The basics of keyword research come down to three questions asked in order: what do people search when they want what I offer, which of those terms can I realistically rank for, and what page does each term deserve? Everything else — the metrics, the tools, the exports — exists to answer those three. Get the basics of keyword research right and the rest of an SEO program has something solid to stand on; get them wrong and you will publish diligently into an empty room.
This is the beginner’s path, in the order you should actually work.
Start With Seeds, Not a Tool
A seed keyword is a short phrase describing something you sell, solve, or teach. You need eight to fifteen of them before opening any software, and the best sources are not keyword tools at all.
- Your customers’ own words. Support tickets, sales call notes, and review text. Customers rarely use your internal vocabulary, and their phrasing is what they type into search.
- Your navigation and product names. Category names are usually seeds already.
- Competitor top pages. Whatever carries their organic traffic is a proven seed.
- The problems you solve. People often search the symptom, not the solution category.
- Search autocomplete and related searches. Free, live, and drawn from real queries.
Write them plainly. “Project management software,” “how to track team tasks,” “gantt chart template.” A mix of what you sell and what people search on the way to needing it.
Expand the Seeds Into a Candidate List
Feed the seeds into a keyword tool to get the full universe of related terms. A multi-seed explorer that accepts several seeds at once and returns up to 150 ideas per search saves considerable time over running them one at a time — you get the whole candidate pool in a couple of passes rather than fifteen.
Set the country index before you look at anything. This is where beginners most often go wrong: a business serving the UK, Australia, or Singapore queries the default US index, sees tiny numbers, and draws the wrong conclusion about demand. Search behavior and volume differ by market, and so does who ranks.
Expect a few hundred to a few thousand rows. That is normal, and most of it is not for you. The next steps are all about cutting.
The Four Metrics That Matter
Ignore the rest until these four make sense.
- Search volume — the estimated number of monthly searches. It is a modeled average, usually across roughly twelve months, from periodic crawls. Read it as a band rather than a fact: a term shown at 1,000 could reasonably be 600 to 1,600, and a seasonal term’s average hides its actual peak entirely.
- Keyword difficulty — a 0–100 estimate of how hard ranking will be, driven mostly by the links pointing at the current top results. Useful for sorting a big list, not for final decisions.
- Cost per click — what advertisers pay for that click. The best cheap proxy for commercial intent. High CPC on modest volume usually beats low CPC on big volume.
- SERP features — whether the results page carries an AI Overview, featured snippet, local pack, or shopping unit. Features push organic links down and reduce clicks, so a 5,000-volume term can deliver fewer visits than a 900-volume term with ten clean results.
All third-party figures are estimates. Once your pages are live, Google Search Console shows what your site actually received, and Google’s own data wins every disagreement about your own performance.
Intent: The Filter That Saves the Most Wasted Work
Search intent is the goal behind the query, and matching it is more predictive of ranking than any metric on the list. Four rough categories cover almost everything.
- Informational — “what is rank tracking.” Wants an explanation. Ranks: guides, tutorials.
- Commercial — “best rank tracker,” “tool X vs tool Y.” Comparing before buying. Ranks: comparisons, roundups.
- Transactional — “rank tracker pricing,” “buy running shoes.” Ready to act. Ranks: product and pricing pages.
- Navigational — a brand name. Someone already knows where they are going.
Never guess intent from the words alone. Open the results page and look. If the top ten are all product pages, a blog post will not rank there no matter how good it is — Google has already shown you what it thinks satisfies that query. Two minutes of looking prevents weeks of writing the wrong format.
Judging What You Can Actually Win
Difficulty scores describe an average of the whole page one. Rankings are won at the soft spot. So for each shortlisted term, find the weakest result in the top ten and study it: how strong is that domain, how many sites link to that specific page, how long and how good is it, when was it last updated.
If the weakest page-one result is a 700-word piece on a site no stronger than yours with a handful of links, that keyword is a realistic target this quarter — whatever the difficulty score says. If the weakest result is a thorough guide on a much stronger domain, put the term on a twelve-month list and pick something else now. Benchmarking against the weakest competitor rather than the leader is the difference between a plan you execute and a plan that intimidates you into inaction.
From List to Plan: One Primary Keyword Per Page
A keyword list becomes a plan when every term has a home. Group your survivors into clusters — sets of terms that share one underlying intent — and give each cluster one page.
- Cluster by intent, not by word overlap. If the same URLs rank for several variants, those variants belong to one page.
- Assign one primary keyword per page. That term sets the title, H1, URL, and opening sentence.
- List three to eight secondary terms under each primary. They live in H2s and body copy and usually deliver most of the traffic.
- Keep a keyword-to-URL map. One row per primary keyword. Check it before commissioning anything, or you will eventually build two pages for the same query and split your own rankings.
- Sequence by winnability. Publish the beatable cluster first, establish footing, then attack head terms from a stronger position.
Saving your shortlist to a project keyword pool rather than a loose spreadsheet is worth the small effort — in SEO Rocket that pool feeds both the writer and the rank tracker, so the terms you validated are the terms you write about and then measure. Filtering, sorting, and CSV export are available free after your first query if you would rather work in a spreadsheet.
Measuring, and Redoing It
Give new pages six to twelve weeks before judging them. Track your primary keywords weekly, on a fixed list, from a fixed location, and read the trend rather than any single reading — daily movement of two or three positions is ordinary noise, not a signal.
After a couple of months, Search Console becomes your best research tool. It shows the queries your pages actually earned impressions for, which frequently includes terms you never targeted and reveals that your intended primary keyword was not the phrase people used. Feed those discoveries back into the list. Keyword research is not a project you finish; it is a loop you run every quarter, and each pass is better informed than the last because it is built on your own data instead of estimates.
