Keyword Research

Keyword research is the work of finding out what your audience actually types into search engines — and how often, and why — before you build content around it. Skip it and you end up writing for phrases nobody searches, or worse, for phrases whose searchers want something entirely different from what your page sells.

What the job actually is

The deliverable is not a spreadsheet of 2,000 keywords sorted by volume. It is a set of decisions: which topics get their own page, which cluster together, what language the titles and headings use, and — critically — what kind of page each query demands. A term like "crm migration checklist" wants a checklist, not your product tour, and no amount of on-page polish fixes a format mismatch. That judgment call belongs to search intent analysis, which is inseparable from keyword research done properly.

Volume, difficulty, and intent are the three dials. Volume tells you the ceiling. Difficulty tells you the price of admission — a 5,000/month term owned by ten DR80 sites is a fantasy for a new domain, while a cluster of specific 50/month phrases is winnable this quarter. That trade-off is the core of long-tail keyword strategy.

Where the numbers come from — and how each source lies to you

Every keyword data source has a bias baked into how it collects numbers. None of them is "the truth"; the skill is knowing which distortion you are looking at.

The honest workflow triangulates: a tool for scale, GSC for ground truth on your own rankings, and the SERP itself for intent. Any single source will steer you wrong eventually.

How to check it on your own site

1. Open Google Search Console → Performance → Queries, filter to positions 5–20, and export. These are terms Google already half-believes you deserve — the cheapest wins available.
2. Pick your ten most important pages and pull their ranking keywords in Ahrefs or Semrush (Site Explorer → Organic keywords). Note where a page ranks for terms it doesn't actually address well — that is unmined demand.
3. Run your main competitor through the same tool and use the content-gap report: keywords they rank for and you don't. Ignore their branded terms.
4. Take your top 5 target terms and actually search them. Read the full first page. If the results are all listicles and you have a product page targeting the term, you have found an intent mismatch, not a keyword opportunity.
5. Map every chosen keyword to exactly one URL before writing anything, so two of your own pages never compete for the same term — the process is laid out in keyword mapping.

Common audit mistakes

• Trusting tool volumes as measurements. They are estimates from models. Use them to rank opportunities against each other, not as traffic forecasts to promise anyone.
• Filtering out everything under 100 searches/month. That filter deletes the exact terms where new and mid-sized sites actually win, and a cluster of specific low-volume terms often outproduces one head term you will never rank for.
• Chasing volume with zero difficulty check. The result is a content calendar full of terms you cannot touch for years. Weigh every opportunity against what your domain can realistically rank for today.
• Doing the research once and framing it forever. Query language shifts — new modifiers, new competitors, new SERP features eating clicks. Re-run the analysis on your money topics at least yearly.
• Never reading the actual SERP. The results page is the only place where volume, intent, and competition are visible simultaneously, and it is free. Tools supplement it; they don't replace it.

FAQ

Which keyword tool is most accurate?

None of them, and anyone claiming otherwise is selling one. Each models demand differently and they routinely disagree. Pick one for consistency, use GSC for truth about your own site, and compare terms relatively rather than trusting absolute numbers.

Is keyword research still worth doing in the AI-search era?

Yes — the format is shifting toward conversational queries and AI answers, but the underlying discipline (what does the audience want, in their words, and can you win it?) transfers directly. What changes is covered in how AI changes keyword research strategy.

How many keywords should one page target?

One primary term plus its natural variants and subquestions. A thorough page picks up dozens of related rankings on its own; deliberately stuffing five unrelated primaries into one URL just produces a page that ranks for none of them.

What is a good keyword difficulty score to target?

There is no universal threshold — a KD30 term is easy for an established site and hopeless for a three-month-old domain. Calibrate against yourself: check the difficulty of terms you already rank for, and hunt slightly above that line.

Where should a complete beginner start?

With the step-by-step process in keyword research fundamentals — seed terms, expansion, intent sorting, and mapping, in order.