SEO & AI SEARCH 19 min read · Updated 31 August 2026

SEO and GEO in 2027: ranking on Google and inside AI answers

What changed, what still works, and the specific technical and content work that gets a page cited by ChatGPT, Gemini, Perplexity and Google AI Overviews — not just ranked.

Careers Ninza SEO faculty

Careers Ninza · Kolkata, India

The mechanics of being found have split in two. Google still ranks pages, and AI systems now assemble answers by retrieving and citing sources. Both matter, they overlap substantially, and the work that serves both is largely the same — just done more deliberately than classic SEO required.

What actually changed

ThenNow

Rank for a keywordBe the source an answer is built from

Traffic was the only outcomeCitation and recall matter without a click

Keywords drove relevanceEntities and relationships drive it

One algorithm to satisfyGoogle, Bing, and several retrieval systems

Content volume workedVerifiable specificity works

The most consequential shift is that informational queries increasingly resolve without a click. A page answering "what is data analytics" gets summarised. A page answering "data analytics course fees in Pune with EMI" gets clicked, because the reader needs specifics only you hold.

The practical response is not to abandon informational content — it still earns citation and brand recall. It is to stop measuring it on clicks alone, and to shift the weight of new content towards queries where a decision or transaction requires visiting your page.

The foundation, unchanged and still decisive

1Server-rendered HTML. If your content only appears after JavaScript executes, many AI crawlers never see it. This single issue silences more sites than any other and it is invisible in a browser.

2Genuine crawlability. Confirm pages are indexed, not merely submitted. Fix noindex leftovers, blocked resources and orphan pages first.

3Speed on real Indian mobile connections. LCP under 2.5s measured on a mid-range Android on 4G, not on your desktop.

4One canonical URL per piece of content. Duplicates split every signal you have.

5Internal linking that carries meaning. Descriptive anchors, every important page within three clicks. Retrieval systems infer topical structure from this.

6AI crawlers permitted in robots.txt. GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended. Blocking them removes you from AI answers entirely.

Entity and schema work: the highest-return item

Models reason about entities and relationships, not keyword density. Schema is how you state what you are unambiguously, and most Indian business sites still have none.

Implement as a single linked @graph rather than isolated blocks, so a machine can follow the connections:

—Organization or LocalBusiness — identity, address, geo coordinates, areaServed, knowsAbout, and sameAs pointing at verified profiles

—WebSite with SearchAction

—WebPage with dateModified, so freshness is machine-readable

—BreadcrumbList for hierarchy

—Product, Course, Service or Event with real Offer prices

—FAQPage — the single most-quoted type in AI answers

—Article with named author and publisher for provenance

Never mark up content that is not visible on the page. It breaches every search engine guideline and is the fastest route to losing rich results permanently.

Content patterns that get cited

Retrieval systems extract claims and attribute them. These patterns make that easy, and they are testable — ask an assistant your key question and see who it names.

1Answer in the first two sentences under each heading. Then elaborate. Models lift those sentences.

2Use question-shaped headings. "What does a data analyst earn in Bengaluru?" outperforms "Salary information" consistently.

3Publish specifics. Named prices, dates, cities, durations. Vague claims are unciteable because nothing concrete can be attributed.

4Put relationships in tables. Comparisons in genuine HTML tables get extracted cleanly; images of tables are invisible.

5Keep a visible last-updated date. Recency is weighted heavily and an undated page reads as potentially stale.

6State limitations honestly. Pages that say what does not work read as more trustworthy to both readers and models, and they get cited for the caveat as well as the claim.

llms.txt and machine-readable summaries

A plain-text file at your root summarising what your site offers, written for assistants rather than crawlers. Adoption is growing across AI tools and the cost is one file.

Include your identity and location, your products or services with real prices, your key pages, and a short factual section correcting things a model might otherwise get wrong — accreditation status, guarantees you do not make, how pricing actually works. That last section is the most useful and the least common.

Local: ranking across Indian cities

For most Indian businesses this is where revenue sits. Someone searching "digital marketing course in Indore" is far closer to buying than someone searching "what is digital marketing".

Google Business Profile

1Complete every field, not just required ones. Completeness correlates directly with local ranking.

2Most specific primary category available, then secondaries.

3Real photographs updated monthly. Stock imagery underperforms visibly.

4Reviews collected steadily, with a reply to every one including critical ones.

5Weekly Posts for batches, offers and events — free surface area most competitors ignore.

6Q&A seeded with real questions and honest answers.

City pages, and the trap

City pages work. Mass-produced city pages get penalised. The difference is whether the page contains anything genuinely local.

A page that is your service page with the city name swapped in is a doorway page, penalised explicitly since 2015. Generating 500 of them is among the fastest ways to damage a domain. Build real pages only where you have local substance — batch timings, local roles, local partners — and cover every other city with one well-built locations hub.

For each priority city include locally relevant content, city-worded FAQs, a Service node with areaServed scoped to that city, internal links to nearby cities in the same state, and an honest statement of whether you operate there physically or serve it remotely.

