Keyword Research for a Place Nobody Searches From

Type a Saint Helena query into any mainstream keyword tool with the country set to Saint Helena and you get nothing. Not a small number. Nothing, or a dash, or a zero that is really a rounding artefact.

The instinctive reading is that there is no market. The correct reading is that the tool is pointed at the wrong country. Nobody on Saint Helena is searching for a reason to visit Saint Helena. The 889 non-St Helenian leisure visitors recorded in the twelve months to October 2024 all searched from somewhere else, and the somewhere else is where the research belongs.

This sounds obvious written down. It is not what most agencies do, and the failure is visible in the output: pages targeting a territory of 3,980 people with keywords nobody can evidence, reported against Google Trends data that turns out to describe VPN routing.

Point the tool at the origin, not the destination

The government's own tourism strategy names its priority source markets, and the answer is not surprising: the United Kingdom and South Africa. Air access reinforces it. Airlink operates a weekly service from Johannesburg, with a seasonal Cape Town rotation and an inter-island Ascension connection roughly every four weeks. If you cannot get here from a country without transiting Johannesburg, that country is a secondary market by structure rather than by choice.

Older visitor-origin data, from a 2020 analysis published in an academic hospitality and tourism journal, put the United Kingdom at around 37 percent of visitors, South Africa at 21 percent, other European origins at 13 percent, and the United States and Caribbean at 9 percent. Treat those as historical shape, not current share. They predate the pandemic collapse and the subsequent partial recovery, and a distribution measured on 2,481 pre-pandemic tourists does not transfer cleanly to a base of 889.

What the numbers do establish is a research configuration. Run everything against United Kingdom and South African settings, in that order of weight. Add Germany, the Netherlands and the United States as tertiary sets if there is budget for it. Never run it against the territory.

The Google Trends trap, in detail

There is a specific reason Trends fails here, and understanding it prevents an entire category of error.

Trends does not report how many people searched. It reports the proportion of searches in a region that matched the term, scaled to 100. When the denominator is a few thousand people, a small absolute movement produces an enormous proportional one.

Now add the second factor. A meaningful share of traffic attributed to this territory does not originate here at all. It arrives through VPN exit nodes and infrastructure associated with .sh addressing. Internet Society Pulse recorded roughly 44,150 registered .sh domains, and the overwhelming majority have nothing to do with the island, because .sh reads as shell and makes an attractive domain hack.

Combine the two and you get the result that has been reported repeatedly since at least 2021: Saint Helena appearing at or near the very top of worldwide interest tables for cryptocurrency, and ranking high for cybersecurity. Checked again in August 2026, the territory still sits at the top of the twelve-month worldwide table for cryptocurrency. A place with 3,980 residents does not lead the planet in crypto search interest. The chart is describing a VPN configuration, and any content plan built on it is planning for people who were never there.

Research task clusters, not keywords

With volume data unavailable in the destination and unreliable in aggregate, the unit of research has to change. Build around the decisions a prospective visitor actually has to make, in the order they make them, and let the source-market tools tell you how each decision is phrased in Britain and in South Africa.

Some of these clusters are large in the source markets even though they are invisible in the destination. Flight and access questions in particular attract far more searching than the visitor numbers would suggest, because a great many people research this trip and never take it.

Cluster	Where the searching happens	Why it matters more than it looks
Getting there	United Kingdom, South Africa	The decision that kills the trip. Also the cluster where third-party pages publish routes that do not exist
Entry permits and paperwork	United Kingdom, Europe	Guidance exists in several places and does not always agree. High risk of a stale answer being retrieved
Money, cards and currency	All source markets	The Saint Helena pound at parity with sterling confuses people who have never encountered a pegged local currency
Napoleonic heritage	United Kingdom, France, Germany	The one topic where the territory has genuine unprompted recognition. French-language demand is real and usually ignored
Diving and whale sharks	South Africa, United Kingdom, Europe	Seasonal, high-intent, low-volume. Small numbers of committed travellers rather than casual interest
Walking and endemic wildlife	United Kingdom, Germany, Netherlands	Serves a specialist audience that reads long content and books far ahead
Relocation and work	United Kingdom, South Africa	Different audience, different rules, and constantly conflated with tourist information on the same pages
Connectivity for longer stays	United Kingdom, Europe	Matters to anyone staying weeks rather than days, and the picture changed materially after 2023

