Three Words for the Same Headache

Indonesian healthcare search happens in three vocabulary layers for the same condition, formal medical, standard everyday Indonesian, and colloquial, and a page written only in the formal term misses most of the volume that actually exists. "Cephalgia" is correct. "Sakit kepala" and "kepala pusing" are what people actually type, and together they represent a substantial multiple of the formal term's search volume.

No Indonesian study has measured the exact ratio between these layers for clinical queries. The direction is clear enough to act on regardless.

Three Layers, One Search Box

The pattern repeats across nearly every common condition. A page built around only the clinical term is invisible to the majority of people searching for exactly the thing it's about.

Voice Search Pushes Even Harder Toward the Colloquial Layer

Google's Search Live, the multimodal voice-enabled AI Mode that launched in Indonesia in April 2026, makes the vocabulary problem more urgent, not less. A typed search tends toward the standard layer, "sakit kepala." A spoken query tends toward the colloquial layer by default, because that's simply how people talk, "kenapa kepala saya sering pusing" rather than a clipped keyword. Content built only around the standard and formal layers is already behind the format shift voice search represents, since the query itself increasingly arrives phrased the way a patient would describe the problem to a friend, not the way a textbook would name it.

This same three-layer pattern holds across other common conditions beyond headache and blood sugar. "Usus buntu" (colloquial/standard) sits alongside "apendisitis" (formal) for appendicitis. "Sakit maag" pairs with "dispepsia." "Darah rendah" pairs with hypotension in its formal form. None of these pairs are exotic or rare, they're the everyday vocabulary gap that a condition-page library needs to close systematically rather than case by case whenever someone happens to notice a mismatch.

Getting the words right doesn't automatically produce a conversion, because different phrasings of the same condition sit at completely different points in a patient's decision process. A content strategy built only around vocabulary, without accounting for where each query type sits in that funnel, ends up over-invested in traffic and under-invested in bookings.

Query Type	Example	Conversion to Appointment
Symptom	"demam berdarah gejala"	Low direct, high volume
Condition	"penyakit diabetes tipe 2"	Low direct
Treatment	"operasi batu ginjal biaya"	Medium, cost intent signals comparison stage
Doctor/Specialist	"dokter jantung terbaik Jakarta"	High, near decision stage
Location	"klinik kulit Surabaya"	Very high

The uncomfortable pattern in this table: patients overwhelmingly search symptoms in whichever vocabulary layer feels natural to them, then move toward doctor and location queries only once they've mostly decided to seek care. By the time a patient searches a specific doctor or clinic, most of the informational content opportunity has already done its job or been missed. A vocabulary strategy that only targets the high-volume symptom layer, without also building out doctor and location content in the same three layers, captures attention without capturing the booking.

Scaling This Across an Entire Specialty List

Manually writing three vocabulary layers for every condition a hospital treats doesn't scale past a handful of pages before it becomes unsustainable. The practical approach is programmatic: build one content template per page type, symptom, condition, treatment, that structurally requires all three vocabulary layers plus the formal medical citation, then populate it condition by condition from a single specialty list rather than briefing each page from scratch. The template enforces the discipline; the content team's job becomes filling it in accurately rather than remembering the rule every time.

This is also where regional language earns its place without needing a dedicated team. A single template field for "regional/colloquial variant, if applicable" gets populated only where a genuine, documented term exists, Javanese and Sundanese first given their population share, rather than forcing an artificial regional term into every single condition page regardless of whether one is actually in common use.

Regional Language Adds a Fourth Layer

Javanese and Sundanese medical vernaculars show up in voice queries too, "mumet" for dizzy or headache in Javanese, "oyag" for unsteady in Sundanese. That's not a marginal consideration in a country where Java and Bali together account for more than 60% of the population. No published study has quantified the search volume this layer carries, but the terms themselves are well documented, and dismissing them because they're unmeasured would be a mistake in the other direction, the same mistake as ignoring the colloquial layer because the formal term is easier to cite in a clinical review.

