Delhi Local SEO Notes

Twelve months is the normal length of an SEO engagement in India and it is a reasonable one — the work does not produce much in eight weeks and judging it at that point is unfair to everybody. But twelve months is also long enough that a bad fit costs you a year, so the questions you ask before signing matter more than the price.

Here are the ones that actually separate a good engagement from an expensive one, in the order I would ask them.

What are the first three things you will change, and why those three?

A proposal that cannot answer this has not looked at your site. Anybody who has spent an hour on a real site has a list — the canonical tags are pointing at old URLs, the service pages are four hundred words, half the catalogue is not indexed. If the answer is a list of deliverables rather than a list of problems, that is a template, not a plan.

Which pages are we trying to rank, and for what?

The answer should be a short table: page, keywords, current position. Not “we will target 30 keywords”. Thirty keywords with no pages attached is a line item; the same thirty mapped onto six specific URLs is a strategy, and you can hold it to account at month six.

Be suspicious of any plan where the homepage is the target for a city keyword. If you want to rank for your city, you need a page about your city.

What does the monthly report contain?

Ask to see a real one from another client with the name removed. If it is a list of activities — posts written, links built, errors fixed — it tells you what was done but not whether it worked. The report you want has positions for the agreed keyword set, impressions and clicks for the agreed pages, indexed page count, and enquiries attributed where attribution is possible.

An activity report is not a performance report, and twelve months of activity reports is how a year goes by without anybody noticing nothing moved.

Who does the work, and will I speak to them?

In a lot of Indian agency arrangements the person who sells is not the person who executes and neither is the person who writes. That is not automatically bad, but you should know the structure, know who writes the content for your industry, and have a direct line to whoever makes technical changes to your site.

What access do you need, and what will you change without asking?

Reasonable answers: Search Console, Analytics, a CMS login, and a written process for anything that touches templates or redirects. Unreasonable: full server access with no change log, or content published on your domain without review.

Equally, decide what you are willing to let them change quickly. An engagement where every title tag edit needs a week of approvals wastes the retainer on waiting.

What happens to the work if we stop?

Content on your own domain stays yours. Links built to your site stay. But accounts created on third-party platforms, tracking configurations, and the documentation of what was done frequently walk out of the door with the agency. Ask for a handover list up front and it costs nothing; ask for it at the end and it rarely arrives complete.

What are the realistic timelines for my specific situation?

A site with existing authority moving from position 15 to position 6 is a different timeline from a new domain trying to reach page one. Any answer that does not distinguish between those two is a sales answer. Expect the honest version to include a quarter where very little visible happens.

What is not included?

The most expensive surprises in SEO contracts are the exclusions: content beyond a certain word count, development work, design, paid promotion of content, local listings management, the cost of tools. Get the exclusions in writing before you sign, not in month four.

The one red flag that matters most

Guaranteed positions. Nobody controls Google's ranking, and anybody who promises a position is either misunderstanding their own work or counting on you not checking. A guarantee of effort, of a specific scope, and of reporting transparency is reasonable. A guarantee of position one is not available for sale.

If you are a business in the city going through this exercise and want an outside read on a proposal before you commit a year to it, the full scope breakdown is set out here.

A companion piece on reading the numbers once an engagement is running: the monthly SEO report line items that actually predict revenue.

Citation consistency is one of the least glamorous pieces of local SEO and one of the few that still reliably moves local pack rankings. For a single-location business it is simple. For a Mumbai business with branches in Andheri, Thane and Navi Mumbai it becomes the thing that quietly holds the whole local presence back.

The problem in one sentence

Across directories, aggregators, your own website and your Google Business Profiles, the same business is listed with slightly different names, three different phone numbers, two spellings of the address and one branch that closed in 2024. Google is trying to work out how many businesses this is, and it is guessing.

