Here is a question I hear from business owners who have had a website for a few years: “Which keyword is this page actually supposed to rank for?” Often nobody knows. The page was written years ago, maybe by someone who has left, and it now ranks for a mix of searches that nobody planned.
Most people assume the answer is in a keyword tool. They look up a few phrases, pick the one with the biggest search volume, and stuff it into the title. That is usually the wrong move for an existing page, because the page already has a history. Google has already decided what it thinks the page is about, and that decision is sitting in your Search Console data.
In this guide, let me explain how I would pick the primary keyword for a page that already exists, using the queries it already gets. I will show you how to pull and group the data, how to choose one main target and a few secondary ones, how to re-optimize without breaking what already works, and how to tell if it helped.
Why an old page ends up ranking for things you never planned
A page is not frozen when you publish it. Over time, three things change around it. You add or remove content. Other pages on your site start covering similar topics. And Google gets better at matching the page to searches that use different words from yours.
The result is what I call keyword drift. The page was meant for “emergency plumber,” but now most of its impressions come from “burst pipe repair cost.” Or a blog post meant for “how to clean a sofa” is getting traffic for “remove water stains from fabric.”
Drift is not always bad. Ahrefs found that the average page ranking first for a keyword also ranks in the top 10 for nearly 1,000 other relevant keywords (Ahrefs, How Many SEO Keywords Should a Page Really Target?). Ranking for many queries is normal. The problem starts when the page has no clear center, so its title, heading and content pull in different directions and none of those queries reach the top spots.
Here are the most common signs that a page has drifted:
- The query with the most impressions is not the one in the title tag.
- Impressions are spread thinly across dozens of queries, with no single query getting more than a few percent of them.
- The page sits in positions 8–20 for most of its queries and never breaks into the top five.
- Another page on your site gets impressions for the same top queries (a sign of keyword cannibalization).
- The page’s intent has moved, for example from “how to” searches to “near me” or “price” searches.
Pulling the page’s real queries out of Search Console
Search Console is the only free source that tells you which searches Google actually shows your page for. Keyword tools estimate. Search Console reports. If you are new to the tool, my guide to the Search Console Performance report walks through every tab, and the beginner lesson on Google Search Console covers setup.
Here is how I would pull the data for one page:
- Open Performance → Search results and set the date range to the last 3 months. That is long enough to smooth out odd days but recent enough to reflect the page as it is now.
- Add a filter: Page → Exact URL and paste the full address of the page.
- Switch on all four metrics: clicks, impressions, CTR and average position.
- Go to the Queries tab and sort by impressions, highest first.
- Click Export and download the list to Google Sheets or Excel.
- Repeat with a last 16 months range if the business is seasonal, so you can see queries that only appear at certain times of the year.
Two things in the data trip people up. First, the most recent days can be incomplete. Google’s help page warns:
“The newest data can be preliminary, meaning it’s still being collected and might change in the next few hours.”
— Google Search Console Help, Performance report (Search results)
Second, the totals in the chart and the totals in the table can differ, because Google aggregates queries by property and pages by page. Don’t panic if your query rows don’t add up to the page’s total clicks. Some queries are also hidden for privacy, so your export will never show every search.
If you want to filter queries more precisely, for example only questions or only searches containing a city name, the regex filter helps. I explain the patterns in Search Console regex filters.
Sorting the export into topic groups
A raw export of 300 queries is hard to read. The trick is to group queries that mean the same thing, then judge each group as one unit. This is a small, manual version of keyword clustering.
In your sheet, add a column called “Group.” Then label each query. I use three simple rules:
- Same need, different words go together. “Boiler repair,” “fix my boiler” and “boiler repairs near me” are one group.
- Same words, different need stay apart. “Boiler repair cost” (someone budgeting) and “boiler repair” (someone ready to call) may need different pages.
- Brand and junk queries get their own label so they don’t distort the picture.
Once every query is labeled, build a pivot table that sums impressions and clicks per group and shows a weighted average position. Now you can see the page’s real shape. Here is a simple worked example.
