There are no items in your cart
Add More
Add More
| Item Details | Price | ||
|---|---|---|---|
By Ravi Verma, Founder - Pre Tutorials, Hazratganj, Lucknow · Published 25 October 2026
Quick answer: We pulled 10,781 unique YouTube autocomplete phrases across 30 digital marketing topics in October 2026. Not one of them was written in Devanagari script. Indian learners type Hindi in Roman letters: 'kaise' appeared 965 times against 380 for 'how to', a 2.5:1 lead. But the 214 video titles that actually rank for those searches invert it — 96 use 'how to' against 43 using 'kaise', and 30 of them use Devanagari that zero searchers typed. The titles that win do not pick a side: 68 per cent carry both languages separated by a pipe.
We run a training institute in Lucknow and a YouTube channel with roughly 29,000 subscribers. In October 2026 we rewrote 130 video titles and needed to answer a question we had been guessing at for years: when an Indian learner searches YouTube for a marketing skill, what language do they actually type?
Everyone has an opinion. Half the creators we know insist on pure Hindi. The other half write English titles and add 'in Hindi' at the end. We could not find a single public dataset that settled it, so we built one.
This post is that dataset and what it showed. The raw numbers are below so you can argue with them.
We are publishing it because the advice circulating in Indian creator circles is contradictory and almost none of it cites anything. We have sat in enough counselling calls with students who were told, confidently and for free, three incompatible things about titles in the same week. A dataset settles less than people hope, but it settles more than opinion does.
We collected YouTube's own autocomplete suggestions — the phrases that drop down as you type — for 30 seed topics, expanded A to Z. Autocomplete is drawn from real search behaviour, which makes it a reasonable proxy for demand. We then scraped the top-ranking video titles for 18 Hinglish queries to see what the supply side looks like.
Seeds covered the topics we teach: facebook ads, meta ads, google ads, seo, wordpress, canva, chatgpt, claude ai, affiliate marketing, whatsapp marketing, youtube, thumbnail, freelancing, automation, lead generation and others. The method is YouTube's own suggestion API, which anyone can reproduce — see YouTube's search documentation.
Zero of 10,781 phrases contained a single Devanagari character. Not one. Across 30 topics and every letter of the alphabet, the Hindi-speaking audience typed Hindi words in Roman script, every time.
This surprised us more than anything else in the dataset. Uttar Pradesh alone has over 199 million people, the overwhelming majority Hindi-speaking (Census of India). The keyboard, not the language, decides the script.
The practical consequence is blunt: if your video title is written in Devanagari, it does not match the text your audience is typing. YouTube is good at bridging that gap, but you are making it work harder for no reason.
The same applies to descriptions and tags, where the cost of covering both scripts is zero. A description can carry the Devanagari phrasing for readers who prefer reading it and the Roman-script phrasing for the search index, and nobody has to choose. Titles are where the constraint bites, because you have one line and roughly sixty visible characters to spend.
One caution before anyone strips Devanagari from an existing channel: our data says nothing about how Devanagari affects click-through once a video is already shown. A viewer scanning results may well respond better to a familiar script than to Roman transliteration. We measured what people type, not what they click.
On the demand side, Hinglish wins clearly.
| Term in autocomplete | Phrases | Share of 10,781 |
|---|---|---|
| kaise | 965 | 9.0 per cent |
| how to | 380 | 3.5 per cent |
| hindi | 411 | 3.8 per cent |
| free | 392 | 3.6 per cent |
| devanagari script | 0 | 0.0 per cent |
Note that 411 phrases contain the word 'hindi' itself. People actively ask for Hindi content — in English letters.
Almost. We checked every common Hindi question word and the distribution is extraordinarily lopsided:
One question word accounts for 88 per cent of all Hinglish question searches in this dataset. If you optimise for a single Hindi word, optimise for 'kaise'. The second-place finisher, 'kya', is a definitional query — the sort answered by a post like AI tools kya hai rather than a tutorial.