Where the inconsistency comes from

Rarely from carelessness. It accumulates:

• A branch manager created a Business Profile years ago, nobody has the login, and it still ranks
• A directory scraped an old address and nobody noticed
• Marketing standardised on “Pvt Ltd” in the legal name while the signage says something shorter
• A landline was replaced by a mobile, updated in four places out of thirty
• A suite number is written as “Unit 4B”, “4-B”, “#4B” and “Shop No. 4B” depending on who typed it

None of these is a crisis alone. Twenty of them across three branches is enough noise that Google cannot confidently associate reviews, links and search behaviour with the right entity.

The fix, in the order that matters

1. Decide the canonical NAP, in writing. One business name, exactly as it should appear everywhere. One address format per branch. One phone number per branch. Write it in a document, circulate it, and treat it as the only correct version. This sounds trivial and it is where most cleanups fail, because without it every person doing the updating invents their own format.

2. Fix your own website first. Every branch gets its own page, with its own NAP in the HTML as text, not inside an image, and its own LocalBusiness schema with the correct address and phone. Your site is the source Google weighs most heavily. Fixing directories while your own footer still shows the old number is backwards.

3. Claim or reclaim every Google Business Profile. Search your brand name plus each locality. Any profile you find that you do not control is either claimable or needs a duplicate report. Unclaimed profiles with old data are the most damaging single item on this list, because they are the version Google shows.

4. Then work the directories. Justdial, Sulekha, IndiaMART where relevant, Yellow Pages India, the Apple and Bing listings, and whichever industry directories matter in your vertical. Update rather than create. A new listing alongside an old wrong one is worse than the old one alone.

5. Mark the closures. A branch that no longer exists should be marked permanently closed on its Business Profile, not deleted. Deleting loses the signal; closing transfers it.

How long it takes and what to expect

A three-branch cleanup across the main Indian directories is typically two to three weeks of intermittent work, mostly waiting on verification. The effect on local pack visibility is usually visible within six to eight weeks, and it tends to show first as a reduction in the gap between branches — the weakest branch starts behaving like the others.

Keeping it clean

Put one person in charge, review the canonical NAP document whenever anything changes, and search your own brand name plus each locality once a quarter. Fifteen minutes, four times a year, prevents the whole problem from reassembling.

Related, from earlier in this series: review velocity and how it interacts with local rankings, at https://indiaseofieldnotes.mataroa.blog/blog/when-your-google-business-profile-outranks-your-own-website-in-mumbai/

For multi-location businesses that want the audit done rather than described, the branch-level work sits here: https://deepbhardwaj.com/seo-expert-in-mumbai/

Before you commission another month of link building, spend two hours reading the links you already have. On most Indian sites that exercise changes the plan, because the profile is telling you something the keyword list is not.

The useful version of this is not a backlink count. It is an anchor text distribution, read against what you are actually trying to rank for.

What you are looking for

Export your referring domains and their anchor text. Then bucket every anchor into five groups:

Exact match — the keyword, verbatim. “SEO expert in Mumbai”. Partial match — the keyword with words around it or a close variant. “SEO consultant in Mumbai”, “experienced Mumbai SEO services”. Brand — your name or company name. Naked URL — the address itself. Generic — “click here”, “read more”, “this guide”, “learn more”.

Now look at the proportions. A profile that grew naturally tends to be brand-heavy, with exact match as a minority. A profile that was built tends to be exact-match-heavy, because whoever built it was optimising for the keyword rather than for the shape of the profile.

The two failure patterns

Over-optimised. Forty per cent of anchors are the exact keyword, almost all from directories and bookmarking sites, all pointing at one page. This is the pattern that gets a page filtered rather than ranked, and the symptom is distinctive: the page has more links than its competitors and ranks below them, and it has been stuck there for months. More exact-match links will make it worse, not better.

Diluted. Ninety per cent brand and naked URL, almost no partial match, and every link pointing at the homepage. Common on sites that got links organically through PR and never did anything deliberate. Here the fix genuinely is more links — but partial match, pointed at the specific pages you want to rank, not the homepage.

Those two profiles need opposite prescriptions, and you cannot tell which you have by counting.

