We Are Now in the 4th Quarter

A state-of-the-nation for travel, tourism and hospitality — and what owners and operators should do before 31 December

The fourth quarter is not where seasons are won. It is where they are confirmed — or quietly lost.

Nine months ago this industry was still debating whether AI was coming. Today it is arguing about what AI costs, who controls it and where it lives on the P&L. Meanwhile the ground under the business has moved: a conflict in the Middle East, a jet-fuel shock, and a traveller who still travels but now counts every coin.

So I did something I recommend to any adviser who publishes as often as I do. I read my own 2026 writing backwards — 84 papers, briefings and articles — and tested each call against what actually happened. The full report, with every source, is linked at the end. This article is the short version.

A year in three acts

Act one: the ground moved. January opened with capability, not controversy: ten role-based AI playbooks, written on the premise that adoption without literacy is just risk with a subscription. Then March. Conflict in the Middle East hit aviation and confidence. Jet fuel surged more than 120% after hostilities began, peaking at US$1,838 a tonne in early April, and some long-haul fares rose as much as 76% year on year. In March I argued that the result would be “not necessarily less travel, but more selective travel”. That call has held.

Act two: AI got a price tag. By June, HITEC month, the conversation had moved from “is AI coming?” to “what does it cost, and who controls it?” That is when Pertlink introduced Token Cost Per Guest — TCPG — the total cost of the AI a property consumes, divided by the guests it serves.

Act three: the decision layer moved. On 27 August Google began rolling out agentic hotel booking inside AI Mode in the US, with ten launch partners. The partner keeps the transaction; Google keeps the conversation that led to it. September became Pertlink’s busiest month, and most of it was about governance: who decides, who checks, and what an AI agent may do with guest money or guest data.

Where we are now

Demand is flat — and uneven. International arrivals grew just 0.4% in the first half of 2026, and UN Tourism has cut its full-year forecast to 1–2% growth, from the 3–4% it projected in January. The Middle East fell 22% in the half; North-East Asia grew 3%, while South-East Asia fell 1%.

Asia-Pacific hotels are growing on rate, not rooms. Regional RevPAR rose 1.3% in July in local currency. The average hides a split: China, including Hong Kong and Macau, fell 2.9%, while India rose 17.6% and Tokyo 15.7% (Figure 1). STR and Tourism Economics now expect Asia-Pacific RevPAR to grow 3.9% this year — occupancy +0.7%, ADR +3.2% — and 2.7% in 2027. A region pricing its way to growth is in a fragile place if the traveller keeps trading down.

Figure 1 — Asia-Pacific hotel RevPAR, July 2026. Year-on-year change in local currency, selected markets. Source: STR/CoStar data via Truist, reported by Skift Daily Lodging Report, 19 August 2026.

Golden Week confirmed the trade-down. Before the holiday, Trip.com reported bookings for stays of seven nights or longer up 123%, while Dragon Trail found just 5% of outbound travellers expecting to stay in a luxury hotel. The official results followed. China’s Ministry of Culture and Tourism counted 826 million domestic trips; on a comparable daily basis, trips rose 6.3% but spending only 4.3%. By Reuters’ calculation, average spending per trip fell 1.9% to 893.92 yuan — the lowest since 2022. Mainland residents’ outbound trips rose 9.3%, and about 42% of Trip.com’s outbound holidaymakers were away for eight days or more, up from about 29%. Hotels felt it: Jefferies, citing InnHome data, reported RevPAR down 0.5% over 1–5 October. Longer, farther, cheaper per night. That is a pricing brief, not a statistic.

Airlines are carrying fewer people, farther. Asia-Pacific airlines carried 33.5 million international passengers in August, 0.6% fewer than a year earlier, even as passenger-kilometres rose 1.8%.

The Philippines has its data — and is still waiting for its growth. The Department of Tourism’s new public dashboard recorded 4,415,860 visitor arrivals to the end of August, up just 0.78%. The World Economic Forum’s 2026 Travel & Tourism Development Index lifted the country eight places to 57th of 110, and the DOT still expects 6.4 million arrivals this year. But a dashboard tells you what happened. It does not tell a traveller — or the AI assistant acting for one — what is open and bookable tomorrow morning. That layer is still missing.

AI is everywhere except the P&L. In h2c’s study of 113 hotel chains, 91% use AI, 28% have an organisation-wide strategy, and 13% report measurable financial returns (Figure 2). In the State of Distribution 2026 report from NYU SPS, HEDNA and RateGain, fewer than one in ten hotels say AI has cut their manual work by more than 30%. At the Destination AI event in Washington, Hyatt’s senior vice president of data and AI, Pat Nestor, asked the question every owner will ask in budget season: where do the hours saved “live on a P&L sheet”?

Figure 2 — Two sides of the AI gap. Left: h2c AI Opportunity Study 2026 (113 hotel chains). Right: Aven Hospitality survey by Talker Research (4,000 travellers in the US, UK and Australia). The surveys are independent and shown side by side, not combined.