Example: a heating company’s boiler repair page
This is a hypothetical example to show the method. Imagine a page at /boiler-repair/ with 3 months of data:
| Query group | Queries in group | Impressions | Clicks | Avg. position |
|---|---|---|---|---|
| Boiler repair (+ near me, + city) | 41 | 18,400 | 390 | 9.8 |
| Boiler repair cost / price | 23 | 9,100 | 120 | 12.4 |
| Boiler not heating water / no hot water | 34 | 6,700 | 95 | 14.1 |
| Emergency boiler engineer | 12 | 2,300 | 60 | 7.2 |
| Brand name queries | 6 | 800 | 210 | 1.3 |
The page was titled “Heating Services | Brand.” It is getting most of its impressions for boiler repair searches, but it is stuck around position 10 because nothing on the page says “boiler repair” clearly. That is a page asking to be re-centered.
The test I use to pick one primary keyword
Ahrefs puts the general rule simply:
“Here’s what we recommend: pick just one primary keyword and enough secondary keywords to cover a given topic in full.”
— Ahrefs, How Many SEO Keywords Should a Page Really Target?
I agree. For an existing page, though, I don’t pick the primary keyword from a tool. I pick it from the strongest group, using four questions. A group should pass all four.
| Question | What to look at | Pass if… |
|---|---|---|
| 1. Does Google already link this page to the topic? | Impressions share of the group | The group has the largest or second-largest share of non-brand impressions |
| 2. Is the page close enough to win? | Weighted average position | Roughly positions 4–20, where a clearer focus can move it |
| 3. Does the page match the intent? | The current top 5 results for the main query | The top results are the same type of page (service page, guide, product list) |
| 4. Is it worth money to the business? | What the searcher wants to do next | The searcher could become a lead or a sale from this page |
In the boiler example, the “boiler repair” group passes all four. The cost group fails question 3 if Google shows price guides at the top. The “no hot water” group is a problem-solving search and probably belongs on a separate troubleshooting article.
Inside the winning group, I then pick the single phrase that is the clearest and most common way people say it. Usually that is the shortest query with the most impressions, such as “boiler repair.” I don’t chase the city variant as the primary keyword unless the page is a dedicated location page.
When two groups are nearly tied
Sometimes two groups are close, for example “sofa cleaning” and “upholstery cleaning.” First check whether Google treats them as the same thing. Search both and compare the top 10. If 5 or more of the same URLs appear, one page can serve both, and I pick the one that sounds more natural to your customers. If the results are different, you likely need two pages. My guide to targeting multiple keywords on one page explains this overlap test in more detail.
Choosing secondary keywords from the same data
Secondary keywords are not extra keywords to sprinkle around. They are the subtopics and wordings that help the page answer the main search fully. Ahrefs makes the point that “Google is good at catching close variations and misspellings,” so you don’t need every variant on the page.
From your grouped export, a query becomes a secondary keyword if it does one of these things:
- It adds a subtopic the page should cover, such as “boiler repair cost” as a short pricing section on the service page.
- It uses a different word customers use for the same thing, such as “boiler engineer” next to “boiler repair.”
- It adds a qualifier that changes the details, such as “same day” or “weekend.”
I usually end up with 3–6 secondary keywords. If the list grows past 10, the page is probably trying to do two jobs. For the difference between true variants and new topics, see keyword variations and synonyms.
Queries that don’t fit go into one of three buckets: a new page (different intent), an existing page elsewhere on the site (link to it), or ignore (irrelevant). Write these decisions into your keyword map so the next person doesn’t undo them.
Checking the rest of your site before you commit
Before you rewrite anything, make sure no other page is already the better match for your chosen primary keyword. In Search Console, remove the page filter, add a Query → Exact query filter with your primary keyword, and open the Pages tab.
| What you see | What it means | What I would do |
|---|---|---|
| Only your page appears | Clear ownership | Go ahead and re-optimize |
| Your page plus a blog post, with the post getting few impressions | Mild overlap | Link the post to the page with clear anchor text |
| Two pages swapping impressions week to week | Cannibalization | Merge, or change one page’s focus |
| A different page gets most impressions | Google prefers the other page | Make that page the target, and give your page a different keyword |
Re-optimizing the page without losing what already ranks
An existing page has earned some trust. You don’t want to throw it away with a total rewrite. My approach is to re-center, not replace. Here is the order I work in:
- Title tag. Put the primary keyword near the start in natural words, for example “Boiler Repair in Leeds | Same-Day Engineers | Brand.” Google’s guidance is that titles should be distinct and descriptive, and it may rewrite them otherwise (why Google rewrites titles).
