Keyword research for a herbal brand, a method that survives the claims rules

Run standard keyword research on a herbal supplement category and the output is predictable: a list sorted by volume, with the highest-volume terms being the ones you are legally not allowed to write about.

“Ashwagandha for anxiety.” “Giloy for immunity.” “Best medicine for diabetes.” “Herbal cure for PCOD.”

High volume, clear intent, and unusable, because writing the page that answers those queries means making the claim.

So the standard method hands a herbal brand a list it cannot act on, and brands respond in one of two bad ways: they write the pages anyway and get the listing pulled, or they give up on search entirely.

There is a third option, and it is a different research method rather than a different attitude.

This is orientation, not legal or regulatory advice. Permitted claims change. Have every piece reviewed by a competent regulatory consultant before publishing.

Step 1: separate the query into its four parts

Every supplement query carries up to four components, and only some of them are claim-bound.

The subject, ashwagandha, triphala, whey protein. Never claim-bound.

The condition, anxiety, diabetes, PCOD, immunity. Always claim-bound.

The practical question, dose, timing, duration, combination, interaction, how to take, how to store. Never claim-bound.

And the commercial question, which brand, how much, where to buy, is it worth it, how to tell a fake. Never claim-bound.

The whole method follows from this: you harvest the subject and then build on the practical and commercial components, discarding the condition component.

And the crucial finding, which is counter-intuitive: the practical queries convert better. A reader asking “ashwagandha kitne din tak leni chahiye” is already past the decision to buy. A reader asking “best herb for anxiety” is at the start of a long journey and will probably go to a doctor or a bigger brand.

Step 2: harvest from where people actually ask

Keyword tools are a starting point and a poor one in this category, because they under-report Indian language queries and long practical questions.

Better sources, in order of usefulness.

Autocomplete and related searches, in all three registers, English, Devanagari and Roman Hindi. Type “ashwagandha” and then each vowel and consonant after it, and record what appears. This is tedious and it is the single most productive hour in the process.

“People also ask” boxes, which are literally a list of answerable questions.

Your own customer support and WhatsApp messages. The questions customers actually ask, in their own words, are the highest-value keyword source you have and the one nobody mines. If you have 200 enquiry messages, that is a content plan.

Marketplace question-and-answer sections on competitor listings. People ask things there they would not type into a search box, and the questions are specific to the product form.

Marketplace reviews, read for complaints, “scoop missing”, “cap leaked”, “no smell”, “expiry too close”. Each complaint is a query somebody else will search before buying.

Practitioner and pharmacy conversations, if you have access. What patients ask repeatedly.

And the glossary problem, terms people encounter on labels and do not understand. “What does 10:1 mean”, “what is a proprietary blend”, “65% boswellic acids meaning”. These have modest volume and extremely high intent, because the person is holding a product.

Step 3: classify by whether you can answer it

Three buckets, and the sorting is the whole discipline.

Answerable fully. Practical and commercial questions. Dose, timing, form comparison, label reading, storage, authenticity, price, cost per serving, how to verify a batch certificate.

Answerable partially, with care. Questions touching a condition where the honest answer is about the ingredient rather than the outcome, and where the piece must include when to see a doctor rather than positioning itself as an alternative. These need review before publishing, every time.

Not answerable. Direct cure, treatment and disease-management queries. Do not write these, do not write euphemised versions, and do not try to rank for them with a page about something else.

That last instruction matters: targeting a prohibited query with an evasive page is worse than not targeting it, because the page is incoherent and the intent is visible to anyone reviewing it.

Step 4: map to intent, not to volume

Sort the answerable list by what the searcher is trying to do.

Informational, what is this, what does this word mean. Builds authority, converts slowly.

Practical, how do I take it, when, how long, with what. Converts well, and is where this category’s real opportunity sits.

Comparison, this versus that, capsule versus powder, brand A versus brand B. High commercial intent.

Verification, is this genuine, how do I check, what should the label say. Very high intent, and almost nobody serves it.

And transactional, price, where to buy. Highest intent, lowest volume, and usually dominated by marketplaces rather than brand sites.

