How to Use Google Search Console for Keyword Research
Every other source in this series estimates what might happen on any site. This one reports what actually happened on yours. It is the only real search data a business has. Most never open it for this purpose.
The Only Real Search Data You Have
Search Console tells you which searches your site appeared for, how often and whether anybody clicked. Those are records rather than estimates. Google is reporting its own data about your own pages, which makes this the single most reliable research source available to any business.
Why it gets overlooked. It is filed as reporting.
Businesses treat it as a monitoring tool, something to check when traffic drops. Very few open it asking what to write next, which is what this page is about.
What it costs. Nothing.
Free, plus already connected on most sites. The barrier is knowing what to look for rather than access.
What it requires first. Live pages.
It only reports searches you already appear for, so a brand new site has nothing in it. That is the one real limitation and block seven covers the rest.
What this page will not do. Explain the report.
What each figure means and how the report behaves belongs to our guide to the performance report. This one is about reading it for opportunities.
Where the value concentrates. Two findings.
Searches you never targeted, plus searches you nearly rank for. Blocks three and four produce most of the useful output between them.
What Makes It Different From A Tool
A keyword tool models what might happen on any site. Search Console records what did happen on yours. That distinction sounds small and it changes what you can conclude, because one is evidence and the other is inference.
What a tool cannot know. You.
No external product knows which searches your particular site appears for, how your pages perform or what your visitors do. It is working from generalities.
What this knows. Exactly that.
Real searches, made by real people, that led to your pages being shown. Nothing modelled and nothing sampled from other websites.
Where the estimates still help. Discovery.
Tools suggest searches you have never appeared for, which this cannot. The two sources answer different questions and both have a place.
The practical consequence. Trust this first.
Where a tool and your own reporting disagree about your site, your reporting wins. That resolves most arguments about whether a subject has demand.
What it does to a proposal. Changes the conversation.
A supplier proposing subjects from estimates alone has ignored the better source. Asking what your own data says is a reasonable question.
When it becomes the main source. Sooner than expected.
Once a site has been publishing for a while, this replaces external research almost entirely for planning what comes next.
Queries You Never Targeted
Your pages appear for searches nobody planned. That list is the most useful research material a business has, because it is demand you did not know existed, expressed in words you did not choose, demonstrated rather than predicted.
Why it happens. Matching on meaning.
A page covering a subject properly gets shown for phrasings nobody listed. That is the tail arriving. It is the clearest evidence coverage works.
What the list reveals. The real vocabulary.
How customers actually describe the problem, which is frequently not how the page describes it. That alone is worth the exercise.
The pattern to look for. Clusters.
Several related unplanned searches all landing on one page. That page is carrying a subject it was never written to serve. It probably should not be.
What to do with a cluster. Consider a page.
If a group of related searches is arriving somewhere unsuitable, that group deserves its own page written deliberately for it.
What to do with a single oddity. Nothing.
One unusual search appearing once is noise. Wait for it to repeat before treating it as a finding.
The other use. Fixing the existing page.
Sometimes the answer is not a new page. It is adding a section to the page already receiving those searches, which is faster and frequently better.
Positions Just Off The Front
Searches where your site appears consistently but not well enough to be found are the cheapest wins available. The demand is proven, the relevance is established and something is already there. Improving that beats starting a new subject from nothing.
Why these are cheap. The hard part is done.
Getting a page to appear at all for a search is most of the work. Moving it from just off the front to visible is a smaller job than building from scratch.
How to spot them. Sort by position.
Look for searches sitting close to the first page without reaching it. We publish no numbered band, since where the cut-off matters varies by search and by how the results are laid out.
Which ones to act on. The commercial ones.
Apply the same filter as everywhere else. A near miss on a search nobody buys from is not worth the effort of closing.
What improving usually involves. Answering better.
Reading what currently ranks above you and asking what it covers that you do not. Frequently the answer is something specific you know and did not think to write down.
What it rarely involves. More words.
Length is not the lever. A page that answers the search more completely wins. That can mean rewriting rather than extending.
Why this suits small businesses. Finite work.
A handful of near misses is a defined task with a visible end, which fits businesses without a content team far better than an open publishing plan.
Appearing Often And Being Ignored
A page shown many times for a search that almost nobody clicks is telling you something. It is one of two problems and they need opposite responses, so working out which you have matters before anybody edits anything.
