Keyword Research

Most people start a page with a topic they personally find interesting. That is exactly backwards. A page does not succeed because you found the subject interesting. It succeeds because a real person, at a real moment, typed a specific string of words into a box, and your page turned out to be the best answer to those exact words. Keyword research is how you find out what those words are before you write a single line, instead of after, when it is too late to matter.

The three numbers every keyword carries

Once you have a candidate keyword, a tool like Ahrefs hands you three numbers that matter more than all the rest. Learn to read them together, because any one of them on its own will lie to you.

Volume is how many times a month people search that phrase. It is the size of the audience, and it is the most seductive and most dangerous number, because big volume pulls everyone toward the same handful of impossible words.

Difficulty, usually shown as KD on a 100 point scale, estimates how hard it is to break into the top ten for that phrase. Really it is a proxy for how much authority and how many links the current winners already have.

Intent is the quiet one, and the most important of the three. It is what the person actually wants. Two people can search phrases that look almost identical and want completely different things, and if you answer the wrong want, none of the other numbers can save you.

Here is what that looks like with real US search data for the phrases in this very chapter. Read it the way a practitioner does, not a beginner.

Keyword	US volume	KD	What it actually tells you
seo	629,000	93	Enormous demand, and a wall. Only a handful of the most authoritative sites on earth rank here. A new site targeting this is bringing a spoon to a gunfight.
keyword research	604,000	93	Same story. Huge audience, brutal difficulty. The phrase this page is named after is one it has almost no chance of ranking for on day one.
technical seo	13,000	70	Smaller, still hard, but a real target once you have built some authority.
content optimization	1,900	29	Modest volume, very winnable. This is where a young site actually starts ranking.

The beginner sees 604,000 and chases keyword research. The practitioner sees KD 93 and understands you do not start there. You start with content optimization at KD 29, and the dozens of quieter, easier phrases around it, and you earn the right to compete for the big words later. Volume tells you the size of the prize. Difficulty tells you the odds. You place your bets only where the two make sense together.

Intent, the number that decides everything

Come back to intent, because it is where most keyword research quietly fails. Search intent is the reason behind the search, and it comes in four broad shapes.

Informational: they want to learn. "What is technical seo." They are not ready to buy anything. They want an answer.
Commercial: they are comparing before a decision. "Best seo tools," "ahrefs vs semrush." Wallet is out, mind is not made up.
Transactional: they want to act now. "Buy ahrefs subscription," "seo audit service near me."
Navigational: they already know where they are going. "Ahrefs login." You will not win these unless it is your own brand.

This matters more than volume, and here is why. If someone searches best seo tools, they want a comparison, a list, honest tradeoffs. If you meet that search with a hard sell for your one product, you lose, no matter how good the page is, because you answered a question they never asked. So before you commit to any keyword, do the thing almost nobody does: open a private window, search it yourself, and study what already ranks. The results are Google telling you, out loud, what it believes the searcher wants. Match that shape, or leave the keyword alone.

How to actually do it

Stripped to the bone, the process is five moves.

Start with seeds. A handful of obvious words for what you do. Selling running shoes: running shoes, marathon training, best shoes for flat feet.

Expand them. Feed each seed into a keyword tool and let it surface the hundreds of real phrases people search around it. Then go where the tools do not reach: type your seed into Google and read the autocomplete, scroll down to People Also Ask, look at the related searches at the bottom. Every one of those is a real question from a real person, handed to you for free.

Mine your own data. If your site already gets any traffic at all, Google Search Console is the most honest keyword tool you own. It shows the exact phrases you already appear for, including ones you never targeted and are sitting on page two of. Those near misses are the fastest wins you will ever find.

Study who is winning. Drop a competitor's page into a tool and pull the keywords it ranks for. They have already paid for expensive research. Read their homework.

