On topic selection, I started with “going with my gut” — writing whatever topic felt good, then finding no traffic after publishing, wasted effort. Later I switched to data-driven selection: finding demand from four sources — GSC queries, on-site search, competitor content, and industry data. Topics got noticeably sharper, and the chance of a “dud” article clearly dropped.
Why data-driven topics are more accurate
Gut-based topics rely on guessing; data-based topics rely on demand. GSC queries are words users actually search; on-site search is words users can’t find on your site; competitor data is directions competitors have already validated. Use the four sources this way: from GSC queries, take words with impressions but no clicks and fill content; from on-site search, take words users can’t find and fill pages; from competitor content, take topics with high traffic and follow up with differentiation; from industry data, take rising-trend topics and position early.
The topic selection flow
Step one, collect: export candidate topics from the four data sources. Step two, filter: sort by search volume, competition, and commercial value. Step three, verify: search to confirm intent and competitive format. Step four, schedule: put them into the topic library by priority. When evaluating, score each item 1 to 5 on four dimensions: search demand, competition, commercial value, resource fit; the highest total goes first.
How to mine each of the four sources
| Data source | Signal meaning | How to use |
|---|---|---|
| GSC queries | Words users really search | For impressions without clicks, fill content |
| On-site search | Words not found on your site | Fill corresponding pages |
| Competitor content | Directions competitors validated | Follow up with differentiation |
| Industry data | Rising-trend topics | Position early |
Consolidate candidates into one table
Don’t rush to write the words mined from four sources; consolidate them into a candidate table with uniform fields: search volume, competition, commercial value, can we do it well. Then sort by total score and write the highest first, avoiding “write whichever catches the eye.”
Don’t skip the verification step
Before scheduling, search each topic once to confirm two things: what users actually want (intent), and what formats currently rank up front (competitors). Get intent wrong and nobody reads what you write; get format wrong — say they want video and you write a long article — also wasted. This ten-minute step blocks half the duds.
The topic library needs dynamic maintenance
The topic library isn’t built once and done. Monthly, add newly mined words, mark written ones as “done,” mark poorly performing ones as “abandoned with reason.” After three months, this library is your most valuable asset: it knows what users want, and it knows what’s useless to write.
An on-site search example
Once I looked at the on-site search log and found many people searching “how to export reports,” but the site had no such content. After adding a tutorial, it not only caught on-site search but also brought organic traffic from external shares. On-site search is the demand closest to you; users have already used search to tell you what’s missing — just fill it in.
Data-driven topics fail sometimes too
However good the data, there’s a blind spot: it only reflects “already searched” demand, and can’t give you opportunities “users haven’t realized yet.” So don’t hand all topics to data; keep a 10-20% allowance for intuition and trend judgment — like an industry trend just starting, with small search volume but worth an early position. Data handles “what’s wanted now”; judgment handles “what will be wanted.”
That unwritten 10-20% allowance is also where in-house writers get room to shine. A topic library fully hostage to data gets more and more alike; leaving a little room can produce differentiated content that catches the early train. Writers’ feel for the industry is exactly what data can’t give but can fill data’s blind spot. Feel plus data, walking on two legs, makes topics both accurate and forward-looking.