- H1. Make the main heading say the same thing in plain words. It doesn’t have to match the title exactly; see title tag vs H1.
- Opening paragraph. Within the first two or three sentences, confirm the page answers the primary search.
- Subheadings. Turn secondary keywords into sections where they truly add something, for example “How much does boiler repair cost?”
- Internal links. Update 3–5 links from related pages to point here with descriptive anchor text.
- Remove what pulls the page off course. Move unrelated sections to their own pages instead of deleting useful content.
Google’s own self-check is a good test for the title and H1:
“Does the main heading or page title provide a descriptive, helpful summary of the content?”
— Google Search Central, Creating helpful, reliable, people-first content
What I would not change: the URL (unless it is truly misleading), sections that already earn clicks for other queries, and content that has backlinks pointing at it. Changing a URL adds risk you rarely need for a keyword refocus.
A quick note on repetition
Re-centering does not mean repeating the keyword. Once in the title, once in the H1 and naturally in the copy is plenty. If your draft says “boiler repair” in every paragraph, read SEO over-optimization warning signs before you publish.
How to tell if the new focus worked
Write down your baseline before you publish: clicks, impressions, CTR and average position for the primary keyword group and for the page as a whole, using the same 3-month window. Then add a note in Search Console so you remember the date. I explain how in using Search Console annotations to track updates.
Google’s SEO Starter Guide says some changes might take effect in a few hours while others could take several months, and it suggests waiting a few weeks before judging a change (SEO Starter Guide). I compare 28 days after Google recrawls the page with the 28 days before the change, and again at 8–12 weeks.
What a good result looks like:
- The primary keyword group’s average position improves and its share of the page’s impressions rises.
- Clicks from the primary group rise, even if total impressions stay flat.
- Secondary groups hold steady or improve.
- Brand queries are not part of the comparison, because they move for other reasons.
For a full method, including how to handle seasonality, see my companion guide on how to measure whether an on-page SEO change worked.
Questions owners ask about retargeting an old page
Should the primary keyword be the query with the most clicks?
Not always. Clicks often come from brand searches or from a query you already rank first for. Look at impressions per topic group and at position. The best primary keyword is usually the group with the most impressions where you are not yet in the top three.
Can I pick a keyword the page doesn’t rank for at all?
You can, but then you are effectively creating a new page in an old URL. If the page has zero impressions for a topic after months, Google doesn’t connect it to that topic. A new, dedicated page is often cleaner.
What if the page’s best query has the wrong intent for my business?
Then don’t chase it. If a service page gets impressions for “how to fix X yourself,” write a separate guide for that intent and link it to the service page. Keep the service page focused on people ready to hire.
How many secondary keywords should one page have?
There is no fixed number. In practice, 3–6 well-chosen secondary keywords cover most service pages and guides. If you need more than 10, check whether two pages are hiding inside one.
How long should I wait before trying a different primary keyword?
I suggest at least 8–12 weeks after Google has recrawled the updated page. Changing focus every few weeks gives Google mixed signals and makes your data impossible to read.
Retargeting your first page this week
In brief: don’t start with a keyword tool. Start with the queries Google already shows the page for, group them by meaning, pick the strongest group that you can realistically win and that matters to your business, and re-center the title, H1 and content around it. Then measure against a clean baseline.
If you have dozens of older pages and no clear map of what each one targets, that is exactly the kind of cleanup our SEO team does every month. Tell us about your site on our get a quote page and we will take a look.
Further reading on primary keyword for an existing page
- How to Measure Whether an On-Page SEO Change Worked
- AI-Generated Alt Text: Speed vs Quality
- Title Tags for Local and Service Pages
Sources for this primary keyword selection guide
- Ahrefs – How Many SEO Keywords Should a Page Really Target?
- Ahrefs – How to Do Keyword Optimization for SEO
- Google Search Console Help – Performance report (Search results)
- Google Search Central – SEO Starter Guide
- Google Search Central – Creating helpful, reliable, people-first content
- Google Search Central – Influencing your title links
- Google Search Central – Debugging drops in Google Search traffic
