Do Australian shoppers still want a human when peak-season support goes wrong?
Yes, when something goes wrong. In a Twilio survey run in April and May 2026, 83% of Australian consumers tried to bypass an AI agent to reach a person, and a Sinch survey found only 2% would choose a chatbot to resolve an issue. Simple jobs, such as a return, are a different story.
Table 1 sets out what we found. Treat every source as vendor research: Twilio, Sinch, Zoom and Qualtrics all sell customer-engagement or experience software, and none of the coverage we read gives an Australian sample size.
Table 1: What Australian shoppers say about human and automated customer support, 2025 and 2026
| Measure | Latest figure | Source and date | Caveat |
| Australians who try to bypass AI agents to reach a person | 83%, spending an average of 101 seconds trying | Twilio 2026 Customer Insights Series, reported by B&T, 27 Oct 2026 (7,652 consumers and 660 business leaders in 18 markets; fieldwork 9 Apr to 1 May 2026) | Vendor research; Australian sample size and method not stated |
| Time saved by AI agents | 63 seconds on average, a net loss of 38 seconds once the bypass attempt is counted | Twilio, via B&T | As above; how time was measured is not stated |
| AI fully resolved the query | 25% said most or all of the time; about half said rarely or never | Twilio, via B&T | Self-reported |
| Frustration with being sent to a bot | 57% felt frustrated or undervalued (37% globally); four in 10 are less likely to choose a chatbot than a year ago | Twilio, via B&T | Self-reported |
| Counterpoint: comfort with AI | 48% comfortable talking to AI; 71% comfortable with AI processing a return or exchange; a third would prefer AI if it resolved the issue faster; 42% comfortable with AI making a purchase or payment | Twilio, via B&T | Shows acceptance of routine, low-risk tasks |
| Choosing a chatbot to resolve an issue | 2% would choose a chatbot; 55% favour resolving support questions by email | Sinch, reported by CMOtech and eCommerceNews Australia, 25 Nov 2025 | Sample, dates and survey firm not stated; the article calls the 55% email with a human, but the Sinch executive's quote says email only |
| Escalation gap | 86% expect AI tools to hand unresolved issues to a person; 38% see that regularly; 51% are frustrated by bots that fail to resolve their issue | Zoom, with Morning Consult, reported by IT Brief Australia, 14 Aug 2025 (3,509 consumers in seven countries) | Over a year old; Australian sample size not stated; the article also uses 86% for a second, different claim |
| Cost of a bad experience | 41% of Australians will cut back or stop spending with a company after a negative encounter | Qualtrics XM Institute, reported by ChannelLife, 28 Nov 2025 | Sample, dates and method not stated; the A$66 billion headline has no published method and is not used |
| Satisfaction with automated service | 78% had used automated customer service systems; 39% were satisfied | Survey of 1,031 Australians cited in a University of Queensland Business School opinion piece, 2 Sep 2026 | Survey not named; secondary mention in an opinion piece |
Sources: B&T (Twilio, 27 Oct 2026); CMOtech (Sinch, 25 Nov 2025); IT Brief Australia (Zoom and Morning Consult, 14 Aug 2025); ChannelLife (Qualtrics, 28 Nov 2025); University of Queensland news (2 Sep 2026). Twilio, Sinch, Zoom and Qualtrics sell customer-engagement or experience software. We read news coverage, not the primary reports.
The pattern is consistent. Australians will let software handle routine, transactional work, and Twilio found 71% comfortable with AI processing a return or exchange. They want a person when the issue is unusual or expensive. Only 25% said AI fully resolved their query most or all of the time, and the 101 seconds spent trying to reach a human outweighed the 63 seconds the AI saved. Zoom's 2025 research found 86% expect an AI tool to hand over to a person but only 38% see that regularly.
