SEO Basics: How to Get Found on Google Without Paying for Ads

Search engine optimisation. SEO. It's the work of getting your pages to show up in Google's unpaid results, the ones nobody's bidding on. That's the appeal. The traffic doesn't stop the day you stop paying.

The common failure looks like this. A business publishes a batch of pages with the target phrase in every place a checklist tells them to put it. Page title. Main heading. That description snippet that shows under the title in search results. The body copy. Months pass. The traffic report is flat. The conclusion is usually "SEO doesn't work any more" or "you need a budget for it now".

Neither is the actual reason. Those pages optimised the parts of SEO that are easy to change and skipped the two that decide the outcome. Does this page give the searcher what they actually came for? And is there any reason for Google to prefer it over the ten pages already answering that query? Tags, structure, speed, schema. That's hygiene. Hygiene makes you eligible. It doesn't make you rank.

Put it another way. The checklist stuff is the price of admission. It gets your page in the room. Whether Google picks yours from the room is a different question, and that one turns on match and merit rather than on how many boxes you ticked.

How Google actually works: crawling, indexing and ranking

Three separate stages. A page can fail at any one of them, and the fix is different each time.

Crawling is discovery. Googlebot, Google's crawler, follows links between pages and reads sitemaps (a file listing your URLs) to find things. If nothing links to a page and it's not in a sitemap, it may simply never be found.

Indexing is storage and understanding. Google fetches the page, renders it, works out what it covers, and decides whether to keep a copy. Google doesn't index everything it crawls. Near-duplicate pages get dropped. Near-empty pages get dropped. Pages carrying a noindex tag get dropped. This is the stage most beginners don't know exists.

Ranking happens per query, at query time. Google picks from the index and orders results by relevance, quality signals, and the searcher's location, language and device.

Here's the distinction that saves months. Being indexed is not being ranked. Search site:yourdomain.com in Google. If your page doesn't appear, you've got a crawl or index problem and no amount of keyword work will help. If it appears but you're getting no traffic, you're eligible and losing. That's a different job entirely. Search Console's URL Inspection tool tells you which of the two you have, per URL, and usually why.

Worth setting straight early. Submitting a sitemap or clicking "Request indexing" is a hint. Not a command. It helps Google find a page. It doesn't oblige Google to keep it.

The unit of SEO is a query, not a website

You don't rank on Google. A page ranks for a query. Your site has no rank in aggregate. It has a list of queries it's eligible for, decided page by page, separately.

That reframing changes the task. "Add SEO to the homepage" isn't a task, because a homepage announces a company rather than answering a question. "Publish a page that fully answers how do I calculate customer acquisition cost" is a task. There's a query. There's a searcher. There's a page with one job.

So the working unit is one page, one job, one cluster of queries that all want the same answer. cost per acquisition formula, how to calculate CPA and what is a good CPA can share a page. CPA vs CPC wants a comparison and deserves its own. The underlying definitions of CPC, CPA and ROAS are a third page again.

The test for whether two queries belong on the same page. Search both. Compare the results. If Google returns largely the same pages for both, it's treating them as the same question and one page can serve both. If the two result sets barely overlap, you need two pages. That test does the job a keyword tool's similarity score only estimates, and it costs nothing.

The four search intents, and why intent beats every on-page tweak

Search intent is why someone typed the query. The four-part model below is an SEO-industry framework rather than an official Google taxonomy. The original academic work described three types and the industry added the fourth. It's still the most useful lens I know for deciding what a page should be.

Intent	What the searcher wants	Example query	Page type that satisfies it
Informational	To understand or learn	what is retargeting	Guide, explainer, tutorial
Navigational	To reach a specific place	google ads login	The actual destination page
Commercial investigation	To compare before deciding	best CRM for small business india	Comparison or review with stated criteria
Transactional	To act or buy now	hire performance marketing agency delhi	Service or product page with a clear next step

The practical move. Search your target query and read what already ranks. Not to copy the content. To read the format. If page one is ten how-to guides, Google has concluded searchers for this query want to learn, and your service page won't rank there however well its title is written. If page one is all product pages, a 3,000-word essay won't rank either.

That's why intent match beats on-page work so decisively. On-page optimisation adjusts how clearly you signal what a page is about. Intent match decides whether the page is the right kind of thing in the first place. You can't tag your way out of being the wrong format.

