An SEO keyword suggestion is a related search term that a tool proposes after you give it a seed word, a URL, or a topic. That is the whole definition. A suggestion is not a recommendation, not a promise of traffic, and not evidence that anyone at Google has ever seen that phrase typed more than a handful of times. It is a candidate. Your job is to decide which candidates deserve a page.
Most people get this backwards. They export 3,000 suggestions, sort by volume, and start writing at the top. Six months later they have twenty pages stuck on page four, competing against sites with ten years of link equity. The suggestions were fine. The selection was the problem.
Where keyword suggestions actually come from
Suggestion engines pull from four distinct sources, and knowing which one you are looking at changes how much you should trust the number beside it.
- Autocomplete and related searches. Scraped from Google’s own suggestion API and the “People also search for” block. High signal that real humans type these, but usually no volume attached.
- Clickstream and modeled volume. Third-party panels plus modeling produce the monthly search volume you see in most tools. These are estimates — typically a modeled average over roughly the last twelve months, not a live counter.
- Competitor keyword sets. The tool crawls SERPs, records which URLs rank for which terms, and hands you every keyword a rival page already ranks for. This is the highest-value source because ranking is proof of demand.
- Question and modifier expansion. Programmatic combinations: “how to”, “best”, “vs”, “near me”, “for small business”. Cheap to generate, wildly variable in quality.
SEO Rocket’s keyword explorer runs multi-seed searches and returns up to 150 ideas per query with volume, keyword difficulty, CPC, global volume, and SERP features attached, against a country-specific index. That last part matters more than people expect. A Singapore site queried against the US index will look like it has no demand at all.
Volume is an estimate — treat it that way
Every third-party volume figure you have ever seen is modeled. Two reputable tools will disagree by 40% on the same term and both will be defensible. The practical rule: use volume for ranking candidates against each other, not for forecasting revenue. A term at 1,900/mo is probably bigger than one at 320/mo. Whether it delivers 1,900 impressions is unknowable until you rank.
Google Search Console is the only source that tells the truth about your own site. Once a page has been live sixty days, GSC impressions beat any tool’s estimate for that URL. Use suggestions to find opportunities; use GSC to confirm what actually happened.
The four filters that cut 3,000 suggestions to 30
Run every export through these in order. The list shrinks fast, which is the point.
- Intent match. Search the term. If page one is all product category pages and you planned a blog post, you lose before you write a word. Format follows the SERP, not your content calendar.
- Weakest-competitor test. Ignore the median. Look at the weakest page in the top ten — the one with the fewest referring domains or the thinnest content. If you can beat that page, the keyword is live. If even position ten has 400 referring domains, skip it this quarter.
- Business proximity. Does the searcher have a plausible path to your product within two clicks? “What is a 301 redirect” has volume and near-zero commercial value for most businesses.
- Cluster fit. A keyword that reinforces a topic you already cover is worth more than an orphan term with double the volume. Isolated pages rarely rank.
Difficulty scores: useful, but not a verdict
Keyword difficulty is a normalized 0–100 estimate, usually derived from the link profiles of the current top ten. It is a fast proxy, not a ruling. A KD of 12 on a SERP owned by Wikipedia, Reddit, and three government sites is harder than a KD of 34 on a SERP full of thin affiliate pages.
Open the SERP. Spend ninety seconds. Ask whether the pages ranking are genuinely good or just old. Old-and-mediocre is the single most reliable opportunity signal in search, and no score captures it.
Grouping suggestions into pages
Ten suggestions rarely mean ten pages. Terms that return substantially the same top-ten results are the same query in Google’s eyes and belong on one URL. Practical threshold: if three or more of the top ten URLs overlap, merge them.
So “seo keyword suggestion”, “keyword suggestion tool”, and “keyword ideas generator” almost certainly collapse into one page. Splitting them creates three thin pages that cannibalize each other and dilute internal links. One strong page beats three weak ones every time — that is not a philosophy, it is what the ranking data shows.
Once grouped, pick the highest-volume term in each cluster as the primary target and treat the rest as sections, subheadings, and natural on-page vocabulary.
A realistic weekly workflow
Here is a routine that takes about ninety minutes and produces four to six briefs.
- Monday, 20 min: Run three seeds — your core product term, one problem your customer has, one competitor domain. Export everything.
- Monday, 30 min: Apply the four filters. Expect to keep 1–2% of what you exported.
- Tuesday, 25 min: Cluster survivors by SERP overlap. Assign one URL per cluster.
- Tuesday, 15 min: Check the weakest page-one competitor for each cluster and note what it is missing. That gap is your brief.
Save the winners to a project keyword pool so the same list feeds your writing and your rank tracking. Re-researching the same terms every month is the most common waste of time in this workflow.
Common mistakes that kill otherwise good lists
Chasing volume alone is the obvious one. Here are the less obvious ones.
Ignoring the country index. Volume, difficulty, and SERP composition are all country-specific. Run your research against the market you actually sell into.
Treating zero-volume terms as worthless. Tools report zero when a term falls below their reporting threshold, not when nobody searches it. Long, specific, high-intent phrases frequently show zero and convert better than head terms.
Expecting rankings to hold still. Daily movement of two or three positions is ordinary noise. Judge a keyword decision on a trend across four to six weeks, never on a single morning’s check.
Publishing before checking what already exists on your own site. Half the “new” suggestions in a typical export are already partially covered by a page you forgot about. Updating that page usually outperforms writing a new one.
Turning a shortlist into published pages
The gap between a good keyword list and traffic is execution volume. Thirty solid targets are worth nothing sitting in a spreadsheet. This is where a workspace that carries the keyword through to a drafted, validated, published page earns its keep — SEO Rocket runs research, writing with hard quality gates, and rank tracking against the same keyword pool at a flat US$50/month, so a shortlist does not go stale between tools.
Whatever you use, keep the discipline simple. Suggestions are cheap. Judgment is the scarce input. Filter hard, cluster honestly, benchmark against the weakest competitor on page one, and measure with Search Console rather than estimates once your pages are live.
