Find Keyword Opportunities Hiding in Your Search Console Data

Most people go hunting for keyword opportunities in Search Console with the wrong tab open. They check Performance, glance at the top queries by clicks, and close the tab because nothing looks new. The actual opportunities in Search Console rarely show up in the queries already sending traffic. They show up in the queries a site is already ranking for, quietly, on page two or the bottom of page one, that nobody has ever looked at because they never generated enough clicks to catch anyone's eye.

I use Search Console for keyword research more than I use any dedicated keyword tool, paid or free, because it is the only source that tells you what Google already thinks a page is about. Keyword Planner and autocomplete tools tell you what people search for in general. Search Console tells you, with actual data, what your specific pages already show up for, at what position, and how often people see them without clicking. That is a different kind of signal, and it points to a different kind of opportunity: keywords a site is one improvement away from winning, rather than keywords it would need to build a page from scratch to chase.

Why impressions without clicks matter more than clicks

The queries report in Search Console Performance shows impressions, clicks, average position, and click-through rate for every query a page has appeared for, going back up to sixteen months. Most people sort by clicks and stop there. I sort by impressions instead, then look at which high-impression queries have a low click-through rate relative to their position.

A query showing up hundreds of times with an average position around eight or nine and a click-through rate near zero is not a dead end. It is a page Google already associates with that topic strongly enough to surface it repeatedly, just not strongly enough, or with a compelling enough title and description, to earn the click. That gap between "Google shows this page for this query" and "people click through" is the single most reliable free keyword opportunity signal I know of, because it removes the guesswork entirely. You are not speculating about demand. Google is already showing you the demand and telling you the page is missing the click.

Finding striking-distance keywords

The industry shorthand for this is striking distance: queries where a page ranks close enough to the first page, or low on the first page, that a realistic content or optimization push can move it meaningfully rather than needing a multi-month campaign. I pull this by filtering the queries report to a wide position window, exporting it, and sorting by impressions descending within that window.

What surfaces is usually a mix of three things. Some queries are close variants of a term the page already targets directly, and the fix is adding that phrasing somewhere prominent, in an H2 or the opening paragraph, rather than assuming the page needs a full rewrite. Some are adjacent questions the page half-answers, where expanding a section to actually address the query head-on can pull it up. And some are genuinely new intent the page never planned for, which sometimes justifies a new page rather than an edit to the existing one.

A worked example

I pulled the queries report for a client's core service page recently, an HVAC maintenance page ranking reasonably well for the exact-match term it was built around. Filtering for queries where the page appeared but sat outside the top handful of positions turned up a cluster around "hvac maintenance plan cost" and "how often should hvac be serviced," neither of which the page addressed directly. It had a pricing mention buried in a sidebar and no maintenance-frequency guidance anywhere.

Both queries had been generating impressions for months. Neither had generated meaningful clicks, because the page technically covered adjacent ground without answering either question in the text Google could quote or summarize. Adding a short section with clear frequency guidance and an honest pricing range moved both queries up within about six weeks, and the click-through rate on the frequency query in particular improved once the page actually contained a direct answer instead of implying one. None of that required new content strategy or an outside tool. The demand was sitting in data the client already owned.

Filtering out the noise

Not every low-CTR, high-impression query is worth chasing. Branded queries, someone searching the company name plus a service term, will often show a similar pattern and mean nothing about content gaps, since brand searchers often already know what they want and click regardless of the snippet. I filter those out early by excluding the brand name and close misspellings from the query list before I look at anything else.

Single-occurrence or very low-impression queries are the other trap. Search Console shows the long tail of near-random phrasing that happened to trigger an impression once, and treating a query with three impressions over sixteen months as a real opportunity wastes time on noise. I set a minimum impression threshold before I even look at a query's position, and that threshold moves depending on the site's overall traffic level rather than being a fixed number, since a low-traffic new site and an established one with years of data need different cutoffs to separate signal from noise.

Comparing devices before you act on anything

One filter people skip entirely is the device breakdown inside the same queries report. A query that looks like a striking-distance opportunity in the combined view can split very differently between mobile and desktop, and average position often differs by a full position or more between the two once you separate them out. I have seen pages sitting at position seven on desktop for a query while sitting outside the top fifteen on mobile for the identical query, usually because a slower mobile load time or a layout issue is quietly suppressing mobile rankings in a way the desktop version never shows.

That split matters for prioritization because most service-business search traffic now comes from mobile, particularly for local and near-me style queries. A striking-distance keyword that looks borderline in the blended report can actually be a near-miss specifically on mobile, worth fixing first, or it can be a desktop-only pattern that will not move the needle much given where the actual traffic sits. I check the device filter before committing to any edit priority list, not after, because it changes which fixes are actually worth doing first.

New page or edit the existing one

The decision that comes out of this data is almost always binary: edit what exists, or build something new. If the striking-distance query is a variation on intent the ranking page already serves, adding a section, adjusting a heading, tightening the title tag for that phrase, is the right move and it is fast. If the query represents intent the page was never built to serve, forcing it in usually dilutes the page rather than helping it, and a dedicated page performs better for both the new query and the one the original page already owned.

I default to editing first, because Search Console data on an existing page is a much stronger signal than volume data on a brand-new keyword idea. The page already has authority Google associates with the topic. A new page for the same intent starts from zero and has to earn that association from scratch, which is slower even when the keyword itself looks appealing in a tool like Keyword Planner.

The one exception is when the striking-distance query represents a fundamentally different stage of the buying process than the page currently serves. A service page written for someone ready to book a job will not naturally absorb an early-stage comparison query without becoming a worse page for its original purpose. In that specific case, a new page built for the comparison intent, internally linked back to the original service page, tends to outperform stretching one page to cover two jobs at once.

Combining Search Console with other free research

Search Console alone will not surface keywords a site has never ranked for even faintly, since it only reports on queries that generated an impression somewhere. For genuinely new topic ideas outside what a site already touches, I go to broader discovery methods like mining autocomplete and People Also Ask, which surface phrasing real searchers use without requiring any existing ranking as a starting point.

The two methods are not competing for the same job. Search Console tells you where to push on ground you already hold. Autocomplete and PAA tell you where new ground might be worth claiming. Running both against the same site over a quarter, rather than picking one, is how I build most content calendars for clients who already have some organic history to work from.

If you want a structured way to act on striking-distance findings without letting them sit in a spreadsheet, the 30-day SEO quick wins plan walks through prioritizing exactly this kind of data into a fixed sequence of edits. And if you'd rather have someone pull and prioritize this analysis for you against your actual traffic and conversion data, that is core to what the Website Optimizer Pack covers.

Search Console will not generate a keyword idea out of nothing, and it will not tell you search volume the way a dedicated keyword tool does. What it does better than any other free tool is show you, with real data instead of an estimate, exactly where your existing pages are close to winning terms they have not won yet. That is a narrower kind of keyword research than most guides describe, and in my experience it is also the fastest one to turn into ranking movement, because you are pushing on doors that are already half open rather than building new ones from the ground up. More of this kind of practical, no-subscription-required research lives in the Free Tool section.