Visitor Survey 2025, Two Numbers That Change Plans

In March 2026 the Pacific Tourism Organisation published Tonga's International Visitor Survey for January to June 2025, together with the Ministry of Tourism and the Tonga Tourism Authority. It is the most recent behavioural data available on people who actually travelled to Tonga, and it contains two findings that should reorder the first ninety days of work for any operator selling to overseas visitors.

One of those findings argues against spending money on digital channels. We are going to lead with that one.

Before any of it means anything, though, the survey's limits need stating, because they are unusual and they change how much weight each number can carry.

Read the method before you read the numbers

The survey collected 88 valid responses from 543 invitations. That is a 16% response rate, and those 88 people represent roughly 1.5% of actual visitor arrivals during the collection window. Data was gathered between February and April 2025. January, May and June were not captured at all, for operational reasons, although the full six-month arrivals count was still used to estimate economic impact.

Eighty-eight people. Hold that number in mind for everything that follows.

This is not a criticism of the Pacific Tourism Organisation. Running visitor surveys in small markets is genuinely difficult, and publishing the sample size and the coverage gaps openly is exactly what a credible statistical agency does. The problem arises downstream, when the figures get lifted into marketing material with the caveats stripped out. During research for this project, one AI-generated market report quoted three of these percentages, attributed them to the Tonga Ministry of Tourism, labelled them verified, and made no mention of the sample size at all. That is how an 88-person survey becomes a fact.

So: treat directional findings as reliable, treat precise percentages as indicative, and never let a number from this survey stand in a proposal without the sample size beside it.

The 39% figure, and why we publish it anyway

Thirty-nine percent of respondents used at least one online channel, meaning social platforms or online travel booking sites, to inform their decisions. Sixty-one percent used none.

An agency has an obvious incentive to bury that. We put it on the service page and in the proposal, for two reasons.

The first is arithmetic. If you scope a digital programme as though it can reach the whole visitor population, you are budgeting against an audience roughly two and a half times larger than the one that exists. When results come in below the implied expectation, the client concludes the agency underperformed, when the actual failure happened at the scoping stage. Publishing the constraint up front makes the target honest.

The second is that it changes what you should measure. On a 39% reachable share, traffic volume becomes close to meaningless as a success metric. What matters is booking value from the people you did reach. That is a harder thing to report and a much more useful one.

There is a caveat worth stating in the other direction, too, because fairness cuts both ways. The survey asked about channels used to inform the decision. It did not measure whether a traveller was influenced by information they encountered without recalling the channel, and self-reported channel attribution is known to underrepresent search specifically, because people frequently do not remember searching. So 39% is probably a floor rather than a ceiling. We still plan against it, because planning against the floor is what conservative scoping means.

Short planning cycles change the build order

More than a third of respondents planned their trip less than one month before departure. For a long-haul destination, that is fast.

Think about what that traveller needs. They are deciding within weeks, probably comparing two or three options, and they need specific operational facts right now: is it running, what does it include, how do I get there, what happens if the weather turns, can I still book. They are not going to read a seasonal essay about the region. They are not going to enter an email nurture sequence and convert in month four.

So the build order inverts relative to a standard content programme.

Asset	Value under a short planning cycle	Priority
Complete, accurate Google Business Profile	High. Surfaces at the decision moment, carries hours, location, photos and reviews in one place	First
Dated operational fact pages, licence, capacity, transfers, season	High. Exactly what a three-week-out traveller is verifying	First
Clear availability and booking path	High. Friction at this stage loses the booking outright, there is no second visit	First
Review volume and recency on the platforms people check	High. 60% are first-time visitors with no prior knowledge of you	Second
Comparison content, this operator versus that one	Moderate. Useful, but the traveller often arrives having already shortlisted	Second
Long-form destination inspiration content	Low. Competes with national publishers who will win, and reaches people earlier than this audience decides	Later or never
Email nurture sequences	Low for acquisition under a 30-day cycle. Useful for repeat and referral instead	Repurpose

That table is not a general truth about tourism marketing. It is what follows specifically from a documented short planning cycle in a market where 60% of visitors have never been before. Change the planning cycle and the priorities change with it.

Who these visitors are, and what it implies about tone

The survey fills in a portrait that is more useful than the headline percentages.

Average stay 6.8 nights. Around USD 148 per person per day. Sixty percent first-time visitors. Nearly half travelling alone. Average reported household income of USD 76,405. Nearly four in five would recommend Tonga, and more than 80% would consider returning. Purpose of visit split as holiday 49%, business 21%, visiting friends and relatives 19%. Tongatapu received 93% of visitors, Vava'u 16% and Ha'apai 4%, with the overlap explained by multi-island trips.

