Keyword research without an expensive subscription

Keyword research has an image problem: it sounds like something you first need a 130-dollar-a-month subscription for. In practice, every good piece of research starts without a tool, with the question of what your customer types in, and increasingly what your customer asks an AI assistant. This is the approach I use myself, from free sources to the point where paid data does make sense.

Step 1: collect questions, not words

Don't start in a tool, start with your customer. What questions do people ask in your inbox, on the phone, in reviews? Type your main product or service into Google and note the suggestions and the "People also ask" boxes. Ask ChatGPT the same question and see which follow-up questions it suggests. That is literally how your audience searches these days, and those questions are gold for your content plan.

If you already have traffic, Google Search Console is your richest free source: under Performance you can see which searches you are already being shown for, including the terms where you are hovering on page two. That is where the quickest wins are.

Step 2: choose one term per page (short-tail vs. long-tail)

The classic mistake is collecting a hundred terms and wanting the homepage to rank for all of them. Work the other way round: every page gets exactly one target term, and the type of term decides the type of page.

Short-tail ("garden set", "seo software") has volume and competition: those are your category and overview pages. Long-tail ("which garden set can stay outside in winter") has less volume but much higher intent: those are your articles. Two pages on the same term make them compete against each other instead of against everyone else, so keep the list unique per page and note for each term which page claims it.

Step 3: when paid data does make sense

Volumes and competition figures only become useful when you have to choose: which of these twenty topics do I write first? Then you want to know that term A gets 1,600 searches a month and term B ten. You no longer need a separate subscription for that measuring side, though: suites such as Semrush and Ahrefs are built for specialists, and for your own site you mostly end up paying for depth you don't use.

At Neuramerce the keyword workbench is built into the platform: real search volumes and cost-per-click figures (from the same data sources the pro tools use), linked to the page that claims the term, so research turns straight into a content plan instead of a spreadsheet.

From list to plan

The end product of keyword research isn't a spreadsheet but a content plan: one page per term, one clear intent per page, and an order based on expected return. Start with the terms where you are already on page two (Search Console), then the questions your customers literally ask, and only after that the big short-tail terms everyone is fighting over.

And don't forget the new question side: people increasingly ask their questions to AI assistants, in whole sentences. How to write for that is in the article on AI SEO, and you can get the basics right today with the twelve SEO tips.