The practical and verification buckets are where a small brand can win, because the large brands write informational and transactional content and ignore the middle.

Step 5: check the language register per query

Not per site, per query. We have written about this separately and it is the step most often skipped.

The same question exists in three registers with different volumes and different competition.

Technical terms pull toward English and Hinglish, because “withanolides” is not usually written in Devanagari.

Traditional vocabulary pulls toward Devanagari, “अनुपान”, “रसायन”, “क्वाथ”.

Voice-originated queries tend to arrive in Devanagari.

And the only reliable test is to search the query in each register and look at what ranks.

Step 6: write the brief before you write anything

And this is the step that separates a content plan from a list.

Each brief should name: the exact query, the register, the intent bucket, what the answer actually is in one sentence, which cautions must appear, which internal links it needs, and what would make this piece better than what currently ranks.

That last field is the one worth labouring over. If the honest answer is “nothing”, the brief should be deleted rather than written.

And a brief with no specific answer in it produces the vague wellness article nobody reads, which is how most supplement content gets made.

What this method produces, and the honest limitation

It produces a long list of modest-volume, high-intent, fully compliant topics, and a short list of nothing.

No single piece will bring the traffic that one prohibited high-volume page would have brought. That is the trade, and it should be stated plainly rather than sold as a hack.

What accumulates instead is coverage of a category’s practical questions, which is defensible, does not get pulled, and brings readers who are holding a product or about to buy one.

And it compounds, because the practical questions do not change much while claims rules and marketplace policies do.

How we did it here

Being concrete, since this site is the example.

We built a cluster map first, categories and the questions within each, then expanded it into a set of briefs with the query, the language register and the intent written into each one before any article existed.

The split across English, Devanagari Hindi and Hinglish was decided per cluster from the query register rather than by translating anything.

And a significant number of briefs were re-scoped during writing because two of them turned out to be the same question, which is normal and is better caught at the writing stage than left as two competing pages.

The part worth copying is not the number of articles. It is that the brief existed before the article, and contained the answer.

The short version

1. Standard keyword research outputs terms a herbal brand cannot write about
2. The highest-volume queries are the claim-bound ones
3. Brands either write them and get pulled, or abandon search
4. The third option is a different method
5. Split each query into subject, condition, practical and commercial parts
6. Only the condition component is claim-bound
7. Harvest the subject, build on practical and commercial, discard the condition
8. And practical queries convert better than condition queries
9. “How long should I take it” is past the buying decision
10. “Best herb for X” is at the start of a journey going elsewhere
11. Keyword tools under-report Indian language and long practical queries
12. Mine autocomplete in all three registers, tedious and most productive
13. “People also ask” is a list of answerable questions
14. Your own support messages are the best source and nobody mines them
15. 200 enquiry messages is a content plan
16. Marketplace Q&A sections carry form-specific questions
17. Marketplace complaints are pre-purchase queries for the next buyer
18. Label-term confusion has modest volume and very high intent
19. Because the person is holding the product while searching
20. Classify into fully answerable, partially answerable, and not answerable
21. Partially answerable pieces need review every time, and a see-a-doctor section
22. Do not target prohibited queries with evasive pages
23. An evasive page is incoherent and the intent is visible
24. Map to intent, not volume
25. Practical and verification buckets are where small brands can win
26. Large brands write informational and transactional and ignore the middle
27. Check the register per query, not per site
28. Technical terms pull to English and Hinglish; traditional to Devanagari
29. Voice queries tend to arrive in Devanagari
30. Search each register and look at what ranks, do not guess
31. Write the brief before the article, with the answer in it
32. Include the cautions, the links, and why this beats what ranks now
33. If the honest answer to that last field is “nothing”, delete the brief
34. The output is many modest-volume compliant topics, not one big win
35. That is the trade, and it should be stated rather than sold as a hack
36. It compounds, because practical questions outlast claims policies

Related guides: how to start a supplement brand in India, supplement label compliance in India. Related reading: why most Indian supplement search demand is not in English, Hindi ya Hinglish, kisme publish karein, writing supplement content that does not breach claims rules, and why publishing 500 articles at once is a bad idea.