Validating a cluster without volume data

If the number is unavailable, evidence has to come from elsewhere. Four sources, none of them a keyword tool.

Read the results page itself. If forums, community threads and long-form independent pages dominate a query, real people are asking it and nobody has answered it well. That is a stronger signal than a volume estimate, because it shows unmet demand rather than measured traffic.

Follow the structure of access. A weekly flight from one hub imposes a research pattern. People search around departure days, around the seasonal Cape Town rotation, and around cruise calls. Seasonality can be inferred from the schedule even when it cannot be measured.

Use the government's own publications as a demand map. Which pages exist, which get updated, which questions the administration keeps answering. A page rewritten three times is answering something people keep asking.

Ask the operators. A dive operator, a guesthouse, a tour guide each know what they are asked before a booking. In a market with a handful of providers, five conversations cover most of the sector. That is not scalable research anywhere else, and here it is the most reliable input available.

One caution about the first of those four. A results page for a Saint Helena query is itself contaminated by the naming problem, so before treating composition as a demand signal, check which St Helena the results are about. A query that returns Napa Valley restaurants and a Colchester charity has not told you anything about unmet demand in the South Atlantic. It has told you the query needs a geographic qualifier before it can be researched at all.

Two things worth knowing about how the searching happens

Beyond which country to point the tool at, two published findings change how a cluster should be written rather than which cluster to pick.

The first is that generative tools are now part of travel research, at least in one measured market. Phocuswright reported in 2025 that 39 percent of United States travellers used generative AI for travel research, up from 28 percent. That figure is American and should not be read as a Saint Helena number or a British one. What it establishes is direction: a share of the research that used to land on a results page now lands inside a generated answer instead, which means the content has to be extractable and not merely rankable.

The second is more actionable. Pew Research Center analysis in 2025 found that around 60 percent of queries beginning with a question word produced an AI summary, against roughly 36 percent of full-sentence queries. Most of the clusters in the table above are naturally phrased as questions. How do I get to St Helena. Do I need a permit. What currency do they use. That phrasing is exactly the shape most likely to trigger a generated answer, which raises the value of a direct answer in the first two sentences of a page and lowers the value of a long scenic introduction.

Worth adding a note of proportion here. Visitor spending across 2024 sat somewhere between £4.9 million and £6.7 million, and daily spend per visitor was measured at roughly £150 to £190 back in 2019. Those are real numbers attached to a small number of people, which is the honest justification for research effort in this market: each visitor is worth a lot relative to the size of the audience, and there are very few of them.

One more hazard to watch when reading a results page. Several agency websites publish pages targeting Saint Helena with no verifiable client, contract or office in the territory, and some of them mix the South Atlantic territory with the Californian town in the same page. If a query returns a stack of those, do not read it as competition or as evidence of demand. It is evidence that a keyword with almost no volume is cheap to target, which is not the same thing.

Language, and one detail worth getting right

British English is the baseline. Both primary source markets use it, and the territory itself does. South African usage overlaps closely enough that a single variant serves both, with vocabulary adjustments rather than spelling ones.

The detail that gets missed is French. Napoleonic heritage is the one subject on which this territory has genuine unprompted international recognition, and a substantial share of that interest is French-speaking. Nobody serving this market is competing seriously for it. That is not a large opportunity in absolute terms. In a market where nothing is large in absolute terms, it stands out.

What the method cannot tell you

Three honest limits.