The gap is not peculiar to hotels. In McKinsey’s State of AI 2026 survey of 1,719 participants across industries, eight in ten say AI has improved their own productivity, yet the share reporting any impact on their organisation’s EBIT is unchanged on last year at 37%. And the bill is rising: 93% of organisations in a separate McKinsey survey have exceeded their AI budgets. These are cross-industry figures, not hotel data. The direction carries; the decimals do not transfer automatically.

The booking is moving somewhere else. Travellers are using AI to decide, not yet to commit: in Aven’s survey, 67% had used AI in travel searches, but only 14% would trust it to book a hotel stay on its own. Meanwhile only 22% of hotel chains systematically track how they appear in AI answers, and 33% have taken no specific action to be found or booked there.

The playbook: four plays for 76 days

Seventy-six days is enough time to change a budget, a governance framework and a guest-facing habit. It is not enough time to do everything. So the playbook comes down to four words.

1. Count it. Hours saved are a feeling. Money is a line item.

• Owners and boards: no 2027 AI budget line without the P&L line it moves — room revenue, labour, energy, distribution cost — and the baseline it moves from. Make Token Cost Per Guest a standing KPI beside RevPAR. Hold a kill-or-scale review of every AI pilot by 15 December.
• Operators: every month, turn “hours saved” into a named decision: a shift changed, a task reassigned to revenue work, a response time improved and measured.

2. Claim it. If the machine describes your hotel wrongly, the machine is now your front desk.

• Owners and boards: decide your stance on AI booking channels — direct, via brand or OTA, or not yet — and the minimum guest data you require back. Name one person accountable for the property’s machine-readable truth: facts, policies, fees, amenities, accessibility and rates.
• Operators: ask 20 real guest questions of ChatGPT, Gemini, Google AI Mode and Perplexity. Record every error, fix it at source within two weeks, and repeat monthly.

3. Cap it. Alignment is decided in the lab. Guardrails are decided by you.

• Owners and boards: approve a Decision Envelope for every AI system — what it may decide alone, what needs a human, and what it may never do. Put AI on the board risk register with a named owner. For EU guests, transparency duties under Article 50 of the AI Act have applied since 2 August 2026: guests should know when they are dealing with an AI.
• Operators: track the AI Override Rate — how often people overrule the machine. A rate that rises signals drift; a rate of zero may mean nobody is looking.

4. Prove it. To a guest who counts every coin, and an owner who counts every line.

• Owners and boards: build 2027 on the forecast range, not the hope, with a downside case for a renewed fuel or airlift shock. Price and package for length of stay and multi-destination itineraries, not the one-night trophy booking. Decide in writing which guest moments stay human. Guests are telling you where the line is: in Mews’s survey of 3,250 travellers in Europe and the US, guests preferred a human at every one of nine touchpoints tested — 66% for the front-desk welcome — and 75% said good hospitality depends on human staff.
• Operators: give arrival teams a one-line context card, so the first 60 seconds feel recognised rather than processed.

The full report turns the four plays into an 11-week operator sprint, running from Monday 19 October to 31 December, with a lead and a “done by” test for each stage.

What would change our view

A playbook is a set of bets. Three signals would move ours:

• China’s outbound momentum. If it holds into 2027, Asia-Pacific should plan for longer, value-led Chinese stays, not a luxury rebound.
• Google AI Mode beyond the US. Expansion to Asia-Pacific markets would move “Claim it” from important to urgent.
• Hotel-sector evidence on AI costs. If hotel surveys begin to match the cross-industry finding that one in five organisations is constraining AI on cost, “Count it” becomes the first conversation in every 2027 budget.

The final whistle

The industry spent the first half of 2026 learning what AI is, and the third quarter discovering who controls it. The fourth quarter asks something more ordinary and more demanding: proof — in the numbers and at the front desk — that it was worth it.

The hotels that win the quarter will not be the ones with the most AI. They will be the ones that can show an owner where it sits on the P&L, show an AI agent the truth about the property, show a regulator who decides, and show a guest — in the first sixty seconds — that someone was expecting them.

The intelligence may be artificial. But the experience is human.

About the author. Terence Ronson is Founder and Managing Director of Pertlink Limited, a boutique hospitality technology and AI advisory consultancy headquartered in Hong Kong since 2000 and operating across the Philippines and Asia-Pacific. He is a founding member of PAIBA (the Philippine AI Business Association), Chair of the AI Education & Training Subcommittee of the HFTP AI Collective, and a member of the Collective’s AI Research & Innovation Subcommittee.

Disclaimer. General information only, not professional advice. Pertlink accepts no responsibility for outcomes from following this playbook or the strategies it contains. Take professional advice before acting.

Made with the help of AI tools, but with a HITL.

