Most keyword research advice assumes you enjoy spreadsheets. You don’t, and you shouldn’t have to. The goal isn’t a list of 400 keywords with search volumes. It’s a short list of topics you can win that map to something a customer would actually type before buying from you.
This playbook is the exact process a RankMill research agent runs. You can do it by hand in an afternoon, or let the agent do it in ninety seconds. Either way, the thinking is the same.
Start with intent, not volume
A keyword with 50 searches a month from people ready to buy beats one with 5,000 searches from people writing a school report. Before you look at a single number, sort every idea into one of three buckets: informational, comparison, and transactional.
Volume tells you how many people are looking. Intent tells you whether any of them are looking for you.
The three buckets, quickly
- Informational: “how to store coffee beans.” Builds trust, ranks slowly, rarely converts on the first visit.
- Comparison: “V60 vs Chemex.” High intent, your product can appear as the answer.
- Transactional: “buy single-origin espresso beans.” Lowest volume, highest value.
Map topics to what you sell
Take your three or four best-selling products and write down the questions someone asks in the week before they buy. Those questions are your first ten posts. This is the step tools skip and the step that matters most.
A coffee roaster selling light-roast beans should own “pour-over ratios for light roasts” long before it chases “best coffee 2026.” The first is winnable and relevant; the second is a money pit.
Score each idea in sixty seconds
You don’t need a tool for this. Rate each topic 1–5 on three axes and add them up. Anything scoring 11 or higher goes on the calendar.
| Topic | Relevance | Winnability | Intent | Total |
|---|---|---|---|---|
| Pour-over ratios for light roasts | 5 | 4 | 4 | 13 |
| Storing beans for freshness | 4 | 5 | 3 | 12 |
| Best coffee 2026 | 2 | 1 | 3 | 6 |
Let an agent keep the list fresh
Do this once and you have a quarter of content. The hard part isn’t the first list, it’s keeping it current as competitors publish and rankings shift. That’s the part worth automating.
A RankMill research agent re-runs this scoring on a schedule, watches what competitors ship, and surfaces new high-scoring topics as they appear. You still approve every one. Here is the shape of what it returns:
# topic proposal from research agent
topic: "Pour-over ratios for light roasts"
intent: comparison
score: 13 # relevance 5 · winnable 4 · intent 4
gap: "no competitor covers roast-level ratios"
The one rule that survives every update
Search engines keep changing. The thing that never stops working: be the most genuinely useful answer to a question a real person is asking. Everything in this playbook is just a way to find those questions faster.
Start with ten. Publish them well. Then let the loop keep the list alive.