The verb that follows 'kaise' is nearly as concentrated. 'kaise kare' appeared 332 times, 'kaise banaye' 175 times, and 'kaise banaen' a further 47. Those three constructions cover more than half of all 'kaise' searches.
Here is the part that changed how we write titles. We took 18 Hinglish queries and recorded the 214 video titles ranking for them. The language split inverts completely.
| Signal | In what people search | In what actually ranks |
|---|---|---|
| 'how to' | 380 (3.5 per cent) | 96 titles (44.9 per cent) |
| 'kaise' | 965 (9.0 per cent) | 43 titles (20.1 per cent) |
| Devanagari script | 0 (0.0 per cent) | 30 titles (14.0 per cent) |
Read that last row again. Fourteen per cent of ranking titles use a script that zero searchers typed. Those videos rank anyway — on watch time, channel authority and topical relevance, not on literal text matching.
That is the honest caveat to this whole study: title wording is one ranking input among many, and clearly not the strongest one. A weak title on a strong channel still wins.
So why does English win on the supply side? We think there are two reasons, and we can only evidence the first.
The first is measurable: 68 per cent of ranking titles (145 of 214) contain a pipe character, and 24 of them carry 'how to' and 'kaise' in the same title. The winners are not choosing a language. They are stacking both phrasings into one title and collecting both audiences.
Real examples from the ranking set:
The second reason is a guess: English-first titles may simply be a habit inherited from an earlier era of Indian YouTube, and the dataset is catching a lag between how people search now and how creators learned to write titles years ago. We cannot prove that from this data.
This was the most useful finding for us practically. The Hinglish-to-English ratio is not constant — it swings hard by subject.
| Topic | 'kaise' phrases | 'how to' phrases | Leans |
|---|---|---|---|
| youtube | 96 | 15 | Hinglish |
| thumbnail | 90 | 14 | Hinglish |
| chatgpt | 70 | 15 | Hinglish |
| canva | 64 | 15 | Hinglish |
| make money online | 7 | 36 | English |
| google tag manager | 6 | 19 | English |
| sales funnel | 1 | 8 | English |
The pattern we read from this, and we are stating it as an interpretation rather than a fact: hands-on creative and platform tasks get searched in Hinglish; technical and business-jargon topics get searched in English.
Someone learning to make a thumbnail asks in the language they think in. Someone searching 'google tag manager' already learned that term in English and has no Hindi word for it. 'Sales funnel' has no comfortable Hinglish form either.
Eight of our 30 topics leaned English: make money online, google tag manager, digital business card, google merchant center, conversion tracking, sales funnel, claude ai and pinterest marketing. The remaining 22 leaned Hinglish.
There is a rough test you can apply before you write a title. Ask whether the topic has a comfortable Hindi form that people say out loud:
Beyond language, the 214 ranking titles share measurable habits:
The gap on 'free' is worth sitting with. It is the fourth most common word in what people search and nearly absent from what ranks. The charitable reading is that creators avoid it because free-seeking viewers watch badly and convert worse, so the titles that survive are the ones that never attracted them. The less charitable reading is that it is simply an unexploited phrasing. We have not tested either, and we did not add 'free' to our own titles on the strength of a correlation we cannot explain.
The median ranking video had roughly 86,000 views, and the biggest had over 20 million. This is a competitive set, not a long tail. Anyone reading these title patterns as a shortcut should hold that number in mind: the titles are a thin layer on top of channels that have already earned their position, and copying the layer does not transfer the position.
We rewrote 130 of our own video titles in October 2026 using one format:
In practice that produces titles shaped like this, taken from our own rewrite:
Two things changed there. The earnings figure came out, because we do not make income claims and because Google and YouTube both treat them as a risk. And the phrasing moved to the construction the data says people actually type for a hands-on Canva task.
