the actual founder work — product, formulation, supplier comms, brand, ads. Something had to give. What gave was product velocity. He hadn’t shipped a new SKU in 11 months.

The thing that triggered the call

A DM came in on a Sunday from a customer who’d had a reaction to a product. The founder didn’t see it until Tuesday. By then the customer had posted publicly. The founder spent 4 hours of his Tuesday fixing the situation that 4 minutes on Sunday would have prevented. He called us Wednesday.

Why a specialist VA, not Gorgias automation

The instinct in e-commerce is to lean into automation — chatbots, AI-first response, deflection. Two reasons that’s the wrong shape for a $2M brand:

1. The brand voice is the moat. AI-first responses break it on every ticket that isn’t FAQ. Even a polite AI deflection reads transactional next to the human voice the brand built.
2. The 20% of tickets that aren’t FAQ are the ones that drive NPS. Refunds, allergic reactions, “this didn’t work for me” — these need judgement and tone, not deflection.

What he actually needed was a human running first-response with AI tools as their assistant, not the other way around. Specialist VA, customer-support background, AEST-aligned for the morning rush.

What we placed

A specialist customer-support VA at $20/hr AUD, 25 hours a week. Filipino, AEST hours, prior e-commerce support experience (an apparel brand for 3 years, then a beauty brand for 2). Vetted specifically for:

• Brand voice — could she match a warm-direct AU register?
• Comfort with judgement on borderline tickets (refund decisions, dispute escalations)
• Familiarity with Gorgias + Shopify admin
• Comfortable using Claude as a drafting assistant without breaking voice

Stage 5 of vetting: drafting responses to 8 sample tickets we built from real support patterns. Three candidates ran it. The founder picked based on the one whose drafts needed the least editing.

The first 30 days

Week 1 — FAQ build + voice training. The VA shadowed every ticket the founder handled. She wrote up the FAQ doc (28 entries) and the canned-response library (12 entries, written in the brand voice). The founder reviewed and edited. Brand-voice prompt for Claude written in week 1 — the VA used Claude to first-draft, then edited to brand voice, then sent.

Week 2 — Took over Gorgias inbound. Email tickets first, live chat second. Routing rules: standard FAQ tickets handled solo with brand-voice replies; refund decisions under $100 handled solo using the policy doc; refunds over $100 escalated; allergic-reaction tickets always escalated. Average response time dropped from 36 hours to under 6 by end of week 2.

Week 3 — Instagram + Facebook DMs. Same routing logic, adapted for social tone. The brand started replying to DMs within 15-30 minutes during AEST business hours. Instagram engagement metrics flagged the change immediately.

Week 4 — Returns + RMA + review moderation. Returns processing in Loop. Review responses on the product page + the Judge.me dashboard. The founder set the rule: every 1-star review gets a personal reply from him; everything else the VA handles.

What didn’t work in month one

Brand voice broke on week-3 social DMs. Instagram tone is different from email tone — shorter, more emoji, less formal. The VA’s email tone leaked over the first week of DM coverage and 3-4 customers commented. We added an Instagram-specific tone subsection to the brand-voice prompt. Fixed by end of week 3.

The “always escalate allergic reactions” rule was over-broad. Some “I had a tingling” reports are normal product reactions, not allergic. The VA was escalating routine sensations to the founder, defeating the time-saving. We refined to “escalate skin-changes (redness, swelling, hives), handle routine sensations with the standard reassurance template.” The founder co-wrote the new rule.

Gorgias macros took two iterations. First version was too templated and read auto-generated. Second version was the canned response with 30% mandatory editing. That landed.

The numbers, 6 months in

• Total monthly cost of the VA placement: $2,200 AUD ex-GST
• Customer support response time: 36 hours → under 4 hours (under 1 hour for live chat during AEST hours)
• NPS: 42 → 60 over 6 months (+18 points)
• Refund rate: held flat at 2.8% (the worry was that faster support would prompt more refund asks; it didn’t)
• Founder hours reclaimed weekly: 20
• New product SKUs shipped in the following quarter: 3 (zero in the year before placement)
• Wholesale account wins (driven by faster wholesale-enquiry follow-up): 4 new stockists

What this looks like for your business

If you’re an AU Shopify (or BigCommerce / WooCommerce) founder doing $1M+ revenue and personally handling support, this placement shape works. Variables:

• Order volume + ticket-per-order ratio — adjusts hours per week
• B2B vs B2C — B2B needs more judgement, longer ticket cycle; the VA tier might shift to senior specialist
• Support tool stack — Gorgias, Zendesk, Front, Help Scout, even direct Shopify inbox all work
• Brand voice complexity — higher-touch brands (luxury, allergen-sensitive, regulated) need more brand-voice training in week 1

The next step is a 30-minute discovery call. We’ll work out which support functions to delegate first based on what’s costing you the most response time today.