Should I choose keywords by search volume?

Volume is a tiebreaker, not the test. A five-hundred-search query that matches buying intent beats a fifty-thousand-search query answered by an encyclopedia page.

What if my site is new with no authority?

Choose queries where the current top results are weak, thin or off-intent, and where you can add first-hand detail. Feasibility outranks volume for the first year.

Do long-tail keywords still work?

Yes, and they are usually the only realistic entry point for a new site. They also reveal the exact phrasing buyers use, which improves the page that later targets the shorter term.

How to Choose Keywords for Your Website: Four Checks

WebSEOGrowKeyword to page Send three candidate keywords

Keyword pick · four checks

How to Choose Keywords for Your Website Before You Write a Word

Choosing keywords means passing each candidate through four tests: does the searcher want what you sell, is the query worth money, can your site realistically compete, and do you already own assets that help. Fail one test and the page becomes a cost. Pass all four, score them, then build one page per keyword.

A shortlist you can defend beats a spreadsheet of nine thousand rows.

Score before you brief. A keyword that fails feasibility can sit in the backlog for a year and still fail.

What is the fastest reliable way to choose keywords?

Start from what you sell, not from a volume tool. List the phrases a buyer would type at the moment they are ready to compare suppliers, test each one for intent, commercial value and feasibility, and assign a single page to each. Volume comes last, because a small query with buying intent pays and a large query with none does not.

The order is not negotiable

Most sites choose keywords after the article is written, then edit the headline until it fits something. That produces pages with no demand behind them and a content calendar nobody can justify. Reverse it: pick the query, decide the page type, then write.

Wrong order

Write a post, look for a keyword it could rank for, add it to the title, wonder why impressions never arrive.

Right order

Pick a query with intent and feasible competition, decide whether it needs a service page, a comparison or a guide, then write to that shape.

Why it matters

A page built for a query answers it in the first screen. A page with a query bolted on answers something else and gets ignored by both readers and search engines.

If you cannot name the query a page exists to answer, you do not have a content plan. You have a posting habit.

Four checks, in the order that saves the most time

Run them in this sequence. Intent first, because a mismatch cannot be fixed by better writing. Feasibility third, because it decides whether the work is worth starting this quarter.

01Intent

Does the searcher want what you sell?

Check the result type before the words

Search the query yourself and read what ranks. If the first page is all definitions and encyclopedia entries, the searcher wants an explanation, and a product page will not satisfy them no matter how well it is optimised. If the page is full of supplier catalogs, price lists and comparison tables, the searcher is shopping.

Then read the query wording. "How to choose a stainless bottle" is research. "Stainless bottle supplier" is procurement. "Stainless bottle MOQ" is a buyer already in conversation with someone. Each wants a different page, and the difference is visible in the phrasing before you open any tool.

Fail condition: you would need to write a different page type to answer it honestly.

02Value

Is the query worth money to you?

Trace the query to a transaction

Ask what happens after the click. If the honest answer is "they read and leave", the query is a traffic asset at best. If the answer is "they request a sample, a quote or a spec sheet", it belongs at the top of the list even at a fraction of the volume.

A practical test: write the next step down. For "stainless bottle supplier" the next step is a quote request. For "what is 304 stainless steel" the next step is another article. Both can be worth writing, but only one pays this quarter, and the plan should say which is which.

Fail condition: no path from the query to an enquiry, an order or a booked call.

03Feasibility

Can this site compete within a year?

Judge the results, not the difficulty score

Look at what currently ranks and ask whether you can produce something clearly better. Weak signals: thin pages, no first-hand detail, no images of the actual product, articles that answer a slightly different question. Strong signals: manufacturers publishing specs, marketplaces with deep category structure, sites with hundreds of pages on the topic.

A new domain should start where the ranking pages are beatable. That usually means longer, more specific phrasing plus an asset nobody else has, such as test data, a size chart, a shipping table or real installation photos.

Fail condition: every result on page one is a marketplace or a manufacturer with a dedicated category.

04Assets

Do you already own something that helps?

Look for existing pages, data and links

The cheapest keyword is one your site already half-answers. A page with impressions but no clicks is a rewritten title away from working. An old post with three external links is a better starting point than a blank draft with none.

