Keyword research is the foundation of every effective SEO strategy. Understanding what your audience searches for — and which queries you can realistically rank for — determines which content to create and how to structure it.
Good keyword targets have three properties: enough search volume to be worth the effort, realistic difficulty given your current authority, and clear commercial or informational intent that matches what you offer.
Run the Keyword Research Tool on your existing pages. It shows which keywords each page currently ranks for and identifies quick wins — keywords where you rank positions 5-15 that could reach the top 3 with improvements to the page.
Run the keyword tool on your top 3 competitors. The keywords they rank for that you do not are your priority targets. Focus on those with search volume over 100 and difficulty under 40 for realistic wins.
Include your primary keyword in: the title tag (near the start), the H1 heading, the first paragraph, at least one H2 subheading, the meta description, and the URL slug. Do not force it — write naturally and the keyword will appear in context. Keyword stuffing hurts rankings.
Keyword research goes wrong at the very first step, because the figures people build strategies on are presented with a precision they do not possess.
Nobody outside Google knows how many times a term was searched. Every tool derives its figure from a mixture of Google's own Keyword Planner data — which is bucketed and heavily rounded, and which groups close variants together — and clickstream panels extrapolated to the whole population. Run the same keyword through three tools and you will routinely get three materially different numbers.
The practical rules that follow:
Difficulty scores are not Google metrics. They are computed almost entirely from the backlink profiles of the current top ten, which means they systematically miss the things that actually decide whether you can win:
Third-party tools tell you what they estimate about the market. Search Console tells you the truth about you — real impressions, real positions, real clicks, from real searches on your actual pages. Almost nobody mines it properly.
Filter the Performance report to queries where your average position is between 8 and 20. These are pages Google already considers relevant enough to show — they are simply not quite good enough to be seen. Moving a term from 14 to 6 is a fraction of the work of ranking a new page from nothing, and the traffic difference is enormous, because clicks collapse steeply below the first few positions.
Sort by impressions and look for queries where you are ranking respectably and almost nobody clicks. That is not a ranking problem — it is a snippet problem. Your title and description are failing to earn the click that your position has already won you. It is the cheapest fix in SEO: rewrite the title, change nothing else.
Every page ranks for far more terms than it was written for — often hundreds. Reading them tells you the language your actual audience uses, which is reliably different from the language you assumed. Frequently the highest-impression query on a page is not the keyword you targeted, and the page can be improved simply by acknowledging the question people were really asking.
Searches for your company name are not growth; they are people who already knew about you, arriving. Filter them out. What is left — searches for what you do — is the number that tells you whether the SEO is working.
Google has already determined what a query means. That determination is visible, for free, in the results — and it is not negotiable. Publishing the wrong content type against a settled SERP is the most reliable way to write something nobody will ever see.
| Intent | The SERP looks like | You must publish |
|---|---|---|
| Informational | Guides, definitions, PAA boxes, videos | A guide. A product page will not rank. |
| Commercial investigation | "Best X", comparisons, reviews, listicles | A comparison. Your own product page is the wrong format. |
| Transactional | Product pages, shopping carousel, ads | A product or category page. A blog post will not rank. |
| Navigational | One brand dominating everything | Nothing. Do not target someone else's brand. |
| Local | Map pack above the fold | A Business Profile. The web page is secondary. |
Two consequences that people resist:
The single most common structural error in keyword strategy is building a page for every keyword variant. Google does not work that way, and has not for years.
Modern search matches meaning, not strings. "Cheap plumber London", "affordable plumber in London" and "low cost plumbers London" are the same question, and one good page will rank for all three and for hundreds of variants you never thought of. Building three pages does not triple your coverage — it splits it.
When several of your pages target the same intent, Google has to choose one. It chooses on its own criteria, and it frequently chooses the wrong one — a stale blog post rather than your service page. Meanwhile the links, the internal signals and the engagement that should have accumulated on one strong page are split across three weak ones, and none of them is strong enough to win.