The page-level distribution matters more than the site-level one

A site-wide anchor summary hides the thing that usually explains a stuck ranking. Break it down per target URL.

The recurring finding on Indian sites: the homepage has four hundred links and the city or service page you are being measured on has eleven. Every directory submission, every profile, every citation went to the root domain, because that is the field the form asked for. Meanwhile the page that needs to rank for “SEO services in Pune” or “industrial valves manufacturer” has almost nothing pointing at it.

That is a fixable problem and it does not require new links at all — it requires internal linking from the pages that do have authority, with anchors that describe the target page rather than saying “click here”.

Check where the links actually land

Fetch every referring URL and record the final destination after redirects. Two things turn up regularly.

Links pointing at URLs you retired. If the redirect is in place they still pass most of their value; if it 404s, that link is doing nothing and has been for however long. On a site that has been through a replatform this is often a double-digit number of wasted links.

Links that are gone. The page that linked to you was deleted, or rewritten, or the whole directory wiped its user submissions while still answering with a 200 status code. A status check will not catch this — you have to look for your own link on the page. Any report that counts a link without having seen it on the live page is counting something that may not exist.

Nofollow is not the dividing line people think

Plenty of nofollow links sit on pages that send real referral traffic and real brand signal. A profile that is ninety-five per cent nofollow from low-quality directories is a problem; a profile with a healthy mix including nofollow from places real people read is not. Record the attribute, do not treat it as the only thing that matters.

What the audit should output

One table: target URL, link count, anchor distribution across the five buckets, and how many of those links are verified live on the page today. That single table usually makes the next quarter's plan obvious — which page is starved, which is over-optimised, which anchors to stop using, and how many of the links you have been paying for actually exist.

The Mumbai service page breakdown for this site, including the anchor split by target URL, sits here: https://deepbhardwaj.com/seo-expert-in-mumbai/

Related reading: reading a Mumbai SERP before writing a page — https://indiaseofieldnotes.mataroa.blog/blog/when-your-google-business-profile-outranks-your-own-website-in-mumbai

Most monthly SEO reports are a list of numbers that went up. They are pleasant to receive and useless for deciding anything. The problem is not dishonesty — it is that the metrics chosen are the ones easiest to report rather than the ones that tell you whether the work is going to produce revenue in three months.

Here is what belongs on the first page, and why.

1. Indexed pages, with the direction of travel

Not a total in isolation. The count, versus last month, with the reason for exclusions on the pages that are not indexed.

This is the metric most reports omit entirely, and it is the leading indicator that precedes everything else. A page not in the index cannot rank at any position. When indexed pages rise, the site's eligible surface area is growing, and rankings follow on a lag. When it falls, something has broken and you will feel it in traffic six weeks later.

2. Queries in positions 5 to 20, counted

Not average position across the site, which is an average of things that have nothing to do with each other and moves for reasons nobody can explain.

Count the queries sitting between position 5 and position 20. That band is where effort converts into clicks — a move from 14 to 8 changes traffic materially, while a move from 62 to 48 changes nothing. If the count in that band is growing month over month, the work is compounding whether or not any single headline keyword has moved yet.

3. Impressions on commercial queries only

Total impressions include every accidental informational match and every brand search. Filter to the queries that indicate someone might buy — the ones containing your service words, your city, price or cost or near me — and report impressions on that subset.

This separates real progress from a blog post that went mildly viral in a country you do not sell to.

4. Clicks and click-through rate on your top twenty pages

Reported per page, not sitewide. A page ranking at position 4 with a 1% click-through rate has a title and description problem, not a ranking problem, and it is the cheapest fix in SEO. Sitewide CTR hides this completely.

Two queries on a page one position with zero clicks is a specific, actionable finding. A sitewide CTR of 2.3% is not.

5. Enquiries attributed to organic, however roughly

This is the number the client actually cares about and the number most reports avoid because attribution is imperfect. Imperfect is fine. Count form submissions and calls where the session source was organic, state the margin of error, and report it anyway. A report without it is asking the client to take the rest on faith.