There is also an Australian cautionary case. In August 2025 the Commonwealth Bank reversed its decision to cut 45 customer service roles after introducing a voice bot, saying its assessment had been in error. The Finance Sector Union said call volumes had been rising and team leaders had been pulled onto phones. That is a union's account, and a bank is not an online retailer, but it shows how deflecting the easy contacts can leave people with the hard ones.
We do not use the A$66 billion loss figure attached to the Qualtrics research, because no method was published. The 41% figure is in Table 1.
How big does the peak get, and when does it hit?
Bigger than the Black Friday weekend, and longer. Table 2 collects the demand figures we could verify.
Table 2: Peak-season demand: spending, shoppers and parcels, 2025 and 2026
| Measure | Latest figure | Source and date | Caveat |
| Black Friday to Cyber Monday spend forecast, 2026 | Record A$7.2 billion for 27 to 30 November, up 5.5% (A$374 million) on last year | Australian Retail Council (ARC) and Roy Morgan, 11 Nov 2026 (SMS survey of 7,654 Australians, 15 to 19 Oct 2026) | A forecast from a survey, not recorded sales |
| Shoppers and planned spend | About 5.8 million intend to shop (about six million last year); average planned spend A$920, up 14.4%; one in four plans to use an AI chatbot to research products and deals | As above | Only 1,838 respondents planned to take part; the AI figure is about product research, not support |
| Online spending, September quarter 2026 | A$25 billion, up 17% on a year earlier; average online basket A$90, down 6.8% and the lowest in the dataset | CommBank iQ, in a CommBank article, 2 Nov 2026 | Card, BPAY and direct debit data plus estimates for buy now, pay later; the dataset is not defined |
| Last year's Black Friday fortnight | Record A$23.8 billion, 19.5% above a typical two-week period | CommBank iQ, as reported in the same article | The article does not give the fortnight's dates |
| Parcels, November and December 2025 | 111 million delivered, Australia Post's busiest peak yet; it is preparing for the same in 2026 | Australia Post, via SBS News, 8 Oct 2026 | Counts parcels, not support contacts |
| Support ticket uplift at peak | No recent public benchmark found | Our searches, 16 Nov 2026 | Use your own last-year ticket counts by day and contact type |
Sources: Roy Morgan ARC finding 10368 (11 Nov 2026); CommBank newsroom, Black Friday test looms as online spending jumps but baskets shrink (2 Nov 2026); SBS News (8 Oct 2026). The ARC forecast and the CommBank iQ figures measure different things and should not be added together.
The Australian Retail Council and Roy Morgan forecast a record A$7.2 billion for 27 to 30 November, but with 5.8 million shoppers against about six million last year, the record rests on higher planned spend per person. CommBank iQ found last year's Black Friday fortnight ran 19.5% above a typical fortnight, and Australia Post is preparing for another peak after 111 million parcels in November and December last year. Support demand follows the parcels, and returns follow the gifts.
We found no recent public benchmark for how far support tickets rise at peak, and the only figures we found were a 2019 analysis of ten brands, which we left out. Your own ticket counts from last Black Friday, by day and by contact type, are a better forecast than any industry average.
Is it too late to staff for Black Friday 2026?
Too late to hire, not too late to prepare. Black Friday is Friday 27 November, 11 days from today, and anyone recruited now will not be trained before it passes.
What still helps is cheap. Write ready answers for your ten most common questions, publish delivery cut-offs (Australia Post's Christmas last-sending dates for Parcel Post run from mid-December to 22 December depending on destination, with Express to 23 December, so check its site), email affected customers early when a parcel slips, show a visible route to a person, and state honest response times. Agree overtime with existing staff now.
If you already have a team in the Philippines, note that Monday 30 November, Cyber Monday, is Bonifacio Day there, a regular holiday. Anyone who works it is entitled to a holiday premium, so agree cover and pay in advance.
How should e-commerce brands staff for the 2027 peaks?
