Striking distance keywords are the queries your site already ranks for at positions 4 through 15: close enough to the top that one on-page fix can move them, far enough down that they earn you almost nothing today. Google has already decided your page belongs in the results. The remaining gap is usually specificity, not authority. That makes these the cheapest wins in SEO, and most sites never look for them.
Versions of this tactic have circulated in SEO teaching for years, usually as "low-hanging fruit." What follows is our corrected version: the actual filter settings, the step almost every version skips, and the real odds.
Why positions 4 to 15, not "page two"
The classic advice says hunt positions 7 to 20. The band that pays is tighter and higher.
Positions 1 to 3 are already won. There's headroom, but it's small, and you'd be polishing your best work instead of fixing your almost-best. Positions beyond 15 usually have a real problem: thin content, wrong intent, or a competition level the page can't survive. Those queries feed your new-content pipeline, not your quick-wins list.
Positions 4 to 15 are different. The page cleared indexing, cleared topical relevance, and entered the consideration set. It's losing on one of exactly two things, and both are fixable in an afternoon. The click curve is the reason this matters: near the top of page one it's steep, so the same impressions produce several times the clicks at position 4 that they do at position 8. You're not building anything new. You're collecting on a decision Google already made.
Find them in Search Console in five minutes
Everything here is free, first-party data. Here's the exact recipe, yours to keep.
Open Search Console, then Performance, then Search results. Set the date range to at least three months (the 28-day default is too noisy to judge position stability). Turn on all four metrics: clicks, impressions, CTR, average position.
Add the position filter:
Average position → Greater than → 3
Average position → Smaller than → 16
If you want revenue-adjacent queries first (you do), add a query filter, Custom (regex), Matches regex:
(buy|purchase|price|pricing|cost|cheap|affordable|best|top|vs|versus|review|near me|hire|quote|for sale|discount|deal)
Sort by impressions, descending. Discard branded queries, navigational queries, and anything clearly informational that slipped through. What's left is your working list. A mid-size service site with a year of history typically surfaces 15 to 40 qualifying queries on the first pass.
One note on Search Console's newer AI filter box: it's a typing shortcut, not an analyst. Google's own documentation confirms it configures the same filters you'd set by hand and can misinterpret requests, so verify whatever it applies before you read the data. The regex is the source of truth.
Split the list before you touch a page
This is the step nearly every version of this tactic skips, and it's where the results come from. Two different problems hide inside a striking distance list, and they need different fixes.
High impressions at position 4 to 10 with low CTR is a snippet problem. Searchers see you and don't click. The page's relevance is fine; its title and meta description are losing the click battle. Rewriting them is the cheapest fix in SEO.
High impressions at position 10 to 15 is a relevance problem. The page is topically close but not explicitly matched. It needs the query answered on the page: usually a dedicated H2 section with a direct, first-sentence answer underneath it.
Match the intervention to the problem. Rewrite the title for snippet losers (or augment it with a differentiator: a year, a number, a specific benefit). Add a genuine H2 section for relevance losers. If the query has standalone demand the page can't genuinely serve, build a dedicated page instead of stretching this one. And one guardrail either way: one target query per page. Stuffing a title to chase three queries at once is how you lose all three.
Before any edit, check for cannibalization. If two of your pages trade positions for the same query, consolidate first. Optimizing one of them harder usually moves nothing and makes the cluster messier. Then, after the edit: two or three internal links pointed at the page with the query as natural anchor text, a recrawl request through URL Inspection, and a dated log entry so you can attribute whatever happens next.
The real odds
Here's the part the thirty-second versions of this tactic leave out. In our experience, roughly half of these edits don't move anything. Recrawl happens within days; movement, when it comes, shows in two to four weeks; and a good chunk of the time the answer is no. That's not failure. It's a portfolio of small bets with positive expected value, and it's why the deliverable framing we use with clients is "keywords you already rank for that we can lift," not "guaranteed wins."
Two more things this tactic won't do. It won't create demand: it harvests existing equity, and after two or three monthly passes the orchard is picked. And it won't help a brand-new site, because a domain inside its first year mostly doesn't have impressions to harvest. If that's you, skip this and go build the content backlog (our first 90 days plan covers that sequence). Come back when Search Console has something to tell you.
If the audit bug bites you while you're in there, our AI SEO checklist covers the full on-page floor, and the AI visibility audit walkthrough extends the same discipline to ChatGPT and AI Overviews.
Make it a loop, not an event
Run the filter monthly. New striking distance queries surface as the site earns impressions, and last month's edits either moved (roll the pattern forward) or didn't (revise or retire the bet). Log every change: query, page, position before, edit made, date, position at 30 days. The log is what turns this from a trick into a system, and it's the difference between knowing your win rate and guessing it.
We've packaged the whole working layer as a free Striking Distance Harvest Kit: the full regex pack (commercial, question, local, and ecommerce variants), the position filter recipe, the failure-mode cheat sheet, the tracking table we use with clients, and a lift-score formula for ranking the bets by expected click gain (scoring approach adapted, with credit, from Harry Sanders of Hawk Academy). It unlocks with an email and it's yours to run monthly, whether you ever talk to us or not.
This loop is also, not coincidentally, month one of the Reforge Method's Assay phase: measure what's broken, lift what's liftable, and let the fast wins buy patience for the slower work of authority. If you'd rather have us run it, that's the engagement, and the striking distance table is in your first monthly report.