Most websites have more traffic sitting in a drawer than they realize. It is not hidden in new topics or new pages, but in search queries Google already associates with you and simply does not show far enough up. Position 6, position 9, position 12: that is where you collect impressions but hardly any clicks. A few places higher, the same row would look completely different.
In SEO these are called striking distance keywords. The idea is always the same: keywords just short of page one or just short of the top spots, where a manageable amount of work makes a visible difference. This piece shows how to find them in Google Search Console yourself, how to estimate their potential roughly, what to do with the candidates, and where the list will mislead you.
Why positions 4 to 15 are the biggest lever
Clicks are not spread evenly down the results page. They fall off steeply from the top, and after position 3 the curve drops especially hard. The exact figures vary a lot by study, industry and layout of the results page, the shape of the curve does not. As a rough order of magnitude: the top three positions share most of the clicks, position 3 sits somewhere around ten percent, from position 8 on it is a few percent, and page two usually gets one percent or less.
That leads to an uncomfortable calculation. A keyword on position 1 can hardly get better. A keyword on position 40 may still bring nothing after a lot of work, because Google clearly does not consider the page relevant for that topic. In between is a band where Google already thinks your page fits and just not that it is the best answer. There, a better title, a missing section or a few internal links are often enough to gain several places.
Why 4 to 15 in particular? The lower bound is easy: anyone already in the top three has made the big jump. The upper bound is a convention. Some use 4 to 10 and only look at page one, others go up to 20. Including 11 to 15 is worth it because those keywords slip onto the bottom of page one whenever the position fluctuates a little, and because the jump to page one makes the biggest relative difference there. Beyond roughly position 20 it is rarely a matter of fine-tuning.
One thing up front so the numbers make sense: the position in Search Console is an average. It is the impression-weighted mean of your page's best placement, across all devices, locations and days in the period. Position 8 does not mean you rank 8th, it means that is where you appear on average. How to read that number properly is covered in depth in Impressions but no clicks.
Step by step: finding striking distance keywords in Search Console
All you need is a Search Console property with a few weeks of data. If that is still missing, see Setting up Search Console.
1. Pick the right report and period. Open Performance and the "Search results" report. Set the date range to the last three months. Shorter periods are too thin for small sites, longer ones mix old positions into the average that no longer apply. Turn on all four metrics at the top: clicks, impressions, average CTR and average position.
2. Switch to queries. Below the chart is the table with the "Queries" tab. That is your raw list. It does not show every search query, though; more on that below.
3. Filter by position. Above the table on the right is a filter icon that narrows the rows by a metric. Choose "Position", "Greater than" and enter 3. That removes every keyword that is already at the top. You draw the upper bound at 15 while reading: once you sort by impressions in the next step, rows far back rarely end up on top, and you can spot the rest in the position column. If you want it clean, export the table and narrow it to 4 to 15 in a spreadsheet.
4. Sort by impressions. Click the "Impressions" column header until the largest values are on top. Impressions stand for demand here: they tell you how often Google showed you for the query at all. A row with 3,000 impressions on position 9 is worth more than one with 80 impressions on position 5, even though the second is closer to the top three.
5. Take out your brand. Queries containing your name distort the list, because you usually rank at the top for them anyway, and where you do not, optimization helps little. Click "Add filter" at the top (or "+ New", depending on the view), then "Query", then "Custom (regex)" and "Doesn't match regex". Enter every spelling of your name, separated by a vertical bar, for example example.?bakery|examplebakery|meyer bakes. Matching ignores case. Remember typos, the names of owners or products, and the domain with its extension.
Since late 2025, Search Console also offers its own brand filter for many properties that detects branded queries automatically. It is a good start, but it does not catch every personal or product name. The regex remains the check you control yourself.
6. Check for seasonality. A keyword sitting on position 7 in summer because hardly anyone searches for it right now is a different case from one that is stuck there all year. Open the date filter, switch to "Compare" and set the last three months against the same period a year earlier. The table then shows both values side by side. If a keyword's impressions swing sharply between the years, you have a seasonal topic. That is no reason to drop it, but a reason to work on it ahead of the season rather than in the middle of it.