What should move off page one

Keyword rankings from a rank tracker as the headline. Useful as a detail, misleading as a summary, and increasingly divorced from what users see given personalisation and local variation.

Backlink counts. Ten links from real sites beat two hundred from a network, and a raw count invites exactly the wrong behaviour from whoever is being measured on it.

Domain authority or similar third-party scores. Not a Google metric, not used in ranking, and noisy month to month.

Bounce rate. Changed definition, frequently misread, and rarely actionable on a service site.

The structure that makes a report usable

Three sections. What moved and why. What we did this month. What we are doing next month and what we expect it to move.

The third section is the one that makes the report a decision document rather than a receipt. It also creates accountability in the only direction that matters — next month's report has to open by saying whether the prediction held.

A report that cannot explain why a number moved is a report where nobody looked. And a report where every number went up every month, without exception, is usually a report where the metrics were chosen to guarantee that outcome.

For the version of this that covers what to look for in the proposal before the reporting even starts, read the full breakdown here: https://deepbhardwaj.com/seo-expert-in-india/

Related reading on separating an audit from a strategy: https://indiaseofieldnotes.mataroa.blog/blog/an-seo-audit-is-not-an-seo-strategy-and-mixing-them-up-costs-you-a-quarter/

Replatforming is the single most reliable way to lose organic traffic in India, and it is almost never the new platform's fault. WordPress to Shopify, custom PHP to WordPress, Magento to Shopify — the pattern is identical. Traffic holds for a fortnight, then falls thirty to sixty percent over six weeks, and by the time anyone connects it to the launch the diagnosis is three months late.

The cause is nearly always the same: URLs changed and the redirect map was incomplete, wrong, or chained.

The four failure modes, in order of frequency

Blanket redirect to the homepage. Every old URL points at the root. It looks tidy, it returns 200, and it destroys every page-level ranking the site had. Google treats a redirect to an irrelevant page as a soft 404 and drops the old URL's equity entirely. This is the worst outcome and the most common one.

Redirect chains. Old URL redirects to an intermediate, which redirects to the new page, sometimes via a www or http hop. Each hop is a small loss and a crawl cost, and chains longer than three hops are frequently not followed at all.

Parameter and trailing-slash inconsistency. The new platform serves both /page and /page/, or both with and without a query string, and nothing canonicalises. Signals split across duplicates.

Whole sections quietly missing. The old site had 900 blog posts; the new one has 400 and nobody produced the list of the 500 that were dropped. Every one is a 404 with inbound links.

How to run the audit in an afternoon

Get the complete old URL list. Not the sitemap — the sitemap only had the pages someone remembered. Pull from Search Console's Pages report and Performance report (export every URL with impressions in the last sixteen months), your analytics landing-page report for the same window, and any server log you can get. Deduplicate.

Request every one of them and record the final destination. A simple script that follows redirects and reports the start URL, the number of hops and the final URL and status code. This one table is the audit.

Sort the table into four groups. Landing on a relevant new page in one hop: correct, ignore. Landing on the homepage: broken, needs a real target. Two or more hops: rewrite the rule to point directly at the final destination. Returning 404 or 410: either restore the page or map it to the closest relevant equivalent.

Decide honestly about the leftovers. Not every old URL deserves a redirect. A discontinued product with no traffic and no links can 410. But a page with inbound links or historical impressions should always land somewhere relevant, because that is where its accumulated authority goes.

Then check what Google thinks

Two Search Console reports tell you whether the fix landed. The Pages report's “Not found (404)” and “Page with redirect” buckets should shrink week over week. The Performance report filtered to the old URLs should go to zero impressions while the new ones pick them up. If old URLs still draw impressions two months after launch, the redirects are not being followed.

Resubmit the new sitemap, and ping IndexNow if the platform supports it, so Bing and Yandex pick up the change quickly rather than waiting to recrawl.