A few things follow from that profile.

The income figure suggests this is not a price-led audience. Discount-driven positioning is probably working against you. What a household at that income level optimises for is certainty: that the thing will run, that the operator is licensed, that the logistics work.

Nearly half travelling alone is more actionable than it looks. Solo travellers have different concerns around safety, single supplements, joining group departures and whether a minimum number is required to run. Most operator content is written for couples and families. Addressing solo logistics explicitly is a low-cost differentiator that almost nobody bothers with.

And the 93% Tongatapu figure against 16% for Vava'u tells you where the underserved intent sits. Vava'u carries the whale-swimming season and a large share of the charter market, but only a small fraction of visitors reach it. Whether that reflects demand or access friction is a question the survey cannot answer, and an operator in Vava'u should probably find out before scaling content spend.

The spending figures, and an arithmetic problem

The survey estimates USD 24 million in economic impact across January to June 2025. Roughly 40% of prepaid expenditure is estimated to flow into the local economy, the rest being captured by overseas intermediaries.

Then it reports prepaid spend of USD 1,055 per person and in-country spend of USD 586 per person, against a stated total average of USD 1,009 per trip.

Those do not add up. USD 1,055 plus USD 586 is USD 1,641, not USD 1,009.

The most likely explanation is different denominators, that is, not every respondent reported both components, so the averages are computed over different bases. That is a normal thing to happen in survey reporting. What matters practically is how you use the figures: cite the components separately with their labels, never sum them, and do not present all three side by side as if they were internally consistent. If you need a single per-day figure, the survey's own USD 148 per person per day across 6.8 nights is the safer construction.

We flag this rather than quietly picking whichever number suits the argument. An operator who builds a revenue model on USD 1,641 per visitor and then discovers the discrepancy has been badly served.

Institutional and government work runs on different rules

If part of your revenue comes from tenders, the arithmetic changes again, and there is a documented inconsistency worth knowing before you plan around it.

Tonga's Public Procurement Regulations 2015 set thresholds in pa'anga: a request for proposals applies between TOP 7,500 and TOP 49,999.99, an expression of interest followed by an RFP applies between TOP 50,000 and TOP 149,999.99, and international expressions of interest with an RFP apply from TOP 150,000 upward. A separate Treasury Instructions document gives a different table, with breakpoints at TOP 100,000 and TOP 500,000.

Both are Tongan government documents. They do not agree. Two independent research passes for this project flagged the same conflict separately, which suggests it is a real inconsistency in the published record rather than a misreading. The practical instruction is simple: do not let any provider present a single threshold to you as settled, and confirm the applicable figure with the procuring entity before scoping a bid.

Value band	Public Procurement Regulations 2015	Treasury Instructions table
Lower band	RFP from TOP 7,500 to TOP 49,999.99	Breakpoint stated at TOP 100,000
Middle band	EOI plus RFP from TOP 50,000 to TOP 149,999.99	Not directly comparable
Upper band	International EOI plus RFP from TOP 150,000	Breakpoint stated at TOP 500,000
Status	Unreconciled. Both are Tongan government documents. Confirm with the procuring entity rather than relying on either in isolation.

There is a second institutional feature worth understanding, because it is competing for the same budget you are.

Donor-funded digital work is already happening in Tonga. The Market Development Facility, funded by Australia's foreign affairs department with New Zealand co-funding, is active in the country, and the Tonga Tourism Authority engaged a New Zealand agency, Tomahawk, for a rebranding programme through it during 2025. Separately, the Pacific Tourism Organisation's Pacific Digital Tourism Transformation Project delivered website and search improvements to five national tourism organisations including Tonga under Activity 3.2 of its second phase, with support running to August 2025 and Trove Tourism Development Advisors as lead specialist.

Two implications. First, there is a funding route that does not come out of an operator's own pocket, and it is worth asking about before assuming a programme is unaffordable. Second, the precedent for foreign agencies delivering digital work into Tonga is already established at national level, which removes the novelty objection from a procurement conversation before it is raised.

The macroeconomic backdrop supports patience rather than urgency. The IMF's 2025 Article IV assessment, concluded by its Executive Board on 10 November 2025, put FY2025 growth at 2.7% with FY2026 projected at 2.3% and medium-term potential at 1.2%. Headline inflation had eased to 1.4% by June 2025, below the National Reserve Bank of Tonga's 5% reference rate, while core inflation had risen to nearly 10%. That is an economy growing slowly, with underlying price pressure, and reconstruction and remittances doing most of the work. Nothing in it argues for aggressive spending, and everything in it argues for cheap structural fixes ahead of expensive programmes.