7. Set a floor. Rows with very few impressions are noise. A single impression on position 5 does not make a candidate. Where to draw the line depends on the size of your site. For a small site, 50 to 100 impressions per quarter can already be a signal; for a large shop you would rather start at a few thousand. The "Impressions", "Greater than" filter handles that in the table.
What remains is a list that is rarely longer than a few dozen rows. That is a good thing. Nobody works on a hundred keywords at once.
Estimating the potential roughly
Sorting a list by impressions is a start, but it does not answer the real question: what does it bring if this keyword moves up? A simple calculation is enough.
Take the keyword's impressions, multiply them by a typical CTR on position 3 and subtract the clicks you already get. For position 3 you can roughly work with ten percent. The result is an estimate of how many additional clicks are on the table.
A made-up example for a bakery website, example-bakery.com:
| Keyword | Impressions (3 months) | Position | Clicks | Potential at position 3, rough |
|---|---|---|---|---|
| spelt bread without yeast | 7,200 | 6.8 | 210 | approx. 500 more clicks |
| buy sourdough bread | 1,800 | 12.4 | 15 | approx. 165 more clicks |
| rye bread recipe | 900 | 4.6 | 55 | approx. 35 more clicks |
The third row is closest to the top and brings the least. The first is further away and brings the most. That pattern is common, and it is exactly why the calculation is worth doing.
Three caveats belong with it, so the number does not promise more than it can:
- The CTR on position 3 is an average. On a results page with an AI Overview, ads or a map pack, position 3 gets considerably less. On a branded query, more. Type the most important candidates into Google yourself once and look at what sits above the regular results.
- Impressions can change. Moving from page two to page one usually brings more impressions, because page two only counts when someone pages forward. For keywords beyond position 10 the estimate is therefore on the cautious side.
- Position 3 is not a promise. The calculation says what a climb would be worth, not that it will happen. It is a sorting tool, not a forecast.
For the order, combine the potential with the distance. A keyword on position 5 with decent potential is often won faster than one on position 14 with a bit more. A workable rule of thumb: start where potential and proximity both look good, and treat candidates beyond position 10 as a medium-term task.
What to do with a candidate
You cannot optimize a keyword. You can only optimize the page that ranks for it. That is why every piece of work on a candidate starts with the same question.
Which page actually ranks? Click the keyword in the table. Search Console sets it as a filter, and the "Pages" tab shows which URLs get impressions for it. If there is a single page, things are clear. If two or three share the impressions, you may have a cannibalization problem; more on that shortly.
Does the search intent fit? Type the keyword into Google yourself and look at the first three results. Are they guides, product pages, recipes, local listings, videos? If your product page ranks for "spelt bread without yeast" while the top spots are all recipes, fine-tuning will not get you past them. Then the question is not how to improve the page but whether you need a different one. If the type of page fits, carry on.
Title and meta description. Is the keyword or a very close phrasing in the title? Does the title promise something that matches the query? For candidates on positions 4 to 7, the title is often the fastest lever, because it affects position and CTR at the same time. What makes a good title is covered in Title and meta description.
Deepen the content. Read the other queries the same page ranks for, and the top results ahead of you. Which sub-questions do they answer that yours does not? Often it is not more text but a specific section: a table with quantities, a paragraph on an edge case, a clear answer in the first sentence. Add in a targeted way rather than making the page longer. What a revision that actually counts looks like is covered in Updating existing content.
Add internal links. Find the pages on your site that already rank well or get a lot of visits and fit the topic. Link from there to the candidate with descriptive anchor text. Descriptive means text that says what the target page is about, not "click here". Internal links are the lever most often forgotten, because the work does not happen on the page itself.
One thing at a time. Note the date you changed something and wait four to six weeks before judging. If you change titles, content and internal links on ten pages at once, you will not know afterwards what worked. One page a week is a pace you can keep.
The typical traps
The list from Search Console looks clean. A few rows in it are not, and if you do not recognize them, you put work in the wrong place.
Branded queries that slip through. The regex only catches what you enter. The owner's name, an old product, an abbreviation regular customers search for: all of these show up as seemingly normal keywords. If a row has a CTR far too high for its position, it is often a branded query.