Source-market volume describes interest in the topic, not intent to travel. A great many people read about this island with no intention of spending two days in transit to reach it. The ratio of researchers to visitors is unknowable and probably large. A cluster that looks healthy in a United Kingdom volume report may consist almost entirely of people who will never board the flight.

Historical origin shares are shape, not share. A distribution measured on 2,481 pre-pandemic tourists cannot be applied to 889 without an assumption nobody can test.

And nothing in this method produces a forecast. It produces a defensible list of topics to write about, in the languages and regions where the people who might travel actually are. That is less than a client usually wants and considerably more than a zero from a keyword tool. The way these clusters are built into pages is set out on our SEO service page for Saint Helena.

---

Frequently Asked Questions

Why do keyword tools show no search volume for Saint Helena?

Because they apply minimum-volume reporting thresholds and round small numbers toward zero, and a territory of 3,980 residents never clears them. The absence is a reporting artefact rather than evidence about demand. The correct response is to change the country setting: the people who search about visiting Saint Helena are in the United Kingdom, South Africa and Western Europe, and volume for those markets is reportable.

Which source markets should be researched?

The United Kingdom first, then South Africa. Both are named priority markets in the government's tourism strategy, and air access reinforces it, since the scheduled route runs weekly from Johannesburg with a seasonal Cape Town rotation. Germany, the Netherlands, France and the United States are reasonable tertiary sets. Historical origin data from a 2020 analysis put the UK at around 37 percent and South Africa at 21 percent, but that predates the pandemic and should be treated as shape rather than current share.

Can Google Trends be used for Saint Helena at all?

No, and no setting inside Trends fixes it. Trends reports the share of queries in a region rather than the count, and a large portion of traffic attributed to this territory arrives through VPN exit nodes and infrastructure associated with .sh addressing. The result is that the territory has repeatedly ranked first worldwide for cryptocurrency interest and high for cybersecurity. Rechecked in August 2026, it still leads the trailing twelve-month worldwide table for cryptocurrency.

Why are there so many .sh domains if the territory is so small?

Because .sh reads as shell, which makes it an attractive domain hack for technology projects and shorteners. Internet Society Pulse recorded roughly 44,150 registered .sh domains, and the overwhelming majority have no connection to the island. Any system inferring geographic relevance from the country-code domain is being fed noise at scale, which is one reason traffic attribution for this territory is unreliable.

How can a keyword cluster be validated without volume data?

Four substitutes, used together. Read the results page: forums and independent long-form pages dominating a query indicate real questions that nobody has answered well. Infer seasonality from the flight schedule and cruise calendar rather than from Trends. Treat the government's publication history as a demand map, since pages rewritten repeatedly are answering repeated questions. And talk to the operators, because in a sector with a handful of providers, five conversations approach a census.

Is French worth targeting?

For one topic, yes. Napoleonic heritage is the subject on which this territory has genuine unprompted international recognition, and a significant share of that interest is French-speaking. Nobody appears to be competing seriously for it. The absolute opportunity is small, which in a market where nothing is absolutely large makes it comparatively interesting.

Does AI search change which keywords matter?

It changes how the content should be written more than which clusters to choose. Pew Research Center analysis in 2025 found around 60 percent of queries beginning with a question word produced an AI summary, against roughly 36 percent of full-sentence queries, and most clusters for this territory are naturally question-shaped. That raises the value of a direct answer in the opening two sentences. Phocuswright reported 39 percent of United States travellers using generative AI for travel research in 2025, up from 28 percent, which is a United States figure and not a Saint Helena one.

What does source-market research not tell you?

It does not distinguish interest from intent. Many people read about this island with no plan to spend two days in transit reaching it, and the ratio of researchers to travellers cannot be established. It also produces no forecast. What it produces is a defensible list of topics, phrased the way people in the source markets phrase them, which is the realistic ceiling for research in a market with 889 leisure visitors a year.

---