Sorting a keyword export by search volume and picking the top rows is the fastest way to build a content plan that never converts. Volume tells you how many people search. It says nothing about whether those people want what you sell.
The four intents
Every commercial query falls into one of four buckets, and each deserves a different page.
- Informational. “What is a CRM.” The searcher is learning. Give them a guide, and expect a long path to purchase.
- Commercial investigation. “Best CRM for small business.” They are comparing. This is where comparison pages, alternatives pages, and honest reviews earn revenue.
- Transactional. “CRM pricing,” “buy CRM software.” They are ready. Send them to a product or pricing page, not a blog post.
- Navigational. “Salesforce login.” They want a specific destination. Unless it is your brand, skip it.
Build the map before the list
Start from your offer, not from a tool. Write down the problem your product solves, then list the ways a customer might describe that problem at each stage: before they know a solution category exists, while comparing vendors, and when ready to buy. That gives you a skeleton of maybe fifteen topics. Only then open a keyword tool, and use it to find the real phrasings and the long-tail variations around each topic.
Reading difficulty honestly
Third-party difficulty scores are estimates built mostly from backlink data. They are a starting filter, not a verdict. Before committing, open the results page and ask:
- Are the top results from sites comparable to yours, or are they all major publishers and government sites?
- Do any results look genuinely weak, such as forum threads, outdated posts, or pages that only partially answer the question?
- How much of the page is taken by ads, AI answers, and features that absorb the click before anyone reaches the organic results?
A keyword with 200 searches a month and three weak results is worth more than one with 20,000 searches where the first screen is entirely paid and aggregated answers.
Cluster, then assign
Dozens of keyword variations usually describe one topic. “How to reduce churn,” “reduce customer churn,” and “customer churn strategies” are one page, not three. Group variants by whether the results pages overlap: if the top ten results are largely the same, one page serves all of them. Splitting them creates two weaker pages competing with each other.
Score what you keep
For each cluster, rate three things on a simple scale: business value (does ranking here produce revenue), win probability (can we realistically compete), and effort. Sort by value first, probability second. The best first projects are usually mid-volume commercial-investigation terms where your product is a legitimate answer and the incumbents are thin.
Revisit quarterly
Results pages shift, competitors publish, and your own authority grows. A term you dismissed as unwinnable in January may be reachable in September. Keep the rejected list. It becomes your pipeline.