Keyword research
Discovering keywords in RankX AI, what each figure means, how the cache makes repeats free, and how discovery feeds rank tracking and topic clusters.
Keyword research in RankX AI takes up to twenty seed terms and returns related keywords with search volume, cost per click, competition and trend. Results feed straight into rank tracking and into topic clusters.
It spends credits only for a fresh discovery. Repeating a recent search inside the cache window is free.
Running a discovery
Give it seeds: the terms your buyers would actually type. Two or three good seeds beat twenty vague ones, because the expansion is only as good as what it expands from.
A cache hit is free by construction. If the same seeds were researched recently, RankX AI returns the stored result without spending. That makes it safe to re-open a research session without worrying about the cost.
If it reports that it is still running, call it again with the same seeds. The operation is self-converging and safe to retry: a retry does not spend twice.
What the figures mean
| Figure | What it is | What it is not |
|---|---|---|
| Search volume | An estimate of monthly searches for that term | A promise of traffic. It is a modelled figure from a third-party source |
| Cost per click | What advertisers pay for the term | A measure of how hard it is to rank |
| Competition | Advertiser competition for the term | Organic difficulty. A term can be cheap to advertise on and hard to rank for |
| Trend | Direction of interest over time | A forecast |
The most common misreading is treating competition as organic difficulty. It is an advertising signal, and the two diverge most in exactly the categories where it matters.
Search trends
A separate, deeper look at up to five keywords: interest over time, momentum, and the related queries rising fastest around them.
It is dispatched and polled. A fresh analysis returns a task id and completes in the background, and reading the result costs nothing. A repeat inside the cache window returns immediately and free.
Related queries only come back for a single-keyword request. That is a property of the source, and it makes single-keyword trend requests worth running separately when the rising-queries list is what you are after.
Trends is the better tool for "is this worth building around" and the worse tool for "what else should we target". Discovery is the reverse.
Turning research into tracking
Saving keywords starts rank tracking for them, up to twenty at a time.
Per-keyword results. A duplicate, or a keyword that would exceed your plan's tracked-keyword cap, is reported on its own row and never fails the batch. So a bulk save of twenty with two duplicates saves eighteen and tells you about the two, rather than refusing all twenty.
The cap is enforced in the database, not in the interface, so a bulk import over the limit is refused rather than silently trimmed. Figures are in plans and limits.
Choose fewer than you can. Every tracked keyword is a recurring cost multiplied by cadence, and a tracked keyword you would not act on is a row you scroll past. See rank tracking for a workable shape.
Linking keywords to clusters
A tracked keyword can be attached to a topic cluster, which is what joins your research to your measurement and your content. Attaching is idempotent: an existing link is never modified, so a retry cannot demote a link you already have.
Clusters are worth doing at the point of saving rather than later. An unclustered keyword is a keyword no brief, prompt or article will ever find.
Research for AI visibility, not just for search
Keyword research in RankX AI feeds two different things, and they want different inputs:
For rank tracking, you want the queries people type into a search box: short, commercial, and phrased as a search.
For AI visibility prompts, you want the questions people ask an assistant: longer, conversational, and constrained by segment, geography or use case. A keyword pasted into a prompt panel measures very little, because an assistant asked a keyword gives a vague answer that names nobody in particular.
Use the research to find the topics; write the prompts as questions. See AI Prompts.
Driving this from an assistant
research_keywords for discovery, get_keyword_trends and
get_keyword_trends_status for trends, save_keywords to start tracking, and
link_keyword_to_cluster to file them.
"Research keywords around 'ai visibility tracking'. Show me the ten with the best volume against competition. Do not save anything yet."
See the tool reference.
Where to go next
- Topic clusters, for the structure everything hangs from.
- Content briefs, for turning a keyword into a piece of work.
- Rank tracking, for measuring the result.
Issue reference
Every check the RankX AI Website Audit runs, with its severity, its weight, what it means and which audit paths can produce a verdict for it.
Topic clusters
The single structure that keywords, tracked prompts and content all hang from in RankX AI. What a good cluster looks like, and why it is worth getting roughly right.