If you want to know how to see what keywords a website is using, the honest answer splits in two. You can read the keywords a site has deliberately placed on its pages for free, using your browser and a few search operators. You cannot see which keywords actually bring it traffic without a third-party keyword database, because that data lives in Google Search Console and only the site owner can open it.
Both halves are useful, and they answer different questions. The on-page half tells you what a competitor is targeting. The database half tells you what is working. Most competitive research needs the second, but the first is free and takes ninety seconds, so start there.
Free method one: read the page source
Open any page, right-click, and choose View Page Source. Then search the HTML for the fields where targeting is visible.
- <title> — the strongest single declaration of intent on a page. Whatever the site wants to rank for is usually right there, near the front.
- <meta name=”description”> — written for clicks, but almost always contains the primary term and a variant or two.
- <h1> and <h2> — the subtopic map. The H2s show you which questions the page is built to answer.
- URL slug — often the cleanest version of the target keyword.
- Image alt text and internal link anchors — anchor text pointing out of the page reveals what the site thinks its other pages are about.
One field to ignore: <meta name="keywords">. Google stopped using it in 2009 and has said so publicly. If a site still populates it, you are looking at a legacy plugin, not a strategy — though occasionally it is a lazy honest list of exactly what they are chasing.
Free method two: search operators
Google itself will show you a site’s keyword surface if you ask correctly.
site:example.com— lists indexed pages. The titles alone give you the topic map.site:example.com "term"— finds every page on that domain mentioning a specific term. Useful for testing whether a competitor has covered a topic you are considering.site:example.com/blog/— isolates a content section so you can see editorial targeting separately from product pages.site:example.com intitle:"keyword"— surfaces pages built specifically around a term rather than mentioning it in passing.
Two limits worth knowing. Result counts from a site: query are rough estimates, not a page census, so do not treat “About 4,320 results” as a number. And this method shows what is indexed, not what ranks — a site can have 4,000 indexed pages and receive traffic to eleven of them.
Free method three: check your own site properly
If the website in question is yours, skip everything above. Google Search Console’s Performance report gives you every query that produced an impression, with clicks, average position, and CTR, across the last 16 months. That is Google’s own data about your own pages, and no third-party estimate comes close.
Filter by page to see which queries a specific URL matches. You will find terms you never targeted, and those are the most interesting rows on the report — Google telling you what your page is actually understood to be about. Queries with high impressions and near-zero clicks are the ones to act on first.
Search Console does hide low-volume queries for privacy reasons and samples aggressively on large sites, so the totals will not reconcile perfectly with your analytics. That is expected, not a bug.
Paid method: keyword databases and site explorers
To see what a competitor ranks for, you need a tool that maintains its own index of search results. These tools crawl millions of keywords, record which URLs appear at which positions, and let you query that index by domain.
Enter a competitor domain into a site explorer and you get their organic keyword list with estimated position, estimated volume, difficulty, and the ranking URL. Sort by traffic contribution and you learn something specific and valuable: the ten or twenty terms actually carrying that business. Sites usually earn most of their organic revenue from a much smaller keyword set than their total footprint suggests.
SEO Rocket’s site explorer runs this on any domain — domain rating, backlinks, referring domains, organic keywords, top pages, anchors, and spam flags — on industry-grade search data with country-specific indexes, at a flat US$50/month. Country selection matters more than people expect. Check a UK or Singapore competitor against the US index and the report will look almost empty, and you will draw exactly the wrong conclusion.
Read the estimates for what they are
Every number a third-party tool reports about someone else’s site is modeled. Volume figures are typically twelve-month averages built from clickstream data and index sampling. Positions come from periodic crawls, so a keyword checked last Tuesday may have moved since. Traffic estimates multiply an estimated volume by an assumed click-through curve, which stacks two layers of error.
That does not make them useless. Relative comparisons hold up well: if a tool says competitor A gets roughly five times the organic traffic of competitor B, that ratio is usually sound even when both absolute figures are off. Use these numbers to rank opportunities and size markets. Do not use them to forecast revenue or to argue about a competitor’s exact traffic in a board meeting.
Turn the list into a plan with a content gap
Seeing a competitor’s keywords is only step one. The step that produces work is the content gap: the terms several competitors rank for and you do not.
Run a gap across three to five direct rivals rather than one, because a single competitor’s list includes plenty of accidental rankings. Terms that two or three rivals all rank for are structural — they represent real demand in your market that you are absent from. SEO Rocket runs content gap across up to five competitors with per-rival position columns, plus anchor-text and backlink gap analysis with a named outreach list and directional cost bands.
Then filter the output hard. Keep terms where the intent matches something you sell or explain, where the weakest page-one result looks beatable, and where the volume justifies a page. Drop branded terms belonging to competitors, informational terms with no path to revenue, and anything where the SERP is dominated by formats you cannot produce.
A ninety-minute workflow
Put it together in one sitting. Spend fifteen minutes on source-code and site: operator checks of your three closest competitors to understand what they are deliberately targeting. Spend thirty minutes in a site explorer pulling their actual ranking keywords sorted by traffic contribution. Spend thirty minutes running a content gap across all of them and filtering to terms you could realistically win. Spend the last fifteen mapping surviving terms to existing pages that could be optimised, versus new pages you would need to write.
You will end with two lists: quick wins on pages you already have, and a prioritised production queue. That is a better competitive plan than most sites ever build, and the raw material was sitting in public the whole time.