Two more things to read while you're looking. The shape of the results (guides, product pages, videos, a map pack). And the features on the page (People Also Ask, shopping carousels, an AI summary above the results). A query dominated by a map pack is a local query, and your answer is a Google Business Profile, not a blog post.

Keyword research with no paid tools

You don't need a subscription to find out what people search. You need to be systematic about the free sources.

Google autocomplete. Type your topic and stop. Then type it followed by each letter of the alphabet. Then prefix it with how, why, what, best, cheap, vs, near me, for. Every suggestion is a query pattern Google has actually seen.

People Also Ask. Search your topic, expand a PAA question, and the box generates more. Keep expanding and you get question-shaped queries in the exact language searchers use. These map close to one-to-one onto H2s and FAQ entries.

Related searches at the foot of the results page. This is where Google shows you the adjacent topics it associates with yours. It's how you find the parts of a subject you didn't know belonged to it.

Reddit, Quora and trade forums. People write questions there the way they'd say them aloud, which is closer to how they search than marketing copy ever is. Search site:reddit.com [your topic] to mine phrasing.

Your own inbox and sales calls. The questions prospects ask before buying are commercial-intent queries with a buyer attached, already in their words.

Search Console's Performance report, once you have any traffic at all. The Queries tab shows what you already get impressions for. Queries sitting on page two with real impressions are the ones you're closest to winning. You're already eligible. You just need a better page. Read the Queries tab knowing that a large fraction of your actual traffic comes from anonymised long-tail queries that never appear in the table at all. Your Search Console numbers are lying to you in five specific ways covers the practical implications, most notably that your "top queries" are the top queries above the privacy threshold, not the top queries overall.

One caution on Google Keyword Planner. Without an active campaign spending money, it reports volume in wide bands rather than precise figures. Use it for relative comparison, not for planning. Group the resulting list with the overlap test above and label each cluster with its intent.

Pick queries you can win: the long-tail rule

A domain registered three months ago won't outrank an established authority site for digital marketing or best credit card. Not with better writing. Not with a perfect title tag. Not with 5,000 words. Those results are held by sites with years of accumulated links, brand recognition and topical track record. No on-page work closes that gap.

What you can win is specificity. Long-tail queries (longer, narrower, usually phrased as a problem) have fewer pages competing for them and often no page that answers them properly.

	Head term	Long-tail, problem-shaped
Example	email marketing	why do my emails go to promotions tab gmail
Monthly searches	High	Low
Competing pages	Enormous, well-linked	Few, often poor
Searcher clarity	Vague	Knows exactly what they want
Winnable on a new domain	No	Yes
Likelihood of an enquiry	Low	High

You're buying position with specificity because you can't yet buy it with track record. Someone typing why is my Google Ads spend so high with no conversions is much closer to hiring help than someone typing google ads. And unlike the head term, that query is realistically winnable.

The pattern to hunt is symptoms. "Why is my X doing Y". "X not working after Z". "How do I fix X when Y". These are the queries where someone who's actually solved the problem beats a content team that hasn't.

On-page SEO basics, done properly

On-page work is real. It's just not the lever people think it is. Done properly it means this.

The title tag. The clickable headline in the results and the text in the browser tab. It's a promise, and it earns the click. Front-load the words that match the query and make it specific. SEO Services | Best Digital Marketing Agency | Home wastes three slots. How to Fix Pages Not Indexed by Google is a promise. Google truncates titles by pixel width rather than character count, so roughly 55-60 characters is a proxy, not a rule.

An H1 that matches the query language. The H1 is the main visible heading on the page. Use the words your reader would use, not internal jargon. Google has said multiple H1s won't break anything, but one is still the better discipline. A page with a single top-level heading has usually decided what it's about.

A first paragraph that answers the question. If someone lands on a page called "how to calculate CPA", the answer should be visible within two or three sentences. Warming up with "in the world of digital marketing, metrics matter more than ever" tells the reader you have nothing and sends them back to the results.

A short, readable URL. /blog/how-to-fix-not-indexed beats /blog/2026/08/post-id-4471?cat=seo.

