Home / Building & SEO / Beyond Keyword Tools: Reverse-Engineering Topic Ideas from Search Console
Beyond Keyword Tools: Reverse-Engineering Topic Ideas from Search Console
Keyword tools tell you what others search for; Search Console tells you what you already have. The latter is worth more, and fewer people use it.
Keyword tools can tell you search volume, but they can't tell you that your site already ranks #18 on a given question. That second piece of information is the actionable one.
Three reports, look at them in order
- Performance report → Queries: which terms are already bringing in impressions
- Performance report → Pages: which pages get impressions but have a low CTR
- Indexing → Pages: which pages aren't indexed yet
The first is the most valuable: terms already ranking in positions 8–20. These terms show that Google considers your page relevant, just not good enough yet. Pushing them up one step is far easier than opening ten new topics.
How to do it
Export the query table and filter for high impressions, low clicks, average position 8–25, and you get an "almost there" list. Then do one thing for each term:
- If a matching article already exists: add a section that answers this question directly, and put the term in the heading
- If not: write a new one, using the query itself as the backbone of the title
The advantage of this method is that you don't have to guess. Google has already told you where it thinks you belong.
How to expand the long tail
A term sitting at position 12 usually comes with three to five variants. List the variants and see which ones have no corresponding content yet:
| Main term | Variant | Content exists |
|---|---|---|
| hysteria2 obfs | hysteria2 obfuscation config | Yes |
| mihomo obfs error | No → write | |
| salamander password | No → can be merged |
What not to do
- Don't force extra paragraphs in for a single term. Adding content is about answering the question, not stuffing in keywords
- Don't chase terms with an average position past 40. That means relevance isn't there, and pushing them is hard work
- Don't check every day. Once a week is enough; the data lags, and checking too often only makes you anxious
A suggested cadence
Set aside half an hour every week: pull the query table, pick two "almost there" terms, and spend an hour each revising or writing. Three months later, clicks on this batch will be noticeably higher than on articles written from scratch over the same period — because they already had a foundation.
Three actionable filters
- High impressions, low clicks, position 8–20: the title or description isn't compelling; changing the TDK is the easiest win.
- High clicks, position 3–8: the content is right but not deep enough — add subsections and tables, and push it further up.
- Low impressions but steady clicks: long-tail demand, worth a dedicated post.
Log every filtered result in one table: keyword, current position, impressions, clicks, planned action. Without that table, "keyword mining" is just browsing.
Time window
GSC data lags by 2–3 days and rolls over 16 months. Read trends over at least 28 days; don't draw conclusions from a single day's data.
One-line summary
The value of mining GSC isn't "finding new terms" — it's "knowing how far you are from page one."