Keyword research for an Indian local business: a method for Hinglish, Hindi and near me searches

Most keyword research advice was written for a SaaS company in Austin. It tells you to open a tool, type a seed keyword, sort by volume and pick the winners. Do that for a dental clinic in Pune and the tool returns “dentist near me” with a big number attached, and nothing at all for the way the clinic’s actual patients search, which is closer to “daant ka doctor kothrud” typed with one thumb on a Redmi, or spoken into the microphone in Hindi.

This guide is the method we use for that kind of business. It fits into one working week, it starts with data Google already gives you at no cost, and it only reaches for a paid tool at the two points where one earns its place. It is written for the owner of a small business (a clinic, a coaching centre, a CA firm, a tiles showroom) and just as much for the freelancer who has been handed that client and a small monthly budget.

Who this is written for

The business has a Google Business Profile, a website of somewhere between one and twenty pages, and a WhatsApp number that does more work than either. Enquiries arrive as “hello sir rate kya hai” at eleven at night. The owner knows the competitors by name and by which one just put up a bigger hoarding. Nobody has ever looked at Search Console.

If that is you, or your client, the keyword universe is small. A CA firm in Jaipur competes on a few dozen queries, in three scripts, across five or six localities. A week is enough to map it properly.

How people in India actually search for a local business

Before any tool, it helps to name the shapes a query takes here, because each one needs a different kind of page to answer it.

Plain English with “near me”

“Dentist near me”, “CA near me”, “coaching classes near me”. These are the queries every tool can see, and the map pack dominates them. Your website barely matters for them; your Business Profile does.

Hinglish in Roman script

“Daant ka doctor”, “ITR file karne wala CA”, “coaching centre ka fees”, “sunday ko open clinic”. This is how most people under forty type on a phone. Tools show little or no volume for these because their databases were built from a population that types in English. People here search them all the same.

Devanagari

“दांत का डॉक्टर पुणे”, “जयपुर में सीए”, “लखनऊ कोचिंग सेंटर”. The Hindi keyboard and voice input on Android produce these, and they are common in tier-2 and tier-3 cities where many people came online on a phone first. Tools are close to blind here.

Voice

Longer, conversational, often a full question: “mere paas sabse accha dentist kaun hai”, “Kothrud mein dentist kitne baje tak khula hai”. The microphone button on Android produces a lot of these, and they behave like questions rather than keywords.

City and area modifiers

Indian cities are big enough that “Pune” is not a useful modifier on its own. People search the locality: Kothrud, Hadapsar, Viman Nagar. In Lucknow it is Gomti Nagar, Aliganj, Hazratganj. Your query map needs the localities you actually serve, not just the city.

One behaviour shapes everything above: the buyer’s next step is WhatsApp. A page that ranks and then asks someone to fill in five fields loses to a page that ranks and shows a green button. Keep that in mind when you decide what a “converting” keyword looks like.

Day one: write down the words customers already use

Start with the language you already have, before opening a single tool. This takes thirty minutes and it produces the seed list every later step depends on.

• Open the business WhatsApp and read the last fifty incoming first messages. Write down the exact phrasing: which service they name, in which script, what question they ask first (“fees kitni hai”, “appointment chahiye”, “GST registration ho jayega?”).
• Ask the person at the reception desk, or the owner, how callers describe what they want. A coaching centre in Lucknow may find that nobody says “coaching centre”; they say “NEET ki taiyari” or “class 12 tuition”.
• Check the Business Profile messages and the Justdial or Sulekha enquiries if those are running.
• List the localities that show up in addresses and in “aap kahan ho” questions.

You should end the session with thirty to sixty phrases. Group them by service, and mark the script of each one. This is your seed list. It is more accurate than anything a tool will hand you, because it is what real buyers said.

Day two: pull Google’s own data

Google gives a local business three data sources for nothing, and most businesses have never opened any of them.

Search Console

If the site is not verified in Search Console, do that first (a DNS record or an HTML tag; it takes ten minutes). Then in the Performance report, set the country filter to India and look at the last three months of queries.

Two filters matter here. First, filter queries by a custom regex of \p{Devanagari} to see every Hindi-script query the site already appears for. On most local sites this list is short and surprising. Second, filter queries containing the locality names from your seed list. The queries that get impressions with a low average position are the ones a page could win with modest work.

If the site is new and Search Console is empty, skip this and come back in a month. Do not let an empty report stop the rest of the week.

Business Profile performance

In the Business Profile manager, the Performance section shows the actual search terms that surfaced the profile. This is the only place you will ever see “near me” data for your own business, and it often shows Hinglish and Hindi terms that no paid tool reports. Copy the whole list into your sheet, even the odd ones.

Autocomplete and People also ask

Open a private window. Set the search URL to India and Hindi with google.com/search?q=&gl=in&hl=hi and start typing each seed phrase, in each script. Write down every suggestion. Then repeat with hl=en. The two sets differ, and both are real.

For the questions your customers ask, search the head term (“dentist Pune”, “CA Jaipur”) and expand the People also ask box four or five times. Those questions are the outline of your FAQ section later in the week.