Descriptive subheads. Each H2 should answer a distinct question someone would actually type. That's how you cover a keyword cluster. By answering more of it, not by repeating the phrase.

Descriptive image alt text, written for someone who can't see the image. Accessibility first, search second.

Two things to stop worrying about. The meta description (the snippet under the title) isn't a ranking factor. It influences whether people click, which matters, but it doesn't move position, and Google often rewrites it anyway. And there's no keyword density figure to hit. Write the page so a knowledgeable reader would call it complete, and the vocabulary takes care of itself.

Why thin content fails, even at 2,000 words

"Thin" doesn't mean short. A 300-word page that completely answers a narrow question isn't thin. A 2,000-word page that circles a topic without ever giving a number, a step or a decision is thin. And it's the more common failure.

Thin means the page doesn't finish the job. The signs:

• No specifics. Every claim is "it depends", with no framework for what it depends on.
• No arithmetic where arithmetic is what the reader needs.
• No answer to the obvious follow-up. If the page explains what retargeting is, the next question is what the audience window should be. Leave it out and the reader goes back to the results.
• No next step. The reader finishes and can't do anything differently tomorrow.

The test I'd apply before publishing. Could a reader act on this without opening another tab? If yes, the page is done. If no, it's thin regardless of length.

This kills the instinct to publish thirty 400-word pages against thirty keywords. Thirty pages that each answer nothing is thirty pages of nothing. Five pages that each close a question will outperform them, and they stand some chance of earning a link, which stubs never do. Word count isn't a ranking factor. Length is a consequence of finishing the job, not a target.

The other way to think about it. A thin page is one the reader has to leave to complete the task they came for. A finished page is one they can act on before closing the tab. That's the only bar that matters.

Internal linking: the lever you control completely

Links from your own pages to your own pages do two things. The anchor text describes the destination, which helps Google understand what that page covers. And the link passes whatever authority your site has towards the pages you most want to rank. No outreach. No permission. No budget. Which is why it's usually the first thing worth fixing on an existing site.

• Descriptive anchor text. Link the words how retargeting audience windows work, not click here or this article.
• Link from your strongest pages. Whichever page attracts the most traffic and links is your most valuable internal source. Point it at the pages you're trying to lift.
• One hub per topic. A pillar page covering a subject broadly, linking out to the specific pages beneath it, each linking back. This makes the topical relationship explicit.
• No orphan pages. Every page should be reachable by clicking from somewhere else on the site. If the only route to a page is the sitemap, that's a weak signal that it matters.
• Link where it genuinely helps. A link that answers the question the reader has just formed is the one that gets clicked.

Technical SEO: the hygiene that makes you eligible

This is the section people over-invest in, because it feels like real work. A technically flawless page that misses search intent ranks nowhere. The checklist that matters:

• Set up Google Search Console on day one, before anything else. It's free. It's the only direct view into how Google sees your site. Every diagnosis in this article depends on it.
• Indexability. Check that robots.txt isn't blocking pages you want indexed, that live pages don't carry a noindex tag, and that canonical tags point where you intend. A staging-site noindex left in place after launch is a common failure and gives no visible symptom.
• Mobile. Google predominantly indexes the mobile version of a page. If content is hidden or broken on mobile, that's the version being judged.
• HTTPS. A trust issue before it's a search issue.
• XML sitemap, submitted in Search Console. A discovery aid, particularly useful on a new site with few external links.
• Reasonable load speed. Slow pages lose users, which is reason enough. As a ranking signal, page experience metrics are a small factor and easily outweighed by relevance. Don't expect speed work alone to move you up the results.

Do these once, properly, then stop. Time spent chasing a perfect Lighthouse score is better spent on one more page that finishes a job, or one genuine link.

Do backlinks still matter for a new site?

The industry oversells links, so be precise about the claim. Google's Gary Illyes has said publicly at industry events since 2023 that links aren't among the top ranking signals and that Google needs very few of them to rank a page. Google has softened the language about links in its own guidance.

That's true and it doesn't rescue a new domain. Because what a brand-new site lacks isn't links specifically. It's evidence of any kind. Nobody searches for your brand. Nothing you've published has been referenced anywhere. There's no history for the ranking systems to weigh. Links are the fastest form of evidence available to you, which is why they matter more at the start than they will once you have a track record.