Sources

1. UN Tourism, World Tourism Barometer release, September 2026 (via Hospitality Net) — https://www.hospitalitynet.org/news/4134434/international-tourism-holds-steady-with-04-growth-in-first-half-of-2026-as-middle-east-conflict-and-rising-costs-weigh-on-momentum

2. Aerohaber, regional detail of the UN Tourism Barometer, September 2026 — https://aerohaber.com/international-tourism-holds-steady-with-0-4-growth-in-first-half-of-2026/

3. Skift Daily Lodging Report (STR/CoStar data via Truist), 19 August 2026 — https://dlr.skift.com/2026/08/19/india-and-tokyo-lead-apac-hotel-performance-in-july/

4. CoStar/STR, Global Hotel Market Forecast Assumptions – Q3 2026, 31 August 2026 — https://www.costar.com/products/str-benchmark/resources/data-insights-blog/global-hotel-market-forecast-assumptions-q3

5. Morgan Lewis, “Impacts on the Aviation Market of the War in Iran”, May 2026 — https://www.morganlewis.com/pubs/2026/05/impacts-on-the-aviation-market-of-the-war-in-iran

6. Skift, “China’s Golden Week Travelers Are Going Farther and Spending Less on Hotels”, 25 September 2026 — https://skift.com/2026/09/25/chinas-golden-week-travelers-are-going-farther-and-spending-less-on-hotels/

7. China News Service (Ministry of Culture and Tourism Data Center), 9 October 2026 — https://www.chinanews.com.cn/cj/2026/10-09/10709628.shtml

8. AsiaOne (Reuters calculations; Jefferies/InnHome data), 9 October 2026 — https://www.asiaone.com/china/china-golden-week-spending-trip-hits-four-year-low-amid-overseas-travel-jump

9. Xinhua (National Immigration Administration data), 8 October 2026 — https://www.xinhuanet.com/politics/20261008/a9cab0a8eda244619079df9ff37837fe/c.html

10.      Shanghai Securities News via Eastmoney (Trip.com 2026 Mid-Autumn and National Day travel report), 7 October 2026 — https://finance.eastmoney.com/a/202610073888869754.html

11. Moodie Davitt Report (AAPA August 2026 traffic results), 28 September 2026 — https://moodiedavittreport.com/asia-pacific-passenger-traffic-softens-in-august-as-airline-cost-pressures-mount/

12.       Philippine Daily Inquirer, “DOT rolls out dashboard tracking Philippine tourism data”, September 2026 — https://newsinfo.inquirer.net/2306203/dot-rolls-out-dashboard-tracking-philippine-tourism-data

13.       The Philippine Star (WEF Travel & Tourism Development Index 2026), 4 October 2026 — https://www.philstar.com/business/2026/10/04/2560712/philippines-tourism-makes-stride-global-ranking-climbs

14.      The Philippine Star, “Philippines on track to hit 6.4 million tourist arrivals target”, 12 October 2026 — https://www.philstar.com/business/2026/10/12/2562494/philippines-track-hit-64-million-tourist-arrivals-target

15.      Hotel News Resource (h2c AI Opportunity Study 2026), 1 October 2026 — https://www.hotelnewsresource.com/article143157.html

16.      NYU SPS, State of Distribution 2026 (with HEDNA and RateGain), 15 September 2026 — https://www.sps.nyu.edu/about/news-and-ideas/articles/press-releases/2026/more-than-50-of-hotels-use-ai-but-under-10-see-real-impact-rategain-nyu-sps-hedna.html

17.       Skift, “‘The P&L Doesn’t Lie’: Hotels See More AI Costs Than Bottom-Line Gains”, 1 October 2026 — https://skift.com/2026/10/01/the-pl-doesnt-lie-hotels-see-more-ai-costs-than-bottom-line-gains/

18.      QuantumBlack, AI by McKinsey, “The state of AI in 2026: On the road to ROI”, 25 August 2026 — https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

19.      McKinsey, “The cost of intelligence: How CIOs can manage AI demand at scale”, 20 July 2026 — https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-cost-of-intelligence-how-cios-can-manage-ai-demand-at-scale

20.     Hotel News Resource (Aven Hospitality / Talker Research), 30 September 2026 — https://www.hotelnewsresource.com/article143136.html

21.       Skift, “Google’s Agentic Hotel Booking Tool Comes to AI Mode”, 27 August 2026 — https://skift.com/2026/08/27/googles-agentic-hotel-booking-tool-comes-to-ai-mode/

22.     Jones Walker, “Yes, August 2 Still Matters” (EU AI Act Article 50), 2026 — https://www.joneswalker.com/en/insights/blogs/ai-law-blog/yes-august-2-still-matters-the-eu-approved-a-high-risk-ai-delay-but-most-trans.html?id=102nbon

23.     Hotel News Resource (Mews guest survey), 7 October 2026 — https://www.hotelnewsresource.com/article143233.html

Terence Ronson

Terence Ronson is the Founder and Managing Director of Pertlink Limited, a boutique hospitality technology and AI advisory consultancy headquartered in Hong Kong since 2000, operating across the Philippines and Asia-Pacific. He chairs the AI Education & Training Subcommittee of the HFTP AI Collective and is an inductee of both the HFTP International Hospitality Technology Hall of Fame and the CHTA Hall of Fame.