Trends, with a city

Google Trends lets you compare two phrasings inside India and narrow by state or city. Compare “dentist near me” with “daant ka doctor” for Maharashtra, or “CA” with “chartered accountant” for Rajasthan. You are not after absolute numbers; you want to know which phrasing is rising and which is flat, so you know which script to lead with on a page.

Day three: build the query grid

Put the whole thing in one sheet with five columns: service, phrasing, script, modifier, and what page or profile element should answer it. A few rows from a Pune dental clinic look like this.

Do the same for a CA firm (“ITR filing charges jaipur”, “GST registration kitne din mein”, “सीए जयपुर”) or a coaching centre (“NEET coaching lucknow fees”, “class 11 physics tuition gomti nagar”, “कोचिंग सेंटर लखनऊ”). Expect sixty to a hundred and fifty rows. That is the whole market for one business, and it fits on one screen.

Two rules while filling it. Keep the Hinglish spelling variants together as one row (“kharcha”, “kharch”, “cost” mixed in) because Google treats them as the same intent. And mark any row where the honest answer is a price. Indian buyers ask for the rate first, and a page that gives a range wins the click over a page that says “contact us”.

Day four: use the paid tools where they help

Everything so far cost nothing, and for many local businesses it is enough. The paid tools help at two points: sizing the English half of the grid so you know which pages to build first, and seeing what competitors rank for that you missed.

We sell shared access to these tools through one dashboard. That means a browser login, no API keys, a fair-use cap that is pooled with other customers, and the occasional slow afternoon when many people are running reports at once. For the one-week job in this guide that is fine; for an agency running thirty audits a day it is not, and we say so on our safety page.

Semrush for the English rows and the competitor list

Set the database to India in the Keyword Magic Tool and run the English service terms. You get volume and difficulty for “root canal cost pune” and its siblings, and, more usefully, the Keyword Gap report against the two or three competitor domains you noted on day one. That report shows the English queries a competitor ranks for that you have no page for. The Position Tracking tool can also track the map pack for a specific city, which is the one thing Search Console cannot show you.

Semrush through us is ₹199/month on its own. Semrush’s retail plans are built and priced for agencies; the current list is on semrush.com/pricing, and the page localises its currency, so check it from an Indian connection. Tool page: Semrush.

Ubersuggest for a second opinion and a quick audit

Ubersuggest is lighter and simpler, and for a local business that is a feature. Set the country to India, paste in the English rows, and use the content ideas view to see which competitor pages already rank for each. Then run its site audit once against the client’s site; a five-page clinic site usually has three or four fixable problems (missing titles, slow images, a page not linked from anywhere) that matter more than any keyword.

It is ₹199/month through us. Its own pricing page is at neilpatel.com/ubersuggest/pricing; we could not load it from our side today, so we are not quoting a figure. Tool page: Ubersuggest.

Mangools when you want a city-level SERP

KWFinder lets you set the location down to a city rather than a country, and its SERPChecker shows what actually ranks in Pune for a term rather than in India overall. For local work that matters: the top ten for “dentist Kothrud” as seen from Pune is not the top ten a tool in Delhi computes. Its difficulty score is tuned for smaller sites, which is what you have.

It is ₹199/month as a single. It is not in the Essential plan; the cheapest plan that carries it is Advanced at ₹1,499/month for 25 tools, which a local business does not need. Buy the single. Its retail pricing is listed on mangools.com/plans-and-pricing; the figures load in the browser rather than in the page source, so we are linking rather than quoting. We have a longer write-up at this post.

AnswerThePublic for the question rows

For the voice and question queries, AnswerThePublic takes a seed term and returns the questions and prepositions people attach to it, filtered by country. Set it to India. Its Hindi coverage is thin, so expect it to be strongest on the English rows and use People also ask for the rest. It is a one-afternoon tool for this job.

The Starter plan is US$20/month on answerthepublic.com/pricing, checked 21 September 2026. Through us it is ₹299/month as a single, and it is only bundled in the Ultimate plan, which is too much plan for this job. Buy the single for a month, export what you need, and stop. Tool page: AnswerThePublic.

What the tools will not tell you

Every one of these tools will show zero or “no data” for most of the Hinglish and Devanagari rows. Do not delete those rows. Your day-two data from the Business Profile and Search Console is the evidence they exist. The tools measure a population that types in English; your customers are a different population, and the grid you built from their own messages is the better source for that half.

Day five: map queries to pages and to the profile

Now decide what answers each row. Most local sites need fewer pages than owners expect.

The Business Profile answers the near me rows

Set the primary category correctly (a “Dental clinic” and a “Dentist” are ranked differently), fill every service with its own description, keep the hours accurate including the Sunday question, and post once a week. Add the top ten questions from your grid to the profile’s Q&A section yourself, with answers. The profile’s description and posts can carry Hinglish naturally; write them the way customers talk.