The preventable version

If a replatform is still ahead of you rather than behind you, the redirect map is built before launch, not after: export the old URL list, map every row to a new URL by hand, and treat any unmapped row as a launch blocker. It takes a day and it saves a quarter.

A fuller walkthrough of the technical checks that belong in this audit is here: https://deepbhardwaj.com/seo-expert-in-india/

Related: robots.txt, staging leaks and stray noindex — the monthly ten-minute audit

Every coaching institute in Delhi NCR competes for the same dozen head terms — “best IAS coaching in Delhi”, “NEET coaching Mukherjee Nagar”, “CAT coaching in Noida” — and those terms are owned by aggregators, review sites and a handful of institutes with decade-old domains. Meanwhile there is a large, well-defined body of search demand in the same category that almost nobody writes for, and it converts better.

The mistake is treating the category as one keyword contest. It is four different ones, and only one of them is crowded.

The four query types

Brand-plus-location. “Vajiram and Ravi fees”, “Aakash Institute Rohini timings”. If you are the institute, these are yours and you lose them only by having no page that answers the question. Fee structure, batch timings, branch address, admission process — each deserves a real page, not a PDF and not a contact form. Parents and aspirants search for fees before anything else, and a site that hides them sends the visitor to a forum where the number quoted is wrong.

Exam-plus-question. “CAT 2027 syllabus change”, “UPSC optional subject with highest success rate”, “NEET biology weightage chapter wise”. This is the largest bucket by volume, it is informational, and it is where institutes should be winning and mostly are not. These queries are cheap to rank for, they build the audience six to twelve months before the enrolment decision, and they are the only reliable way to accumulate the topical authority needed to eventually compete on the head terms.

Comparison. “Vajiram vs Vision IAS”, “online vs offline coaching for NEET”, “best coaching for CAT in Delhi or online”. Uncomfortable but high-intent. Aggregators own these because institutes refuse to write them. An honest comparison page that says who your programme is not right for outperforms an evasive one, and it ranks because it actually answers the query.

Location-plus-category. The crowded bucket. Worth one strong page per branch and per major exam, not worth building a strategy around.

Where the effort should go

Three-quarters of the content effort belongs in the second bucket. A page per syllabus topic, per exam change, per commonly confused rule — written by faculty, not by a content agency working from competitor pages. The signature of institute content written without subject knowledge is that it is correct and generic; it ranks for nothing because there are four hundred equally generic versions of it.

The pages worth the most are the ones only you can write. Cut-off analysis from your own students' data. What actually changed in this year's pattern and what it means for preparation sequence. Why a particular optional subject is being over-recommended. This content is defensible, it gets shared in aspirant communities, and it earns links without asking.

Two structural things

Result pages are an asset most institutes waste. “Our results” as a wall of photographs contributes nothing. The same information as a structured page — selection counts by exam and year, rank ranges, batch sizes — answers “institute name results 2026” queries and is the strongest trust signal the site has.

Batch and admission pages need dates in text, updated. A page still advertising a batch that started four months ago tells a visitor the institute does not maintain anything, and it tells Google the page is stale.

On timing

Enrolment is seasonal and search demand for informational exam queries peaks six to nine months before the exam, not at admission time. Content published during the admission rush arrives after the audience has already decided. The publishing calendar should run about two quarters ahead of the sales calendar, which is the single scheduling change that most improves results in this category.

Institutes that do this consistently end up ranking for head terms as a by-product, two or three years in, because they have built the topical depth that the aggregators only simulate.

If the site needs the technical foundations sorted before any of this content work pays off, that is worth checking first — the SEO services in India side of the work is usually a prerequisite rather than an alternative.

Related: https://indiaseofieldnotes.mataroa.blog/blog/thin-service-pages-why-so-many-indian-websites-stall-at-position-15/

Every Delhi NCR service business eventually asks this: do we build separate pages for Delhi, Noida, Gurgaon, Faridabad and Ghaziabad, or one strong page for the whole region? Both answers are defensible and the wrong choice is expensive, so it is worth thinking about properly rather than copying whatever the last agency did.