What legitimately earns them for a small business:

• Original data. A small survey of your own customers, or a summary of pricing across your sector, is linkable in a way a general guide never is. Nobody links to "what is SEO". People link to numbers that don't exist elsewhere.
• A free tool or template people actually use. A calculator, a spreadsheet, a checklist. These accumulate links for years.
• Being quoted as a source. Journalist request services, trade press, industry publications. A practitioner with a specific opinion is a useful quote.
• Partner, supplier and membership pages. Suppliers list stockists. Trade bodies and chambers of commerce list members. Legitimate, and often already available to you.
• Local press and community. Sponsorships, events, local business coverage.
• Podcasts and guest contributions, where you're adding expertise rather than mass-pitching.

Source quality matters far more than count. One link from a respected publication in your sector does more than a long list of directory entries nobody visits. And a directory link farm isn't neutral. It's a risk. Ignore "Domain Authority" and similar scores while you're at it. They're third-party estimates published by SEO tool companies, not anything Google publishes, and buying links to move one is optimising for a scoreboard nobody official keeps.

Local SEO: the fastest unpaid win if you serve a place

If your customers come to you or you travel to them, this outranks everything above in priority order.

Local map results run on a partly separate system from standard web results, weighted towards relevance, distance and prominence. The asset is a claimed, complete Google Business Profile. Correct primary category. Accurate hours. Real photos. Service area. Description. Category choice does more work than most owners expect, because the primary category strongly shapes which searches you're eligible for at all.

Then reviews. Gathered steadily and genuinely. Ask every satisfied customer at the moment the work lands well. Reply to all of them, including the bad ones, in the way a future customer reading the reply would find reassuring.

Keep your name, address and phone number identical everywhere they appear. Your site. The profile. Directories. Social pages. Inconsistency creates ambiguity about whether these listings describe the same business.

For a plumber, clinic, restaurant or local agency, this is the shortest available path between effort and an enquiry. And unlike the content work, it doesn't depend on ranking a single page.

The reason it beats content in the early months is simple. A profile, some reviews, a category set correctly. That's a weekend of work. A ranked blog post is a quarter of work with a much wider band of outcomes. Do the fast, high-certainty thing first.

How long does SEO take to show results?

Treat what follows as a shape to plan against, not a forecast. Real timelines vary enormously with competition, existing authority and how well the intent match lands.

Days to a few weeks. Indexing. New pages on a healthy site get crawled and indexed quickly. Confirm it in Search Console rather than guessing.

One to three months. First impressions on long-tail queries. Impressions arrive before clicks. Positions will be volatile while Google works out where you belong, and that volatility is normal.

Three to six months. Long-tail pages that match intent well start holding positions and the first clicks arrive. This is where you learn which of your intent guesses were right.

Six to twelve months. Compounding traffic on a new domain, assuming consistent publishing and some links. That's an assumption, not a guarantee. It only holds if the earlier work was done properly.

Beyond twelve months. Competitive commercial terms, if ever.

New domains commonly show a lag before ranking for anything competitive. Google denies operating a formal "sandbox" that quarantines new sites. What's observable is that a domain with no history gives the ranking systems nothing to go on, and history takes time to accumulate. Either way the implication is the same. Budget in quarters.

The corollary matters more than the timeline. Because SEO is slow, the cost of targeting the wrong queries is six months, not six days. That's why checking intent before you write is worth more than any optimisation you do after publishing.

What not to do

Keyword stuffing. Repeating a phrase unnaturally, hiding text, stuffing footers with city names. It doesn't work, and it reads as spam to humans too.

Doorway pages. Twenty near-identical pages differing only by city or keyword, all funnelling to the same place. Explicitly against Google's spam policies.

Buying links, or trading them systematically. Paid links that pass ranking signals, excessive reciprocal exchanges and large-scale guest posting purely for links are link schemes under Google's spam policies. The risk is asymmetric. You can lose rankings you already had.

Mass-producing pages to fill a keyword list. Google's scaled content abuse policy targets content produced at scale primarily to manipulate search rankings, and it applies regardless of whether humans, automation or both produced it. Using AI to research, structure, draft or edit a page you then make genuinely useful isn't a violation, and Google has said so. Pointing a model at a spreadsheet of 500 keywords and publishing 500 pages nobody would choose to read is. The question the systems are trying to answer is whether the page exists for a reader or for a ranking. That question doesn't change based on what typed it. If you're using AI in your workflow, that's the line to hold.

