पाठशाला Pathshala · वृद्धि Vṛddhi, Growth · Lesson 12 · Build

SEO for Indian startups: the long game done right

Search is slow, compounding and crowded with aggregators. Win it by choosing queries by intent, writing in the language buyers search in, and building pages an aggregator cannot copy.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

An old wooden cabinet holds rows of narrow drawers with worn handles in a workshop. — Photograph: Erik Mclean · Pexels

Search “physiotherapist for back pain in Indiranagar” and page one belongs to directories, marketplaces and listing sites. The clinic with the best physiotherapist on the street is on page three. A founder who decides to “do SEO” against that page one will spend a year losing. A founder who reads it first will pick a different fight.

Search is the channel with the slowest start and the longest tail. A page that ranks keeps sending buyers without a media bill, for years; a page that does not rank costs as much to write and returns nothing. In India two facts shape the game more than anywhere else: the first page for most commercial queries is held by aggregators with enormous link profiles, and a large and growing share of searchers think and type in Indian languages or in Hinglish, where those aggregators are thinnest. This lesson is a strategy built on both facts, with a tree for deciding what to write next.

Why search is a long game

Google’s SEO starter guide is candid about time: a change can take anything from a few hours to several months to be reflected in search, and it suggests waiting a few weeks before judging the effect of a change. New sites take longer. That makes search a poor channel for learning fast and a strong channel for compounding once you know what buyers want, which is why it belongs after the [first channel](/library/finding-first-channel-that-works) has been found, not instead of it.

It is also a channel where most effort is wasted. Ahrefs’ study of about fourteen billion pages found that 96.55 per cent get no traffic from Google, and named three reasons: the topic has no search demand, the page has too few links pointing at it, or the page does not match what the searcher wanted. Every part of the strategy below is aimed at one of those three. Choose queries with demand. Choose queries where a page without a huge link profile can still win. And give each query the page its intent calls for.

Read the intent before writing

Every query carries an intent, and the fastest way to learn it is to look at what ranks now, because Google has already made its guess. Four kinds matter to a startup. How-to queries, from someone trying to do something: “how to file GSTR-1 for nil return”. They want steps, a worked example, sometimes a calculator. Comparison queries, from someone choosing: “Tally vs Zoho Books for a small trader”. They want a table and a verdict. Transactional queries, from someone ready to act: “buy organic jaggery online”. They want a product page and a price. Local queries, from someone who needs a person or a place: “CA near me in Andheri”. They want a map, a listing and reviews.

A page that answers the wrong intent cannot rank however well it is written. A product page will not rank for a how-to query; an essay will not rank for a transactional one. And the intent decides who you are competing with. How-to queries are often held by publishers and forums with thin or dated answers, which is where a startup with first-hand knowledge can win quickly. Transactional and local queries are where aggregators are strongest, because they have breadth, reviews and years of links. That is where a startup has to change the question.

The vernacular opening

The KPMG and Google report Indian Languages: Defining India’s Internet, published in 2017, counted 234 million Indian-language internet users against 175 million English users in 2016 and projected the Indian-language figure to reach 536 million by 2021. It also found that 68 per cent of internet users considered local-language digital content more reliable than English content. The projections are dated and should be read as direction, not as current fact. The direction is the point: a large share of Indian buyers search in Hindi, Tamil, Marathi, Bengali, Telugu and the rest, or in Hinglish written in Roman script, and the English-first aggregators serve those queries poorly.

A man leans at the window of a train in Jaipur and talks on his mobile phone. — A great many Indian buyers search the way they speak, in Hindi, Tamil or Hinglish. The aggregators are thinnest there. Photograph: Mahmut Yılmaz · Pexels

Doing it properly means a real page, not a translation widget. A fluent writer in the buyer’s language, using the words buyers use, which for many business topics means English terms inside Hindi sentences. A separate URL for each language version. Google’s guidance on localised versions explains how to tell it that two pages are versions of each other with hreflang annotations, language codes such as hi or hi-IN, each version linking to all the others; it also notes that Google does not use hreflang or the HTML lang attribute to detect the language of a page. The language is detected from the content, so the content has to actually be in the language.

How to outrank an aggregator

Aggregators win on breadth: a page for every city, every category, every combination, and the links that come with being large. A startup cannot match that and should not try. It wins on depth for a narrow query, by offering what a template page cannot. First-hand data: what three hundred of your customers actually paid, waited or filed. A working tool: the calculator, the checker, the generator that does the job the searcher came to do. A specialist’s answer: the physiotherapist’s explanation of which back pain needs a scan, written by the physiotherapist. Local truth: the clinic’s actual hours, the parking, the route from the metro, the reviews it has earned.

The tempting shortcut is the aggregator’s own method at small scale: hundreds of near-identical pages for every city and every keyword. Google’s guidance on helpful, people-first content names that pattern among its warning signs, along with publishing on many topics in the hope some will rank, summarising what others have said without adding value, and writing to a word count. A startup that copies the aggregator’s method has the aggregator’s weakness without its links. Write fewer pages, each one plainly the best answer to its query.

Depth still needs links, because a page nobody points to is one of Ahrefs’ three reasons for zero traffic. A startup earns them the way it earns customers: by being useful to people who already have an audience. The trade association that will link a calculator its members use every quarter. The CA or consultant who will send clients to a checklist that saves them an afternoon. The newsletter that will cite a dataset nobody else has published. The customer who will link a case page that makes them look good. The local chamber, the college incubator, the industry body’s resources page. Each of these is a conversation, not a campaign, and one link from a body the buyer trusts is worth more than a hundred from sites nobody reads. Bought links and link exchanges are the opposite: they risk the whole site for a page that would have ranked on its merits. Keep a short list of twenty organisations whose readers are your buyers and give each a reason to link before asking for one.

An aggregator has a page for everything. A startup wins with the one page on its subject that nobody else could have written.

The technical minimum, and what to ignore

The technical part of SEO for an early startup is short. One page per question. A unique, descriptive title on each page in the buyer’s phrasing. Descriptive URLs. Internal links from related pages using text that says what the target page is about. Images with descriptive alt text. A site that loads quickly on a mid-range Android phone on a mobile network, because that is how most Indian buyers will arrive. Google Search Console set up on day one, so that impressions, queries and positions are recorded from the start. The starter guide is equally clear about what not to spend time on: Google does not use the meta keywords tag, content length alone does not affect ranking, keywords in the domain name do little, and heading order does not matter to search. Spend the saved hours on the page’s substance.

The monthly search review

On the first Monday of each month, open Search Console and the lead sheet and write five things on one line with the date. Total clicks from search and the change from last month. The ten queries with the most impressions, and the position for each. Any query where you sit between positions eight and twenty, which is the list of pages to improve this month. Leads from search pages, by page. And one decision: which page to improve, which new query to write for after walking it through the tree above, and which language version to add. Judge a new page after a full quarter, not a week. Over two years that line is the record of whether search is compounding for the company, and the [content library](/library/content-marketing-that-compounds) that feeds it should be planned from the same list.

Search guidance changes and the language figures are from 2017; both were checked on 10 October 2026. The sources are below. Run your top ten customer questions through the tree this week.