Inventory what you have: pages receiving impressions for adjacent phrases, product data you can publish, photos, certificates, FAQs from sales calls. Those become the evidence a competing page cannot copy, which is what makes the fourth check worth more than a difficulty score.

Fail condition: the answer is "nothing yet, we would start from zero" on a competitive query.

Score the shortlist from one to five

Four checks, four scores, one total out of twenty. The point is not precision. It is forcing every candidate onto the same scale so that the argument in the meeting is about the numbers rather than about who likes which idea.

Scores are relative to your site, not to the market. A five for feasibility means beatable by you, not easy in absolute terms.
Candidate query	Intent	Value	Feasibility	Assets	Total
stainless bottle supplier	5	5	3	4	17
custom logo water bottle bulk	5	4	4	3	16
bottle mockup template free	2	1	2	1	6
what is 304 stainless steel	3	2	4	5	14
bpa free bottle manufacturer	5	4	4	2	15

Read the totals as a queue, not a verdict. Anything at sixteen or above gets a page this quarter. Twelve to fifteen goes to the backlog with the missing score named, so the fix is obvious: build the asset, collect the data, or wait until the site has more depth. Below twelve, write it down and move on, because a low score now rarely becomes a high score later without a new asset.

The one exception is a query scoring five on value with a two on feasibility. Do not delete it. Keep it visible in the plan and revisit it once the surrounding cluster has pages, since feasibility is the score that changes most as a site grows.

Have a list but no ranking preference?

Send up to twenty candidate phrases. We will score them on the same four checks and return the order we would build in.

One keyword, one page

Once a query scores well, it needs an owner. Not a section of a page, not a paragraph on the homepage, and never two pages in the same cluster. The mapping below is deliberately small: three queries, three pages, no overlap.

Query

stainless bottle supplierbuy

custom logo water bottle bulkbuy

what is 304 stainless steellearn

Page type

Category page17

Service page16

Guide article14

Job on the page

Specs, MOQ, quote formsell

Logo process, samplessell

Grades, then link to catalogroute

Two rules keep the map honest. First, every page has one primary query written next to it in the brief, so a writer cannot drift. Second, if a new query fits an existing page, it becomes an H2 or a supporting phrase on that page, not a new URL. Most keyword sets collapse by a third once this rule is applied, which is a good outcome: fewer pages, each with more reasons to rank.

When the map is done, the on-page work becomes mechanical: the primary query goes in the title, the H1 and the first hundred words; supporting phrases become headings and body references; the internal links point at the next step in the journey rather than at the homepage. That handoff is exactly what Keyword Research Services covers, and it is the reason we refuse to write content before the map exists.

Where the four checks stop being enough

On a small site, the checks above are the whole job. At a few thousand URLs the problem changes: you are choosing between clusters rather than keywords, and the scoring has to account for cannibalisation across hundreds of pages that already exist.

That is the point where a structured research pass earns its fee. You need a full inventory of pages and queries, a mapping exercise to find where two pages compete for the same intent, and a decision about which URL keeps the query and which gets rewritten or merged. Doing that by hand in a spreadsheet works up to a few hundred rows and then quietly stops working, because the overlap is invisible until you sort it properly.

Our On-Page SEO Services run exactly these four checks at cluster level, then deliver the one-to-one page map plus the merge list for pages already competing with each other. You keep the scoring sheet, so next quarter's review starts from a document rather than from scratch.

Questions we get about choosing keywords

Five answers, in the words we use on calls.

How many keywords should one page target?

One primary intent per page, supported by five to fifteen related phrases. Two pages chasing the same primary query compete with each other, and the usual result is that both sit below where either could have ranked alone.

Should volume come first?

No. Volume is a tiebreaker between two viable candidates. A query with a few hundred monthly searches and clear buying intent beats a broad informational query by an order of magnitude on return.

What if my site is brand new?

Choose queries where the ranking pages are thin, outdated or answering a slightly different question, and where you hold an asset such as real product data or installation photos. Feasibility outranks volume for the first year.

Do long-tail phrases still work?

Yes, and they are usually the only realistic entry point for a new domain. They also reveal the exact wording buyers use, which later improves the category page that targets the shorter phrase.

How often should the list be reviewed?

Quarterly. Intent shifts, competitors publish and seasonality moves volumes. A list older than six months describes the past rather than the plan.