Queries from bots and tools. Some rows do not come from people. Queries in quotation marks, with operators like site:, inurl: or intitle:, with a minus in front of a word or with odd characters at the start are usually generated by rank trackers, scrapers or SEO tools. They bring impressions without real demand behind them. Filter them out, for example with a second regex filter that excludes quotation marks and colons.
Anonymized queries. Search Console does not show every query. Google leaves out rare queries for privacy reasons, and for many sites they account for a noticeable share of clicks. The list is a selection, not the complete picture. That is not something you can fix, but it is a reason not to overweight individual rows.
Cannibalization. If two of your pages take turns ranking for the same keyword, the average position jumps around and neither really moves up. In the table that looks like a candidate on position 9; in reality you have a structural problem. The fix is not to improve one page but to decide which one is responsible and to link the other to it or merge the two.
The average across devices and countries. Position 8 can mean position 4 on desktop and 12 on mobile. Or position 5 in Germany and 20 in Switzerland. If a candidate looks odd, filter by device and country once before you touch the page.
Data delay. Search Console lags two to three days behind, and the most recent days are often incomplete. If you change something on Monday and check on Wednesday, you see nothing or the wrong thing. An effect on position tends to show after weeks rather than days anyway. If you are looking for a drop rather than a candidate, Lost rankings is the better starting point.
Where to start
- Open the performance report, three months, "Queries" tab, all four metrics on.
- Filter "Position greater than 3", sort by impressions, exclude your brand via regex.
- Weed out bot queries with quotation marks and operators.
- For the top twenty rows, estimate the potential roughly: impressions times ten percent, minus current clicks.
- Pick three candidates where potential and proximity to the top both look good.
- For the first one, find the ranking page, look at the results page, check the search intent.
- Make one targeted change, note the date, wait four to six weeks, then compare.
How Traffalyzer takes this over
The search above takes half an hour. Repeating it every week is the part where it usually falls apart. Traffalyzer handles the mechanical side from your own Search Console: the "Keywords close to page 1" card in the growth plan shows how many keywords sit on positions 4 to 15, roughly how many clicks a jump into the top three would bring, and which keyword is the biggest lever. The Keywords tab has the full list with the "Positions 4–15" filter.
Traffalyzer takes branded queries out automatically based on the domain name, and you add your own terms such as personal or product names. Queries with quotation marks and operators are filtered out. The threshold at which a keyword counts adapts to the size of the site, so small sites get a list too. And every week the weekly letter has a "Task of the week" for one of these keywords. Which page you change for it, and how, remains your call.
Frequently asked questions
What are striking distance keywords?
Search queries your site already ranks for, but just below the top spots, usually positions 4 to 15. For these keywords Google already considers your page relevant, so a manageable amount of work is often enough to gain a few places and noticeably more clicks.
Why positions 4 to 15 in particular?
Because CTR drops sharply after position 3 and is close to zero on page two. In the top three the big jump is already done; beyond position 20 Google usually does not consider the page relevant enough for the topic for fine-tuning to be enough. The bounds are a convention, some work with 4 to 10 or 5 to 20.
Can I filter by position in Search Console?
Yes, via the filter icon on the right above the table. Choose "Position" and "Greater than" or "Smaller than". The filter applies to the rows the report has loaded, so at most 1,000. Sort by impressions first so the important rows are included. For a fixed band like 4 to 15, exporting to a spreadsheet is the safest way.
How do I exclude my brand?
With a filter on the query: "Custom (regex)" and "Doesn't match regex", then every spelling of your name separated by a vertical bar. Many properties now also have a built-in brand filter. It is convenient but does not recognize every personal or product name, so it is worth checking the list afterwards.
How long does it take for the position to improve?
Usually a few weeks. Google has to recrawl the page, and Search Console shows changes with a two to three day delay. You can request reindexing of a changed page via URL inspection; that speeds up crawling, not evaluation. Judge after four to six weeks at the earliest.
Should I write a new page for a striking distance keyword?
Almost never. The existing page has already built trust for the topic; a new one starts from zero and, at worst, competes with the old one. A new page is only worth it if the search intent calls for a completely different kind of page, such as a recipe instead of a product page.
How many keywords should I work on at once?
Few. One page a week is a good pace. If you change many pages at the same time, you cannot say afterwards what worked, and you lose the chance to learn from your own data.