The same logic applies to written content, with one adjustment. A blog post has a title, a URL slug and a set of headings, and they do not all need the same language. We write the visible heading in the language the topic leans toward, keep the slug in Roman-script Hinglish because that is what people type, and use the subheadings to carry the phrasings the title could not fit. A post answering 'instagram reels viral kaise kare' can carry an English subheading about the algorithm without confusing anybody — readers code-switch far more comfortably than title-writers assume.
We are deliberately not publishing a before-and-after traffic claim. The rewrite is days old at the time of writing, and attributing a view count to a title change that soon would be dishonest. We left 46 titles untouched because they were running YouTube's native A/B title tests; those form a natural control group and we will report the comparison when there is enough data to mean something.
We would rather state these than have them pointed out:
None of this needs paid tools. The method, in full:
Cross-check anything surprising against Google Trends and your own YouTube Analytics research tab, which shows what your specific audience searches for. For the Google side of the same question, Google's title link documentation is worth reading.
If you want the practical application rather than the method, we have written it up for YouTube Shorts SEO and for SEO basics separately.
Write them in both, separated by a pipe. In our October 2026 dataset of 214 ranking titles, 68 per cent used a pipe and 24 carried 'how to' and 'kaise' in the same title. The audience searches in Roman-script Hinglish 2.5 times more often than English, but the titles that actually rank skew English — so covering both phrasings is the safer choice for an Indian audience.
Our data suggests there is no search reason to. Across 10,781 YouTube autocomplete phrases collected in October 2026, exactly zero contained a Devanagari character. Indian viewers type Hindi in Roman letters. Fourteen per cent of ranking titles do use Devanagari and rank anyway, so it is not a penalty — it simply does not match what anyone is typing.
'Kaise'. It appeared in 965 of 10,781 phrases, against 132 for 'kya' and one or fewer for kitna, kaun, konsa, kyu and kahan. That is roughly 88 per cent of all Hinglish question searches in our dataset. The most common constructions are 'kaise kare' (332 phrases) and 'kaise banaye' (175).
Title text is only one ranking input. Watch time, click-through rate, channel authority and topical relevance matter more, which is why 14 per cent of ranking titles in our sample use Devanagari that nobody searches for. The strongest explanation we can evidence is that 68 per cent of ranking titles carry both languages, so they are not really English-only titles at all.
We only measured YouTube autocomplete and YouTube rankings, so we cannot claim it transfers. The underlying behaviour — typing Hindi in Roman script because of the keyboard — is not platform-specific, so we would expect a similar pattern on Google. But that is an expectation, not a finding, and we have not tested it.
We would redo it every six months. Autocomplete reflects live behaviour and shifts with new tools and new terminology — 'claude ai' barely existed as a seed topic two years ago and now returns 347 unique phrases. A snapshot older than a year should be treated as historical.
If you want to learn this kind of research properly rather than copying a format, our AI and digital marketing training covers keyword research, YouTube SEO and the analytics behind both, taught live in Hazratganj and online. The Claude AI Mastery Program runs 18 days at Rs 999 (MRP Rs 8,999); the full course list is on the Pre Tutorials store, and the flagship programme is at Claude AI Mastery Program.
Related reading: YouTube se paise kaise kamaye, Instagram Reels viral kaise kare, SEO best practices checklist, digital marketing course in Hindi and best AI tools for digital marketing.
Free career counselling call book kijiye: tidycal.com par slot chuniye
Call ya WhatsApp: 95544 53555 · Mon–Sat, 8 AM – 6 PM
Address: 5th Floor, Pratap Bhawan, behind Leela Cinema, Hazratganj, Lucknow 226001
Google: 4.9 stars from 500+ Google reviews (at the time of writing) · YouTube @pretutorials, 29K subscribers
Tirgiji Tech Solutions Pvt Ltd. Hum placement support dete hain — portfolio, interview preparation aur referrals. Job, salary ya income ki koi guarantee nahi di jaati.
Launch your Graphy