The same customer types “sasta inverter for home” into Google and then reads your entirely English product page without complaining about it.
That is not a contradiction and it is the thing most advice on this subject misses. The language somebody searches in and the language they are willing to read are two different behaviours in the same person, and they do not have to match.
Which means the answer is almost never “translate the website”. It is narrower and cheaper: find the queries people actually type, and decide separately whether the page they land on needs to change language at all.
For most businesses the honest answer is that two or three pages need a vernacular version, the rest do not, and the biggest win is in the words you target rather than the words you publish.
The two behaviours, separated
Think about what each one is for.
Searching is a private act of describing a problem. People do it in whatever language the problem lives in their head, which for most of India is not formal English. So the query arrives as Hinglish, or in Devanagari, or as a spoken sentence, and it frequently contains the word a customer would use rather than the word your industry uses.
Reading is an act of evaluation, and English carries a signal in Indian commerce that has nothing to do with comprehension. An English page can read as more established. Some buyers who search in Hindi would be mildly put off by a business that only speaks Hindi to them, because of what it implies about scale.
So the two decisions come apart. Target the vernacular query. Then ask, separately, whether this specific page converts better in the reader’s first language, which depends entirely on who the page is for.
On finding those queries, I covered the method in keyword research for Indian businesses, including why autocomplete shows you the real phrasing if you start typing the way customers speak. This post is about what to do once you have them.
The gap worth knowing about
Two numbers, from unrelated sources, and I will be honest about their provenance.
IAMAI and Kantar survey data, which I have only seen quoted secondhand, puts the language people use to access the internet in India at roughly 43% English, 24% Hindi, 6% Tamil, 3% Marathi.
Separately, the W3Techs survey of website content languages places Hindi among languages used by under 0.1% of websites.
Set those against each other. Something like a quarter of Indian internet users are working in Hindi, and Hindi appears on well under one in a thousand websites. Whatever the exact figures, the direction of that gap is not subtle.
That is the whole opportunity: a page in Devanagari competing against almost nothing can rank faster than an English page competing against everybody.
I did see a claim that more than half of Indian searches are now in Hindi and regional languages. It cited unnamed aggregated research and I could not trace it to anything, so I would not repeat it, and you should be wary when somebody quotes it at you in a pitch.
Which businesses actually have this demand
Be honest about your own category before spending anything, because the demand is not evenly spread.
Likely yes. Anything bought by a household rather than an office: inverters, water purifiers, two-wheeler service, tuition, tailoring, agricultural inputs, local healthcare, construction materials. Anything sold outside the top eight cities. Anything where the buyer is over fifty. Anything where the purchase is explained to a spouse before it is made.
Likely no. B2B software. Anything sold to a procurement team. Professional services bought by companies. Export businesses. If your buyer reads contracts in English all day, they will search in English too.
Test rather than assume. Open Search Console, look at the queries report, and sort for anything that is not standard English. If transliterated terms are already arriving, you have demand with no page built for it, which is the easiest work available to you. If there is nothing after a year of traffic, believe that.
Why translating the whole site is the wrong first move
Two reasons, and the first is about Google and is worth reading carefully.
Google’s spam policies include a category called scaled content abuse, and the examples name automated transformations explicitly: generating many pages from scraped or existing content “including through automated transformations like synonymizing, translating, or other obfuscation techniques”, where little value is provided to users.
Read the scope precisely, because this gets misquoted in both directions. Google is not saying translation is prohibited. The target is generating large volumes of unoriginal low-value pages, by any method. A careful Hindi version of forty pages you wrote yourself, reviewed by a human who speaks Hindi, is not what that policy is aimed at. Running your 300-page site through machine translation into six languages to multiply your page count is squarely in the neighbourhood.
The second reason is simpler and costs you more often. A translation agency will give you correct, formal, written Hindi. Nobody searches in formal written Hindi. The register people type in is closer to how they speak, with English words left in place where English words are what they actually use. A translator who “corrects” those to their pure Hindi equivalents has just removed the terms you were trying to rank for.
What to write instead
Pick the two or three pages where the decision gets made. Usually that is the page for your main service, the pricing or cost page, and whichever explainer answers the question people ask before buying.
Then write them rather than translating them. Give somebody who speaks the language the English page and the query list, and ask them to answer the same question for the same customer. You will get different emphasis and different objections handled, which is the point.
Three specific traps.
Do not translate your brand or your product names. They are the terms people type. Leave them in Roman script even inside a Devanagari page.
Leave the English technical words alone where that is what people say. If customers say “warranty” and “service centre” in the middle of a Hindi sentence, your page should too. Purity is not the goal, recognition is.
Tag the language properly so the right version is served to the right person. The mechanics are the same ones I went through for a two-language site, including the bit about automatic redirects being the most expensive mistake in the whole area.
And keep both versions, with the English one as the default. You are adding a route in, not replacing the front door.
Before lunch
Thirty minutes, and it will tell you whether any of this applies to you.
Open Search Console, go to the queries report, set the window to the last twelve months and read through for anything transliterated or in Devanagari. Write down what you find.
Then type your main service plus your city into Google, in the phrasing a customer would speak rather than the one your brochure uses, and look at what comes back. If the results are thin or irrelevant, that is an unserved query with a buyer attached.
Then pick one page. Not the site, one page, the one where people decide. That is this quarter’s experiment and it will tell you more than any survey about whether the rest is worth doing.
If you’d like to work out whether your category genuinely has vernacular demand before you spend on translation, drop me a line on email, WhatsApp, phone or LinkedIn and we can have a quick chat, and my notes on SEO cover the rest. I’m taking on a small number of contracts at the moment, so I’ll tell you straight whether it’s something I could help with.