What to measure now

SignalWhereWhat it tells you

Impressions vs clicksSearch ConsoleDivergence means AI answers are absorbing traffic

Citation presenceQuery the assistants monthlyWhether models name you at all

AI referral trafficGA4 referral sourcesActual visits from AI surfaces

Branded search volumeSearch ConsoleWhether exposure is building recall

Local pack visibilityManual checks per cityCity-level competitiveness

Enquiries per sessionYour own CRMWhether the traffic mix improved

Impressions rising while clicks stay flat is the AI-answer signature. It is not necessarily failure — it can mean growing citation and recall — but you cannot see that if clicks are your only metric.

WHERE THIS APPLIES

Local search work applies city by city, and the same technical and schema foundation serves every one of them simultaneously. Careers Ninza teaches this live online, so learners join from Kolkata, Delhi NCR, Mumbai, Bengaluru, Hyderabad, Pune, Chennai, Ahmedabad, Jaipur, Chandigarh, Lucknow, Indore, Nagpur, Coimbatore, Asansol, Durgapur, Siliguri, Patna, Ranchi, Bhubaneswar, Guwahati, Gwalior, Noida and Gurugram. See all locations →

Market opportunity

The gap in Indian SEO is not knowledge of keywords — it is technical implementation. Most sites we audit have no structured data, client-side rendered content, no llms.txt, and robots.txt files that block the very crawlers feeding AI answers. Each is fixable in days.

That means the people who can implement entity and schema strategy, diagnose rendering problems and optimise for citation are scarce relative to demand, and the work is provable in a way most marketing is not. It is also one of the most defensible freelance skills, because the results compound for the client and the expertise transfers across every industry.

Why Careers Ninza

—Worked on a real site. Twenty hours of audit, fix, publish and measure — yours or one we provide.

—Schema and entity work taught properly, including the linked @graph approach rather than isolated blocks.

—AI citation on the syllabus. Content patterns, llms.txt, and how to test whether assistants name you.

—Local ranking for Indian cities including the doorway-page trap most agencies walk into.

—Practitioner mentors doing this work on live sites this quarter.

—₹9,999 for 20 hours, live, evenings and weekends, with lifetime recordings.

SEO & AI Search Mastery is 20 hours at ₹9,999, worked on a real site — technical audit, topic architecture, schema, AI citation and measurement.

SEE THE COURSE

Frequently asked questions

What is GEO and how is it different from SEO?+

GEO, generative engine optimisation, is about being cited inside AI-generated answers rather than ranked in a list of links. It overlaps heavily with SEO — strong technical foundations, schema and clear content serve both. The differences are answer-first writing, entity clarity, and permitting AI crawlers in robots.txt.

How do I get my website cited by ChatGPT, Gemini and Perplexity?+

Allow their crawlers in robots.txt, render content server-side so it is visible without JavaScript, answer questions in the first two sentences under question-shaped headings, implement Organization, FAQPage and Article schema, publish specific verifiable facts such as prices and dates, and keep a visible last-updated date.

Is SEO still worth doing in 2027?+

Yes, but the traffic mix has changed permanently. Informational queries increasingly resolve without a click, while queries requiring a decision — pricing, comparisons, local availability, booking — still send visitors. Shift new content weight towards those and measure impressions and citations alongside clicks.

What is llms.txt and should I publish one?+

A plain-text file at your domain root summarising what your site offers, written for AI assistants rather than crawlers. Include your identity, location, products with real prices, key pages, and a factual section correcting things a model might otherwise state wrongly. Adoption is growing and the cost is one file.

Does schema markup actually help ranking?+

Schema is not a direct ranking factor, but it substantially improves how machines understand your page, which drives rich results, local pack eligibility and AI citation. In 2027 it is among the highest-return technical work available, largely because so few sites implement it properly.

Should I create a separate page for every Indian city?+

No. Mass-produced city pages that only swap the city name are doorway pages, penalised explicitly since 2015, and generating hundreds is among the fastest ways to damage a domain. Build real pages only where you have local substance, and cover the rest with one well-built locations hub.

How do I rank in local search in Indian cities?+

Complete every field on your Google Business Profile, choose the most specific primary category, add real photos monthly, collect reviews steadily and reply to all of them, post weekly, seed the Q&A section, and keep your name, address and phone identical across every directory.

How should I measure SEO now that clicks are unreliable?+

Track impressions and clicks separately in Search Console, since divergence signals AI absorption. Query the major assistants monthly and record whether they name you. Watch AI referral traffic in GA4, branded search volume, local pack visibility per city, and enquiries per session from your own CRM.

Why is server-side rendering important for AI search?+

Many AI crawlers do not execute JavaScript, so content that only appears after JS runs is invisible to them. This single issue silences more sites than any other and it is undetectable in a normal browser, because your browser renders the JavaScript that the crawler never runs.

What does the Careers Ninza SEO course cover and cost?+

SEO & AI Search Mastery is 20 hours at ₹9,999, worked on a real site: technical audit and fixes, topic clusters and intent mapping, on-page and internal linking, schema and entity markup as a linked graph, getting cited by AI answer engines, local ranking for Indian cities, and Search Console measurement.

We teach this, live

Every article here comes from something we teach. Sit in on a free masterclass and judge the mentors yourself.

FREE MASTERCLASS CALL +91 90888 39993