What the survey cannot tell you

Worth being explicit about the gaps, because they are where most of the overclaiming happens.

It does not measure search behaviour. There is no data on what people searched, which engine they used, or whether they encountered an AI-generated answer. Anyone extrapolating search volume from this survey is inventing.

It does not cover cruise visitors, who numbered 16,214 across eight lines in 2024 and behave completely differently, arriving with hours rather than days and buying accordingly.

It does not represent January, May or June, which excludes both the start of the year and the beginning of the whale season.

It cannot support sub-segment analysis. With 88 responses, splitting by market and purpose leaves cells too small to mean anything. If someone shows you a breakdown of Australian business travellers from this survey, that is a handful of people.

And it says nothing about conversion rates, cost per acquisition or channel attribution, because it was not designed to.

What to actually do with it

Three things, in order.

Rescope against 39%. Whatever your digital target is, model it against the reachable share rather than the full visitor count, and state that assumption in writing. If the numbers still work, you have an honest plan. If they do not, you have found that out before spending rather than after.

Build for the thirty-day decision. Business profile, dated operational facts, frictionless booking path, current reviews. Push inspiration content down the list. It competes against publishers you cannot outrank and reaches people earlier than this audience decides.

Report in booking value, not traffic. On a base this size, traffic is noise. Enquiries, booking requests and revenue in your banking currency are the only figures that survive scrutiny, and every percentage you report should carry its numerator and denominator alongside it.

The survey is a good piece of work with clearly stated limits. Used honestly it is the most useful planning document available for this market. Used carelessly, it becomes 88 people speaking for an entire country.

Tessar Napitupulu writes about scoping visibility programmes against verified audience size rather than aspirational reach in Found Before They Search.

---

Frequently Asked Questions

How large was the sample in Tonga's 2025 visitor survey?

Eighty-eight valid responses from 543 invitations, a 16% response rate representing approximately 1.5% of actual visitor arrivals during the collection window. Data was collected between February and April 2025, and January, May and June were not captured. Directional findings are reliable; precise percentages should be treated as indicative and always quoted with the sample size.

What share of visitors to Tonga used online channels to plan?

Thirty-nine percent used at least one online channel, meaning social platforms or online travel booking sites. Sixty-one percent used none. The figure is probably a floor rather than a ceiling, because self-reported channel attribution tends to underrepresent search, but conservative scoping means planning against the floor.

How far in advance do visitors to Tonga book?

More than a third of respondents planned their trip less than one month before departure. For a long-haul destination that is a short cycle, and it favours accurate dated facts, a complete business profile and a frictionless booking path over long-horizon inspiration content and email nurture sequences.

How much does a visitor to Tonga spend?

The survey reports approximately USD 148 per person per day across an average stay of 6.8 nights, and estimates USD 24 million in economic impact for January to June 2025. It also reports USD 1,055 prepaid and USD 586 in-country per person against a stated trip average of USD 1,009, figures which do not reconcile arithmetically, most likely because of differing denominators. Cite the components separately and do not sum them.

Which islands do visitors to Tonga actually reach?

Tongatapu received 93% of visitors, Vava'u 16% and Ha'apai 4%, with totals exceeding 100% because many trips cover more than one group. Vava'u carries the whale-swimming season and much of the charter market while reaching roughly one visitor in six, a gap that could reflect either demand or access friction. The survey cannot distinguish between the two.

Does the survey tell us anything about search behaviour?

No. It contains no data on what people searched, which engine they used, or whether they encountered AI-generated answers. Any search volume estimate extrapolated from this survey is invented. It also excludes cruise visitors, who numbered 16,214 across eight lines in 2024 and behave differently, arriving with hours rather than days.

Can we break the survey down by market and purpose?

Not reliably. With 88 total responses, splitting by source market and travel purpose leaves cells containing a handful of people, which cannot support conclusions. A breakdown of, for example, Australian business travellers from this dataset describes a very small number of individuals and should not be treated as a segment profile.

What visitor profile does the survey describe?

Sixty percent first-time visitors, nearly half travelling alone, average reported household income of USD 76,405, average stay 6.8 nights. Purpose split as holiday 49%, business 21%, visiting friends and relatives 19%. Nearly four in five would recommend Tonga and more than 80% would consider returning. The income level suggests an audience optimising for certainty rather than price, and the solo travel share is high enough that addressing single-traveller logistics explicitly is a genuine differentiator.