The case against five locality pages

Most NCR locality pages are the same page with the city name swapped. Search engines treat that set as near-duplicate content, pick one, and largely ignore the rest. Meanwhile the five pages compete with each other for the same queries, so none of them accumulates the internal links, the external links or the engagement signals that would let any single one rank.

I audit this pattern constantly. A business has eleven city pages, ten of them have never received an organic visit, and the eleventh is outranked by the homepage for its own city.

When separate pages genuinely are correct

They work when each page has something real and different to say. A physical office or service centre in that city. Locally different pricing, timelines or coverage. Staff who work from there. Clients, projects or case references from there. Locally specific regulation – and in NCR this is more common than people assume, because Delhi, UP and Haryana genuinely differ on registrations, permits and compliance.

If you can write four hundred words about that city that could not be written about any other, the page earns its existence. If you cannot, it does not, and building it makes your site weaker.

The structure that works for most NCR businesses

For a service business with one office and region-wide delivery, the strongest structure is usually: one substantive page per service, each covering how the service works across NCR, plus honest coverage detail – which areas are serviced same-day, which take longer, what the travel or logistics reality is. Then a single well-built location page for the office you actually have.

That gives search engines one clear strong page per service instead of five weak ones per city, and it gives the visitor the answer they actually want, which is whether you can serve them where they are.

The Google Business Profile point people miss

Local pack rankings in NCR are driven overwhelmingly by proximity to the searcher and by your Google Business Profile, not by whether you have a page titled with their city name. A business in Faridabad will struggle to appear in the local pack for a Noida searcher regardless of how many pages it builds, because the distance is the dominant factor.

This is why locality pages disappoint so often: businesses build them expecting local pack visibility and get organic visibility they could have had from one better page. If local pack presence across NCR is genuinely the goal, that is a question about physical locations and verified profiles, not about page count.

A practical test before you build anything

Take the city page you are considering. Write the first four hundred words without using the city name. If what remains is indistinguishable from your main service page, you do not have a locality page, you have a duplicate. Do not publish it.

If you already have the thin pages

Do not delete them in a batch. Consolidate: pick the strongest page, merge the genuinely unique content from the others into it, and 301 the rest to it. You keep whatever equity the old URLs had and you stop the internal competition. Give it four to six weeks before judging the result.

The order I would work in

Get one service page per service right, with real NCR coverage detail. Get the Google Business Profile for the office you actually have properly built out. Only then consider a second location page, and only if you have a second location. Locality pages are the last thing to build, not the first, and most NCR businesses never need more than one.

For a view on which structure suits your particular service mix, Deep Bhardwaj has worked through this with Delhi NCR businesses on both sides of the argument.

Related reading: why Indian B2B sites lose rankings after a CMS migration – consolidation and migration break in the same ways.

Every Mumbai business that has looked at its Google Business Profile has been told to get more reviews. Almost none of them have been told the part that actually matters, which is the rate, the recency and the substance – not the total.

A restaurant in Bandra with 800 reviews, none in the last eight months, will routinely be outranked in the local pack by a competitor with 140 reviews arriving steadily at ten or twelve a month. The count is a vanity number. The flow is the signal.

Why rate beats total

The local pack is trying to answer a question about the present tense: which of these places is currently operating, currently busy, currently good. A review from three years ago is weak evidence for any of that. A review from last Tuesday is strong evidence for all of it.

This is why the burst strategy fails so reliably. A business runs a campaign, collects sixty reviews in ten days, sees a brief lift, and then watches it decay over the following quarter as the reviews age and nothing replaces them. Worse, a sudden spike against a flat baseline is exactly the pattern spam detection is built to catch, and filtered reviews can take the legitimate ones down with them.

Steady beats spiky. Eight reviews a month, every month, outperforms ninety-six reviews in January.

What the text is doing

Star ratings are only part of what gets read. The review text is indexed, and the words in it associate your profile with the services and the localities people search for.