Build a trained standing team for the baseline, let software take the simple contacts, and keep a small flexible layer for the burst. Table 3 shows the dates that matter, including where Philippine holidays collide with Australian trading.
Table 3: Peak trading dates and Philippine public holidays, 2026 and 2027
| Event or date | 2026 | 2027 | Philippine holiday status |
| Black Friday | Fri 27 Nov | Fri 26 Nov | No holiday either year |
| Cyber Monday | Mon 30 Nov | Mon 29 Nov | 2026: Bonifacio Day, a regular holiday, on the same day. 2027: no holiday on the day, but Bonifacio Day (regular) is Tue 30 Nov |
| Feast of the Immaculate Conception | Tue 8 Dec | Wed 8 Dec | Special non-working day both years |
| Christmas Eve | Thu 24 Dec | Fri 24 Dec | Special non-working day; Philippine 13th month pay is due by 24 Dec |
| Christmas Day | Fri 25 Dec | Sat 25 Dec | Regular holiday |
| Boxing Day | Sat 26 Dec | Sun 26 Dec | No holiday either year |
| Rizal Day | Wed 30 Dec | Thu 30 Dec | Regular holiday |
| Last day of the year | Thu 31 Dec | Fri 31 Dec | Special non-working day |
Sources: Philstar, 2026 holidays released (Proclamation 1006, 5 Sep 2025); Proclamation 1427 for 2027, read via the Lawphil transcription, which lists 24 December under a special working day heading although its recitals and press reports call it special non-working (confirm in the Official Gazette); Finder (12 Nov 2026) and ARC for the 2026 sale dates. Black Friday 2027 is the fourth Friday of November. Eid holidays are not yet dated. A worker who works a Philippine holiday is paid a premium, so agree cover and pay in advance.
Size the standing team to an ordinary month plus the predictable uplift. A Philippines-based agent can be rostered to cover Australian evenings and weekends, with Manila three hours behind Sydney while daylight saving runs. The burst is a separate question. A directly hired offshore agent is an ongoing role, so it suits the baseline better than a six-week spike. For a pure spike, a local temp, an overtime roster or a managed seasonal team may fit better.
Work back from Friday 26 November 2027. We would start the search around June and aim to have anyone new in seat by the end of August, which leaves about twelve weeks of ordinary traffic to learn your products, tone and tools before the peak. Use a smaller sale in October or early November as a rehearsal. That timeline is our recommendation, not a measured benchmark.
What does peak cover cost: a local temp or an offshore agent?
A full year of one offshore agent costs about 12% more than one local agent for a single 13-week stint, and the offshore agent is still there in February. Table 4 shows the modelled figures.
Table 4: Modelled cost of adding one support agent: a local temp for 13 weeks, a local hire, and an offshore agent for 12 months
| Option | Modelled cost | What it assumes | Caveat |
| One local agent for 13 weeks (pro rata) | A$20,311 | A$60,000 salary: super 12%, NSW payroll tax 5.45%, workers' compensation 1%, a desk (A$715 a month) and A$1,200 of equipment, pro rated from A$81,242 a year | Leaves out recruitment, training, any casual loading or labour-hire margin and public holiday penalties, so a real temp costs more |
| Same, small-business variant | A$17,250 | No desk and below the payroll tax threshold, pro rated from A$69,000 a year | Same exclusions |
| One local agent, permanent, first year | A$93,242 | A$81,242 steady state, plus A$1,200 of extra first-year equipment and an 18% recruitment fee (A$10,800) | The 18% fee is our assumption within a typical 15% to 25% range |
| One offshore agent, 12 months | A$22,800 | A$1,400 monthly rate plus the flat A$400 monthly management fee (A$21,600), plus A$1,200 of software seats | Leaves out the one-time placement fee, which is unpublished; an ongoing role, not a seasonal one |
Source: Pear Tree's own model, an estimate and not independent data. The admin/VA row (A$60,000 local salary; A$1,400 offshore monthly rate) stands in for a support agent, and actual pay varies by role and experience. In steady state a local hour costs A$49 (1,664 productive hours) against A$15 offshore (1,498 hours after a 10% productivity discount). The offshore 12-month figure is 12% above the local 13-week figure (32% above the small-business variant).
