Search Console signals that actually predict which article to write next
Your top-queries report tells you what already worked. The queries worth writing about are further down, ranking between eight and twenty on a page that was never written for them.
The most reliable signal in Google Search Console for "what should I write next" is a query that already brings you impressions, sits at an average position somewhere between eight and twenty, and lands on a page that was never written to answer it. It is reliable because all three parts are evidence: demand exists, Google already considers you relevant, and the gap between the query and the page is a specific thing you can fix by writing one article.
Most content plans start in the wrong report. Sorting queries by clicks shows you what already worked, which is comforting and almost useless for planning. Everything at the top of that list is a page you have already written, ranking where it is going to rank. The decisions live further down.
The four signals, and what each one means
1. Striking distance: position 8 to 20, impressions above your noise floor
A query at position twelve means Google has decided you are a plausible answer and put you on page two, where roughly nobody goes. That is not a ranking problem to be nudged; it usually means your page mentions the topic while something else answers it. A dedicated article that answers exactly that question is the intervention, and it is the highest-yield one on this list.
Ignore positions one to three — there is nothing to win. Ignore anything past forty; that is Google being polite, not Google being interested.
2. Query and page disagree
Filter a query, then look at which of your URLs it lands on. If the answer is your homepage, your services index, or a category page, the article for that query does not exist yet. Google is guessing on your behalf, and it is guessing with the best page it can find rather than the right one. This is the clearest "write this" signal in the tool, because the missing page is named by the query itself.
3. Impressions rising while clicks stay flat
Growing impressions with static clicks means demand is increasing and you are on the edge of it. It reads as a failure — the click-through rate is falling — but the underlying movement is a topic getting bigger. Treat it as an early warning that a subject is worth a proper page, not as something to fix with a new title tag.
4. High click-through rate on a query with low volume
Ten impressions and four clicks is not traffic; it is a precise statement that when people searching this exact thing see you, they choose you. It is the best evidence you have about which adjacent topics are worth writing, because it is behaviour rather than volume.
| Signal | What it looks like | What it justifies |
|---|---|---|
| Striking distance | Position 8–20, steady impressions | One article answering exactly that query |
| Query/page mismatch | Query lands on homepage or a category | The missing page, named by the query |
| Rising impressions, flat clicks | Impressions up 3×, clicks unchanged | A proper page before the topic peaks |
| High CTR, low volume | 10 impressions, 4 clicks | More on the same narrow subject |
Three traps that make the data lie
The last few days are incomplete. Search Console data finalises with a lag. Compare a 28-day window against the 28 days before it, not "last 7 days" against "this week so far" — the second comparison will show a decline every single time, because the newest days have not finished arriving.
Average position is an average. A query showing position 9.4 may be sitting at 4 on mobile and 15 on desktop, or at 3 in one country and 30 in another. Before you act on a striking-distance query, split it by device and country. Sometimes the article you were about to write has already been written and simply is not loading well on phones.
Filtered totals do not sum. Google samples and anonymises rare queries, so the sum of your query rows will not match your total clicks. This is normal and not a bug in your export. Use the rows to compare with each other, not to reconcile with the total.
A 30-minute version you can run monthly
- Set the date range to the last three months and compare it to the previous period.
- Open Performance → Queries, sort by impressions, and filter position to between 8 and 20.
- For each of the top ten rows, click the query and read the Pages tab. Note whether the landing page is a real answer or a stand-in.
- Every row where the landing page is a stand-in is a candidate article. The query is the title brief.
- Sort the candidates by impressions and write the top two or three. Not ten — the point is to publish the ones with evidence behind them, then look again next month.
Three months is a deliberate choice. A 28-day window on a small site puts most queries into single-digit impressions, where one person's week of searching looks like a trend.
What this method does not tell you
It only sees Google Search. Queries that never reach you, topics you have no page for at all, and everything happening inside AI assistants are outside this report by construction. Search Console is an excellent instrument for the demand you are already touching and blind to the demand you are not.
So treat it as one input. It is the one with receipts — every candidate article can be traced back to a row with impressions, a position and a landing page, which is more than most content calendars can say. Use something else to find topics you have never been visible for, and keep the two lists separate so you always know which decisions have evidence under them.
The discipline worth keeping is small: never put an article on the calendar without being able to name the row that justified it. If you cannot point to the query, the impressions and the page Google is currently using instead, you are not planning. You are guessing with a spreadsheet open.