A review that says “great service” contributes a star and nothing else. A review that says “got my iPhone screen replaced in Andheri West, done in forty minutes, fair price” contributes a star plus service terms plus a locality plus an implicit speed claim. Ten reviews of the second kind will do more for visibility than a hundred of the first.

You cannot script this without breaking the rules, and you should not try. What you can do is ask a question rather than make a request. “Would you mind mentioning which service you had done and how it went?” produces substantive text naturally. “Please leave us a 5-star review” produces one-word reviews.

The ask, and when it works

The window is narrow. Satisfaction decays fast – a request sent within a few hours of a completed job converts several times better than one sent a week later.

In practice the highest-converting method in Mumbai is a WhatsApp message with the direct review link, sent the same day by the person who actually did the work, in the language the customer was speaking. Not a bulk SMS from a number nobody recognises. Not an email that lands in Promotions. The personal channel, the same day, from a familiar name.

Ask everyone, not just the delighted. A profile with nothing but five-star reviews reads as manufactured to humans and to the algorithm. Some four-star reviews with real detail make the whole profile more credible.

Responses are not optional

Every review should get a reply, ideally within a day or two. Responses are read by search engines and by prospects, and they are the one part of the review ecosystem you fully control.

A good response references the specific service and the location naturally, because that text is indexed too. A good response to a negative review acknowledges the problem, states what changed, and offers a channel – and it is aimed at the next reader, not the complainant. Prospects read the one-star reviews first and they are really reading your reply.

What to actually track

Four numbers, monthly. New reviews received. Average age of the most recent ten. Percentage that mention a specific service or locality. Percentage that received a response. Track those four and the total review count takes care of itself.

A related failure mode worth reading if a rebuild is planned: why Mumbai businesses lose local pack visibility after a website redesign.

Businesses that want the review flow built into the operational routine rather than run as a campaign usually need the Business Profile, the citations and the website's local signals aligned at the same time – the pack and the organic results respond to different inputs and both have to be fed. Details and contact here: https://deepbhardwaj.com/seo-expert-in-mumbai/

GEO for Indian Brands: Making Your Site Quotable by AI Assistants

A growing share of commercial research now happens inside an assistant rather than on a results page. Someone asks ChatGPT or Perplexity to shortlist suppliers, compare two options, or explain what a service should cost, and the answer names three or four sources. If your brand is not among them, you did not lose a ranking – you were never in the consideration set.

Generative Engine Optimisation is the work of being the source an assistant reaches for. It overlaps with SEO but the unit of success is different: not a click, but a citation.

What actually gets cited

Having watched which pages get pulled into assistant answers about Indian services and products, the pattern is consistent. Cited passages tend to be:

Specific rather than general. “Prices vary depending on requirements” is unquotable. “Audit-plus-implementation retainers for a mid-sized Indian service business typically run between X and Y per month, with the variance driven mostly by content volume” is quotable, because it answers the question the user asked.

Structured as a direct answer. A heading phrased as the question, followed immediately by a two-to-three sentence answer, then the detail. Assistants extract the answer block. Content that buries the answer in paragraph six of a narrative essay does not get extracted, however good the essay.

Numbered, dated and attributable. A figure with a source and a date survives extraction. A figure floating in prose gets dropped as unreliable.

Honest about limits. Passages that include the cases where something does not apply are cited disproportionately often, because they read as expertise rather than marketing. This is the single most counter-intuitive finding and the easiest to act on.

The entity problem, which is where most Indian brands lose

An assistant will not confidently name a brand it cannot resolve as a consistent entity. If your business appears as three different names, with three different addresses and two phone numbers, across your site, your Google Business Profile, your directory listings and your LinkedIn page, the model has no stable thing to refer to.

Fixing this is unglamorous and high-leverage:

• One exact legal or trading name, spelled identically everywhere, including punctuation
• One address format, one phone number format
• Organization or LocalBusiness structured data on the site that matches those values exactly
• sameAs links from the structured data to your verified profiles
• Founder and key personnel named consistently, with their own bio pages

The goal is that every mention of the brand across the web reinforces a single entity rather than fragmenting into several.