The related trap is chasing "E-E-A-T" as a score. Experience, expertise, authoritativeness and trust are concepts in Google's quality rater guidelines. Descriptions of what the ranking systems try to approximate, not a number applied to your site. You can't optimise a score that doesn't exist. You can be the person who has actually done the thing, and show it.

The four numbers worth measuring

Search Console gives you all of them free.

1. Impressions. Are you eligible at all? Zero impressions for a query means you're not in the running, which is an intent or indexing problem rather than a ranking one.
2. Average position. Positions 11-20 with real impressions are the highest-value queue on the site. You're one improvement away from page one. Position 60 means start again.
3. Clicks and click-through rate. Impressions without clicks means the title and description aren't earning the click, or a featured snippet or AI summary above you is absorbing the answer. Rewriting a title gives you feedback within days.
4. Conversions. Traffic that never enquires, buys or subscribes is a hobby. Tie organic sessions to outcomes in analytics. The sheet that tells you whether marketing is making money is the same sheet here.

On the third point, zero-click behaviour is increasing as AI summaries appear on more queries. The direction is clear even though the widely-quoted percentages are contested. The practical response is to target queries where the searcher needs something a two-line summary can't supply. A tool. A quote. A decision. A specific document. Rank-tracker charts and "SEO health" percentages from audit tools are noise dressed as management information.

A worked example: is one long-tail page worth writing?

Every input below is a placeholder. Replace each with your own figure. The point is the structure of the calculation, not these numbers.

Say you're a B2B services business considering a page targeting how to calculate customer acquisition cost for saas, with roughly 200 searches a month. Assume you reach position 4 after eight months and take 7% of the clicks. Click-through rate by position varies enormously with the query and with how much sits above the organic results, so treat 7% as a placeholder rather than a benchmark.

• Monthly visits: 200 × 0.07 = 14
• Enquiries at a 3% rate: 14 × 0.03 = 0.42 a month
• Clients at a 25% close rate: 0.42 × 0.25 = 0.105 a month, or 1.26 a year
• At an average first contract of ₹1,50,000: 1.26 × 1,50,000 = ₹1,89,000 a year in revenue

The enquiry rate is the assumption you're least sure about, and it's carrying the entire result:

Enquiry rate	Enquiries/year	Clients/year	Revenue/year
1%	1.68	0.42	₹63,000
2%	3.36	0.84	₹1,26,000
3%	5.04	1.26	₹1,89,000

Two things this shows. Fourteen visitors a month looks like failure on a traffic dashboard and isn't, which is why traffic is the wrong headline metric. And a one-percentage-point error in the enquiry rate triples or thirds the answer, so the honest output is a range, not a figure.

Two things it doesn't show. ₹1,89,000 is revenue, not profit. Compare it against your gross margin and against what the page cost to produce. And the page doesn't compound. It earns roughly the same each year without further spend, which is recurring, not compounding. What compounds is the site. Links and topical track record make the next page easier to rank.

Scale it to ten pages of which four rank well: 4 × ₹1,89,000 = ₹7,56,000 a year at the top assumption, ₹2,52,000 at the bottom one. Six pages earned nothing. That hit rate is the thing to plan for, not the winners.

Now the comparison people expect. Buying those 14 clicks at ₹80 each costs ₹1,120 a month, ₹13,440 a year. On the figures above that looks like paid returning fourteen times its cost while the page is a waste of effort. And if you can genuinely buy those clicks at that price and they behave the same way, buy them. Two things stop that being the whole story. First, the 3% was assumed for someone who chose your result while researching a problem. There's no basis for assuming a bought click on the same query behaves identically, and the table shows how much weight that assumption bears. Second, a query with 200 monthly searches isn't a channel. Scaling paid means paying for each of the next fifty queries, every month, indefinitely. Scaling organic means writing fifty pages once and paying nothing per click afterwards. The real trade is a fixed upfront cost with an uncertain outcome against a variable cost with a predictable one. Most businesses should run both, which makes which paid platform to start with a separate decision.