The local figure is generous to the local side, because it leaves out recruitment, training, any casual loading or labour-hire margin and public holiday penalties, so a real temp costs more. In steady state the model puts a local hour at A$49 and an offshore hour at A$15. The comparison only favours offshore if you can use the person across the year. If the quiet months would leave an agent idle, a temp or overtime is cheaper. Ramp-up matters too: ELMO's 2026 benchmark puts the time before a new hire is productive at 44 days across all roles, which would leave a temp hired on 1 November productive around mid-December.
What do agents need to know before the returns wave?
How to tell a fault from a change of mind. The ACCC says businesses cannot take away the right to a refund or replacement for faulty products, and that policies such as "no refunds" or "no refunds or exchanges on sale items" are illegal to rely on. For a major problem the business must offer a refund or replacement, for a minor one it must fix the problem, and a change-of-mind policy only binds you if you have one. Train agents on that split and give them a clear escalation path.
Customer data matters as well. If your business is covered by the Privacy Act, APP 8 and section 16C keep you accountable for how overseas staff handle personal information, even where you took reasonable steps, so use a written contract, least-privilege access, two-factor authentication and a VPN. This is general information, not legal advice.
The bottom line for Australian e-commerce brands
Shoppers will let software handle status checks and often returns, but they want a person when an order goes wrong, and that evidence comes from vendors with unpublished samples. Plan a standing baseline, deflect the simple contacts, keep a flexible burst layer and rehearse before the peak. Pear Tree places directly hired support agents from the Philippines and South Africa for a one-time placement fee plus a flat $400 a month per hire, which suits a baseline and not a six-week spike. We sell offshore support, so test any provider against your own ticket history.
Frequently asked questions
Do Australian shoppers prefer a human or a chatbot for customer support?
For problems, mostly a human. In Twilio's 2026 survey, 83% of Australian consumers tried to bypass an AI agent to reach a person, and in a Sinch survey only 2% would choose a chatbot to resolve an issue. They accept software for routine jobs: 71% were comfortable with AI processing a return or exchange. All of this is vendor research without a published Australian sample size.
Is it too late to hire customer support staff for Black Friday 2026?
Yes for hiring. Black Friday is Friday 27 November, and anyone recruited now would not be trained in time. You can still write answers for your ten most common questions, publish delivery cut-offs, email customers early about delays, show a clear route to a person and agree overtime with existing staff.
When should an e-commerce brand start planning support for the 2027 peak?
We suggest starting the search around June 2027 and having new agents in seat by the end of August, which leaves about twelve weeks of ordinary traffic before Black Friday on Friday 26 November 2027. That timeline is our recommendation, not a measured benchmark.
How much does an offshore support agent cost compared with a local temp?
In our own model, one offshore agent for 12 months costs A$22,800, against A$20,311 for one local agent for 13 weeks, so about 12% more. The offshore figure excludes the one-time placement fee, and the local figure excludes recruitment, training, casual loading and public holiday penalties. The offshore agent suits a year-round baseline, not a six-week spike.
Do Philippine public holidays clash with Australian peak trading?
Sometimes. In 2026, Cyber Monday on 30 November is Bonifacio Day, a regular holiday in the Philippines, and further holidays fall on 8, 24, 25, 30 and 31 December, so agree holiday cover and pay in advance. In 2027, Bonifacio Day is Tuesday 30 November, the day after Cyber Monday.
About the author
Nick, Co-Founder of Pear Tree, leads sales, partnerships and business development. Pear Tree connects Australian and New Zealand businesses directly with vetted offshore professionals from the Philippines and South Africa, with no agency middleman.