Practical checklist

1. Add an FAQ block to every commercial page, using the questions customers actually ask – not invented questions engineered around keywords
2. Put a two-sentence direct answer immediately under each H2, before the elaboration
3. Give every page a clear “last updated” date and keep it honest
4. Publish at least one genuinely comparative page per service, including where a competitor or alternative is the better fit
5. Name your authors, with credentials, and give them bio pages
6. Audit your name, address and phone consistency across every listing you can find
7. Keep llms.txt and a clean robots.txt so assistant crawlers can read what you want read

What not to bother with

There is no keyword density trick for assistants, no meta tag that requests citation, and no submission form. Attempts to manipulate assistant answers with keyword-stuffed pages fail for the same reason they fail in search: the extraction is semantic, not lexical.

The durable version of this work looks like publishing genuinely better-structured, more specific, more honest content than competitors – which is also what has always worked in search, applied with a different output format in mind.

Brands that want an entity and citation audit run across their site and listings can get it from Deep Bhardwaj: https://deepbhardwaj.com/seo-expert-in-india/

Related reading: https://indiaseofieldnotes.mataroa.blog/blog/how-ai-search-engines-choose-which-indian-websites-to-cite/

Every site migration that goes badly goes badly for the same reason: somebody started making changes before they had a complete record of what existed. Once the old site is gone, you cannot reconstruct what you had, which means you cannot tell what you lost or how to get it back.

This is the export list. Do all of it before anyone merges anything.

1. Every indexed URL

Search Console's Pages report, filtered to indexed, exported in full. This is Google's view of your site, which is the view that matters.

2. A full crawl of the live site

Screaming Frog or equivalent. URLs, status codes, titles, meta descriptions, H1s, canonicals, meta robots, word count. This is your site's view of itself.

3. The XML sitemap, as a file

Not a link to it. The actual file, saved.

4. Top pages by organic traffic, twelve months

From analytics, not from Search Console alone. You need the pages that produce revenue, not just the ones that get impressions.

5. Top queries by page, sixteen months

Sixteen, not twelve, so you have a year-over-year comparison available after launch. Search Console caps at sixteen months and it disappears as it ages.

6. Current rankings for your target keyword set

A dated snapshot from whatever rank tracker you use. Without a pre-migration baseline, every post-launch conversation becomes an argument about whether a drop is real.

7. Backlink data

Every referring domain and the specific page it points to. When you change URLs, you need to know which ones have external links worth preserving with a precise redirect rather than a lazy one to the homepage.

8. The existing redirect rules

Sites accumulate redirects over years. If you replace the server config wholesale, you break chains that were carrying value from URLs retired in 2019.

9. Structured data currently in use

Export the JSON-LD from your template types. Redesigns routinely drop schema because it lives in a template nobody thought to port.

10. robots.txt and any noindex directives

Saved as files, with a note about why each rule exists. Half of them will turn out to be load-bearing.

11. Core Web Vitals field data

The current values, per page group, from Search Console. Field data has a 28-day rolling window, so the pre-launch numbers vanish if you don't capture them.

Then build the map

With all eleven exports in hand, build the redirect map: old URL, new URL, one row per URL, no blanks. Every old URL gets a destination that is genuinely equivalent in content. Not the homepage. Not a category page. The equivalent page, or the closest thing that exists.

Test the map against staging before launch. Test it again the hour after launch. Then watch Search Console's coverage report daily for two weeks.

The one that catches everyone

Staging environments get indexed. Before launch, confirm the staging site is behind authentication, not merely noindexed – and after launch, confirm the production site is not still carrying the staging robots rules. That single inverted check is responsible for more migration disasters than every other item on this list combined.

Migration work is boring, and the payoff is entirely invisible when it goes right. That's the job.

Reference: https://deepbhardwaj.com/seo-expert-in-mumbai/

Related: Why Mumbai businesses lose local pack visibility after a redesign