A 90-day starting plan

Weeks 1-2, foundations. Set up and verify Search Console. Check indexability: robots.txt, stray noindex tags, canonicals. Submit a sitemap. Confirm the mobile version renders your actual content. If you're local, claim and complete the Google Business Profile now.

Weeks 3-4, research. Build a list of 20-30 long-tail, problem-shaped queries from autocomplete, People Also Ask, related searches, forums and your own inbox. For your top ten, search each and note the format that ranks. Cut anything where page one is entirely major authority sites.

Weeks 5-10, publish. Write five pages, one per cluster, each in the format the results told you to use.

Week 11, connect. Build a hub page and link it to all five. Link the five to each other where it helps the reader. Add links from your existing strongest pages. Remove orphans.

Week 12, links and review. Pursue three real link opportunities. A trade body listing. A supplier partner page. One contribution or quote. Then open Search Console and sort by impressions. Any page with impressions is already in the running and can be improved. Any page with none has an intent or indexing problem, and the site: check tells you which.

At day 90 you won't have meaningful traffic. You'll have data telling you which of your guesses about intent were right, which is the actual deliverable of a first quarter. Quarter two is where you double down on those.

FAQs

How long does SEO take to work for a new website?

Plan in quarters. Indexing takes days to weeks, first impressions on long-tail queries usually one to three months, and traffic worth reporting somewhere between six and twelve months of consistent work. Competitive commercial terms take years or never arrive. The useful checkpoint at month three is impressions, not clicks. Impressions prove you're eligible, which is the first thing to establish.

Can I do SEO myself without hiring an agency?

Yes, for the fundamentals. Search Console setup, keyword research from free sources, intent matching, writing pages that finish the job and internal linking are all doable by one person who knows the subject. Outside help earns its fee on link acquisition, technical problems on large sites, and diagnosing why a site that used to rank has dropped.

How many backlinks do I need to rank on Google?

There's no number, and chasing one leads to buying them. For long-tail, low-competition queries you may rank with almost none. For competitive commercial terms you need a genuine authority position built over years. A more useful question than "how many" is "which sites in my sector could plausibly have a reason to reference me", because that question has actions attached.

Does Google penalise AI-written content?

Not for being AI-written. Google's policy here is about mass production for rankings, not about authorship, and it applies to human-written page farms too. Drafting and editing a page that genuinely helps someone is fine. The practical test to apply yourself. Would a reader who landed on this page by accident still have found it worth reading?

Why is my website not showing up on Google search?

Search site:yourdomain.com. If nothing appears, it's a crawl or index problem. Check robots.txt. Look for stray noindex tags. Confirm the pages are linked from somewhere on the site. Submit a sitemap. If pages do appear but attract no traffic, you're indexed and simply not ranking, which is an intent-match and authority problem instead.

Is SEO still worth it now that Google shows AI answers?

For queries a two-line summary fully answers, the click was always low value. What changes is the mix. Definitional and trivia queries lose clicks, while queries needing a tool, a quote, a local supplier or a decision still send people to pages. Shift your targets accordingly and judge the channel on enquiries, not sessions.

Key takeaways

• Rankings turn on whether the page matches what the searcher wanted and whether there's any reason to prefer you over the pages already there. Tags, speed and schema only make you eligible to be considered.
• Check site:yourdomain.com before doing anything else, because a page that isn't indexed and a page that's indexed but not ranking are different problems with no shared fix.
• A site with no history can't outrank established authority sites on broad head terms, so the winnable strategy is long, specific, problem-shaped queries where few pages answer properly.
• Search the query and match the format already ranking. A service page will never beat guides for a "how to" query, no matter how well it's written.
• A page is thin when it doesn't finish the job, which has nothing to do with word count and everything to do with whether the reader can act when they reach the end.
• AI is a legitimate drafting tool and an illegitimate page-printing press, because what Google's spam policies act on is production at scale for rankings, not the identity of whoever or whatever typed the words.
• Judge the channel on impressions, average position, clicks and conversions from Search Console, and expect the first quarter to produce data rather than revenue.

Related reading: What Is a Marketing Funnel? A Simple Guide With Real Examples · How to Build an Email List From Zero · Marketing Metrics Explained: CPC, CPM, CTR, CPA and ROAS in Plain English