Group and prioritise. You now have a messy pile of hundreds of phrases. Cluster them into topics, and score each cluster on the only question that matters: is the volume worth it, can we realistically rank given the difficulty, and does it bring us the right person. High relevance at winnable difficulty beats high volume almost every time.

One page, many keywords

A trap worth naming, because it wastes more effort than almost anything else in SEO. Beginners make one page per keyword. They write "best running shoes," then "top running shoes," then "good running shoes" as three separate pages, and then wonder why none of them rank.

Google stopped matching strings a long time ago. It understands that best running shoes, top running shoes and good running shoes are the same want wearing slightly different clothes. It wants to rank one excellent page for all of them. Tools show you this as the parent topic: the broader keyword that the current number one page for your phrase actually ranks under. If your keyword's parent topic is different from the keyword itself, that is the tool quietly telling you to target the parent with one strong page and let it collect the variations underneath. One page, many keywords. Always.

Where new sites win

If all of this makes it sound like the big words are locked up by big sites, that is because they mostly are. The way in has always been the long tail: longer, more specific, lower volume phrases that the giants cannot be bothered to chase.

"Seo" is a wall. "How to do keyword research for a small bakery" is a door. Far fewer people search it, but the ones who do want exactly what that page offers, the competition is thin, and you can genuinely rank while your site is still young. Win a hundred of those small, specific searches and you have both real traffic and the authority to start reaching for the bigger words. Nobody starts at the summit. They start at the doors nobody else bothered to knock on.

Read the results before you commit

A keyword on a spreadsheet is a guess. A keyword you have actually searched is a decision. Before you commit to targeting anything, run it through Google yourself and read the top ten, because the results are Google showing you, for free, exactly what it thinks that search deserves.

Look at three things. The format that ranks tells you what to make: if all of page one is listicles, a listicle is the price of entry. The strength of the sites tells you whether you can win: ten giants means wait, a couple of ordinary pages means go. And the extra features tell you the real intent: a People Also Ask box is a ready-made list of the follow-up questions to answer on your page, a shopping row means the search is commercial, a featured snippet is a definition or step list you can try to take with a cleaner version. Ten minutes reading a real search results page will save you from writing ten pages aimed at the wrong thing.

Finding questions and modifiers systematically

The fastest way to turn one seed into a hundred real phrases is to attack it with modifiers. Each modifier is a different moment in a person's life, and each one is a page you could own.

Add question words: how, what, why, when, which. "How to clean running shoes." Add comparison words: vs, or, best, top, alternatives. "Nike vs Adidas running shoes." Add qualifiers: cheap, best, near me, for beginners, under 5000. "Best running shoes for beginners." Add a year when freshness matters. Free tools like Google autocomplete and the People Also Ask box are modifier machines: type your seed, add a single letter, and a dozen real searches appear. You will never run out of specific, winnable phrases once you learn to expand a seed this way.

The mistakes that quietly kill it

A handful of errors sink more keyword research than any missing tool ever could.

Falling for volume. The biggest number is almost always the hardest and the vaguest.
Ignoring intent. Ranking for a keyword whose searchers want something you do not sell brings traffic that bounces and never converts.
One keyword per page. Splitting variations of the same want across separate thin pages, and quietly cannibalising your own rankings.
Never revisiting. Treating research as a one-time event instead of a living list that your own Search Console keeps feeding you. Get these four right and you are already ahead of most sites chasing the same words.

What AI search changes

One shift you cannot ignore, because it is quietly rewriting this whole discipline. People are starting to search in full questions and sentences, spoken to an assistant or typed into ChatGPT, not in the clipped two word phrases we trained ourselves to target. "Best budget running shoes for flat feet under five thousand rupees" is now a single search, not four separate ones.

So keyword research is becoming question research. Alongside your keyword list, build a question list: every real, specific question a person might ask an AI about your topic. Those questions are the phrasing the answer engines are built to match, and the page that answers a genuine question cleanly, in plain words, is the page that gets quoted back in the answer. The words are getting longer and more human. Your research simply has to follow them there.