The symptom is diagnostic and easy to spot in Search Console: the same query returning different URLs of yours over time, and none of them ranking well.
There is no target percentage. Google does not count how many times a term appears and compare it to a threshold, and it has not done so for a very long time. Tools that report "your keyword density is 0.8%, aim for 2%" are measuring something that is not being assessed.
What replaced it is entity and concept coverage. Google evaluates whether a page demonstrates understanding of a topic — which means covering the things a knowledgeable person would inevitably cover. A genuine page about Core Web Vitals mentions LCP, CLS, INP, field data and the 75th percentile, not because someone stuffed those terms in, but because you cannot write competently about the subject without them.
This flips the writing process. Rather than asking "have I used the keyword enough times", ask "what would an expert inevitably mention that I have not?" Then write about that. The terms take care of themselves.
Repetition, meanwhile, has an actual cost. A page that says "emergency plumber London" eleven times reads like a machine wrote it, which damages the engagement signals that are assessed — and at sufficient volume it is a spam pattern in its own right.
Keyword tools return the terms that already have data. They cannot show you the question somebody asked yesterday for the first time, and they systematically miss how people actually phrase things.
And a shift worth planning for: queries are getting longer. As people become used to asking assistants full questions rather than typing three words at a search box, the phrasing moves towards natural language. Content organised as clear questions with direct answers stated plainly near the top is what a retrieval system can extract and cite. That is the same discipline that answers a PAA box — it just matters in more places now.
Keyword difficulty is a score (usually 0-100) that estimates how hard it is to rank on page 1 of Google for a given keyword. It is calculated based on the strength of the pages currently ranking — their domain authority, backlink profiles and content quality. A low difficulty score means the current top results are weak and can be displaced with good content. A high difficulty score means the top results are from authoritative sites that are hard to outrank.
Long-tail keywords are longer, more specific search phrases that have lower search volume but also lower competition and higher conversion intent. 'SEO' is a short-tail keyword with enormous volume and extreme difficulty. 'How to fix slow page speed on Shopify' is a long-tail keyword with modest volume but low competition and high intent. For most sites, long-tail keywords are where rankings are achievable.
Focus each page on one primary keyword and 2-5 closely related secondary keywords. Do not try to rank one page for many unrelated keywords — Google ranks pages for specific topics and writing for too many keywords dilutes your topical focus. Create separate pages for separate topics.
They are estimates, not measurements. Tools derive them from Google's Keyword Planner data, which is bucketed and heavily rounded and groups close variants together, blended with clickstream panels extrapolated to the wider population. The same keyword will return materially different figures in different tools. Use volume to rank opportunities against each other, never as a traffic forecast — and check the seasonal trend rather than the average, which hides it completely.
No. Google matches meaning rather than exact strings, so one strong page will rank for hundreds of variants of the same question. Building a page per variant splits your links, internal signals and engagement across several weak pages instead of concentrating them in one strong one, and creates cannibalisation — where Google has to pick between your own pages and frequently picks the wrong one. Build one page per intent.
There is no target density and Google does not assess one. What matters is whether the page demonstrates genuine coverage of the topic — the concepts and entities a knowledgeable writer would inevitably mention. Repetition has a real cost: a page that repeats its keyword eleven times reads as machine-written and damages the engagement signals that are actually assessed.
Often, yes. Zero frequently means "below the tool's reporting threshold" rather than "nobody searches this". Brand-new terms and highly specific commercial queries routinely report zero volume and still convert, because the small number of people typing them are the ones closest to buying.
Run the Keyword Research Tool and get actionable results in minutes. Pay as you go.
Research Keywords →aiwebpageseo.com is a data-driven SEO and AEO (Answer Engine Optimisation) platform providing a free suite of technical website tools. Rather than relying on AI-theorised assumptions, the platform analyses live URL performance, delivering objective diagnostics, page speed metrics, CLS debugging, and site crawl data alongside actionable technical tutorials.