One page per service, with the locality in the copy

A root canal page that mentions Kothrud, gives a price range, and answers “RCT mein kitna time lagta hai” in a short FAQ will outrank a generic “our services” page. If you serve three localities, mention all three on the page rather than making three thin copies of it.

A Hindi page where Devanagari rows cluster

If your grid has fifteen or more Devanagari rows around one service, write that page in Hindi, with a proper hreflang pair to the English one. We cover the mechanics in SEO for Hindi websites. Do not machine-translate the English page and call it done; a Hindi reader can tell, and so can Google.

FAQ blocks for the question rows

Every service page gets four to six questions from the grid, answered in one or two sentences each, in the language the question was asked. Mixed-script FAQs (“Root canal ka kharcha kitna hai?”) are fine and read naturally to the audience.

Writing for voice and WhatsApp

Voice queries are questions, so headings phrased as questions match them. “Kothrud mein dentist kitne baje tak khula hai?” as an h3 with a one-line answer beneath it does more for voice than any schema trick.

Every page also needs a click-to-chat link near the top, with the number visible as text as well, because many people still copy the number into WhatsApp themselves. Track it: a wa.me link with a UTM tag, or a separate number for the website, tells you in a month which pages are producing conversations. Conversations are the conversion metric for a local business; form fills are rare and tell you little.

Track it monthly, not daily

Rankings for a local business move with the map pack and with what Google thinks the searcher’s location is, so a daily rank check will only make you anxious. Once a month, do three things: export the Search Console queries again and diff them against the grid, copy the Business Profile search terms into the sheet, and count WhatsApp conversations per page. New phrases that appear in the profile data become new rows. Rows that never got an impression after three months get demoted.

If you kept Semrush, its Position Tracking with the city set to Pune and the map pack switched on is worth a monthly look. If you did not, Search Console and the profile are enough.

What this costs and when not to buy from us

The Essential plan carries both tools this method leans on. It is ₹399/month for 5 tools, and the two you would use here, Semrush and Ubersuggest, come to almost exactly the same figure bought as singles (₹199 and ₹199 a month). The plan’s other three tools are writing aids, useful if you or the client will be drafting a dozen service pages in English, otherwise not. So the honest guidance is: pick the plan if you want the writing tools thrown in for the same money, and pick the two singles if you do not. It makes no difference to the bill.

Mangools and AnswerThePublic are not in Essential. Buy them as singles for the month you need them; upgrading to Advanced or Ultimate to get them makes no sense for one local business. Our full price list is on the pricing section, and for a wider view of what Indian freelancers actually run, see our India guide.

Payment from India is UPI or card, and access is through the dashboard at my.bundledseo.com within a few minutes of paying. Because it is shared access, the limits are shared too, and the vendor can reset the account from time to time. Plan the tool work for one or two sittings rather than leaving reports open all day.

And when should you not buy anything? If the site is a one-page brochure and the business has fewer than five services, days one to three and five are the whole job. The Business Profile and Search Console will carry you for the first year. Spend the money on a photographer for the profile instead.

Frequently asked questions

Do I need a Hindi website for a local business in India?

Not always. Look at the grid: if Devanagari rows cluster around one or two services with real impressions in Search Console or the Business Profile, a Hindi page for those services is worth building. If Devanagari rows are scattered and few, put Hindi and Hinglish in the FAQ blocks and the profile posts and leave the site in English for now.

Should I write pages in Hinglish or in proper Hindi?

Match the query. Hinglish typed in Roman script is answered well by English pages with Hinglish FAQs and headings. Devanagari queries are better answered by a real Hindi page. Mixing both on one page is normal for this audience and Google handles it, as long as the page is not a machine translation.

Why does the keyword tool show zero volume for my Hinglish keywords?

The tool’s database was built from queries typed mostly in English, so transliterated Hindi shows as no data even when your Business Profile reports it every week. Treat the profile’s search terms and Search Console as the source for those rows, and use the paid tool for the English rows and competitor research.

How many keywords should a small clinic or CA firm target?

Sixty to a hundred and fifty rows in the grid is typical, mapping to five to fifteen pages plus the Business Profile. Anything larger for a single-location business usually means duplicate variants that should be merged into one row.

Is the Essential plan enough, or do I need a bigger plan?

For this method, Essential carries the two tools that matter, Semrush and Ubersuggest, and its price is close to the two singles bought separately. Mangools and AnswerThePublic are only in larger plans, so buy those as singles for a month rather than upgrading. See the pricing section for the current figures in rupees.

How long before I see results in Google Business Profile and search?

Profile changes (categories, services, hours, Q&A) usually show in the profile’s search-term data within two to four weeks. New service pages take one to three months to settle, longer for a new domain. Track monthly and judge by WhatsApp conversations, not by a rank number that moves with the searcher’s location.

Does shared tool access affect the keyword data I get?

No, the data is the same as a full subscription; what is shared is the daily report limit and the login. Plan your Semrush and Ubersuggest sessions for one or two sittings, export to your sheet, and you will rarely notice the cap. Heavy afternoon use by other customers can slow reports, so early mornings tend to be quicker.