The right amount of investment here is proportionate, not exhaustive. A hospital serving patients primarily in Central Java gains real value from a Javanese-language pass on its highest-traffic condition pages. A hospital in Jakarta serving a more linguistically mixed population may reasonably decide that Bahasa Indonesia's own colloquial layer already captures most of the practical benefit, and that a full regional-language build-out isn't worth the resourcing yet. Both decisions can be correct, depending on where the actual patient base sits.

Timing Compounds the Vocabulary Problem

Getting the words right doesn't help if the content isn't indexed before anyone searches for it. A 2020 study validated a Pearson correlation of 0.9371 nationally between dengue-related search terms and actual reported case counts, with provincial correlations ranging from 0.43 to 0.89. Search spikes precede case increases, which makes Google Trends a genuine early-warning signal, and it means dengue content needs to be indexed before the October to April peak in most of Java, not published once cases are already climbing.

Ramadan follows a different but equally predictable logic, even without the same volume data behind it. Fasting patients managing diabetes or hypertension search for medication-timing adjustments, digestive complaints tied to iftar meals, and fatigue management, all of it clustered in the weeks immediately before and during the fasting month. No Indonesian study has quantified exactly how much search volume this generates, which puts it at UNCERTAIN rather than VERIFIED on the confidence scale used throughout this research. The qualitative signal is strong enough to act on regardless: content addressing these specific concerns, written in the vocabulary layer patients actually use for fatigue and digestive discomfort, should be live and indexed four to six weeks before Ramadan begins, the same lead time recommended for dengue.

---

Frequently Asked Questions

Should we just use the colloquial term instead of the formal one everywhere?

No, use both, on the same page. Lead with the term patients actually search, and include the formal term for clinical accuracy and for the credentialed reviewer's sign-off. The two aren't in competition, a well-built page covers all three layers deliberately.

How do we handle regional language without hiring separate writers per region?

Start with the two or three highest-population regional terms relevant to your patient base, Javanese and Sundanese cover the largest share given Java and Bali's combined population. A single natively-reviewed pass, rather than a dedicated regional content team, is a reasonable starting scope.

Exactly when should dengue content go live relative to the season?

Four to six weeks before the October to April peak in most of Java, to allow indexing time before search volume climbs. Publishing once cases are already rising means missing the window the correlation data shows is predictable.

Does standard keyword research tooling capture colloquial and regional terms well?

Inconsistently. Most tools are built around formal and standard-Indonesian query patterns. Regional and highly colloquial terms often require direct testing, checking actual autocomplete and voice-query behavior, rather than relying on reported search volume alone.

Is there hard data on how much more volume the colloquial term gets over the formal one?

Not for clinical terms specifically, no Indonesian study has measured this precise ratio. The directional evidence, from broader keyword volume patterns, is strong enough to act on, but a headline multiplier figure would be overstating what's actually been measured.

Should doctor and location content also be written in three vocabulary layers?

Doctor and location queries are already close to how patients naturally phrase them, "dokter jantung terbaik Jakarta" doesn't have a meaningfully different formal equivalent the way a symptom does. The three-layer discipline matters most for symptom and condition content, less for the bottom-funnel query types.

Can a template-based approach feel generic across dozens of condition pages?

Only if the template is treated as a fill-in-the-blank shortcut rather than a discipline. The template should enforce structure, all three vocabulary layers, formal citation, clinical review, while the actual clinical content, symptoms, causes, and when to seek care, still needs condition-specific accuracy that a template can't substitute for.

How do we know which regional term is actually in common use versus obscure?

Direct testing against real query behavior, autocomplete suggestions and voice-query patterns specifically, since standard keyword tools underrepresent this layer. A native-speaking reviewer familiar with the target region is a more reliable check than any keyword volume estimate for this specific layer.

Structuring an entire content programme around this, doctor profile pages, review workflows, and the specific schema that ties them together, is the E-E-A-T chapter Found Before They Search walks through in more depth, alongside the programmatic content strategy chapter this vocabulary framework itself draws from. The free chapter is available at arfadia.com/resources/ebook-found-before-they-search, also on Amazon, Google Play Books, and Apple Books.

Hospitals and clinics building out a condition-page library around this structure can start with the content architecture work inside our Healthcare SEO service.

---