The first possibility. Too far down.
You appear, though not where anybody looks. This is really a near miss and belongs in block four rather than here.
The second possibility. Visible and unconvincing.
You appear prominently and people choose somebody else. What they see in the results is not persuading them, which is a different fault entirely.
How to tell them apart. Check the position.
The report shows roughly where you sit. Prominent with poor uptake is the second problem. It is the more interesting one.
What causes the second. The wrong promise.
A title and description that do not match what the searcher wanted. Or that read less usefully than the results around them.
The other cause worth knowing. Answered already.
Some searches get answered on the results page itself, so nobody needs to click. That is not a fault in your page and no amount of editing changes it.
What to do about it. Rewrite what is shown.
How titles and descriptions get written belongs to our on-page material rather than here. It is usually a short job.
Comparing One Period Against Another
A single period tells you the current state. Two periods compared tell you the direction, which is usually the more actionable information. This is the feature businesses use least and it produces the clearest findings.
The comparison worth making. Like against like.
The same span this year against last year, so seasonal patterns do not masquerade as changes. Most trades have a shape to their year that would otherwise mislead.
What appears in a comparison. New queries.
Searches present now that were absent before. Those show what your recent work actually did, rather than what it was intended to do.
The other useful column. What disappeared.
Searches you used to appear for and no longer do. That is an early warning worth investigating before it shows up as a drop in enquiries.
What to check after publishing. The new page.
Compare the period since a page went live against the one before. That tells you whether it picked up the subject it was written for.
What not to conclude. Cause from coincidence.
A change in one period following a change on the site does not prove the connection. Several other things were also happening.
The habit worth forming. Once a month.
A short look at what is new and what has gone, recorded somewhere. That accumulates into a picture nothing else provides.
What It Cannot Tell You
This is the best source available and it is not complete. Knowing where it stops prevents the two mistakes businesses make with it, which are trusting it as a full picture and abandoning it when the numbers look odd.
It only shows where you appear. The main limit.
Searches you have never appeared for are invisible here, which is precisely why external research still has a role for discovering new subjects.
Rare searches are withheld. By design.
Uncommon queries are held back to protect the privacy of the people making them. So the smallest and most specific searches are underrepresented, which affects local businesses most.
The figures are not exact. Processed.
Counts are rounded and grouped in various ways. Read them as scale and direction rather than as precise totals.
It says nothing about money. No outcomes.
A search producing clicks might produce enquiries or might produce nothing. Connecting the two requires your own records of who got in touch.
It cannot tell you why. Only what.
A position changing is reported and never explained. The reason sits with what you did, what competitors did or what changed in search itself.
What that means practically. Use it as evidence.
Strong evidence about your own site, alongside judgement rather than instead of it.
Turning It Into New Pages
Findings are only worth having if they become work. The output of this exercise should be a short list of specific actions, each attached to a page, added to the plan you already have rather than kept in a separate document nobody opens.
The three actions available. Improve, extend or build.
Improve an existing page that nearly ranks. Extend one that is attracting a related subject. Or build a new page where a cluster clearly deserves one.
Which to do first. Improve.
Near misses are the fastest return and the least work. Do those before writing anything new.
Where the findings go. The keyword map.
Amend the map rather than starting a new list. Our content strategy guide covers keeping that plan alive.
What to record alongside. The reason.
Why each item is on the list, in one line. In three months nobody will remember. An unexplained list gets ignored.
What this pairs with. Competitor gaps.
Your own data shows where you nearly succeed. Looking at what competitors miss shows where nobody has succeeded yet.
Where the rest sits. Two hubs.
The keyword research guide for this series. The Google Search Console guide covers the tool itself.
You are already
appearing for
searches nobody
planned for.
Every other source estimates what might happen on any site. This one records what did happen on yours. The searches your pages already show up for, in words you never chose, are demonstrated demand rather than somebody's model of it. Most businesses have never opened this asking what to write next.
What we look for in your own data:
Where an external tool and your own reporting disagree about your site, your reporting wins.
Every guide.
One subject.
What the words mean, what the numbers in a keyword tool actually are, how to find the searches worth targeting, how to turn a list into a publishing plan and what to do when a page you relied on slips.
