Rank tracking

Rank Tracking is RankX AI checking your keywords' positions in ordinary Google results, to depth 20, on a cadence you choose within your plan's floor. Each check also captures Google's AI Overview for that query if one rendered, so the tracked keyword set is what widens both surfaces.

Adding keywords

Add them individually, in bulk, or from keyword research. Each keyword carries four settings:

The phrase itself cannot be changed after it is added. To change it, untrack the old one and add the new phrase. That is not a limitation to work around: a keyword's history is a series of measurements of one query, and editing the query underneath it would silently invalidate the series.

Plans cap tracked keywords per Website, enforced in the database, so a bulk import over the limit is refused rather than trimmed. Figures are in plans and limits.

The three position states

Never merged, and the distinction is the point:

RankX AI also reports why a scheduled check was skipped when one was, which is usually one of: the cadence had not elapsed, the wallet was short, or the schedule was paused.

Cadence, and what it costs

A rank check is charged per keyword, per check, and the AI Overview capture is a speculative hold on top of it that is released in full when no overview rendered. So a tracked keyword settles at one of two prices depending on whether Google showed a summary, never in between.

The arithmetic that matters:

keywords x checks per month

Two floors apply and the higher wins: your plan's floor, and 7 days while on trial, on every plan. Choosing a slower cadence than the floor is always allowed; the floor is a lower bound on the interval rather than a target.

Prices are in the credit cost reference.

Untracking is reversible

Untrack is a soft stop: the keyword stops being checked, its slot is freed against your plan limit, and its history is kept. Re-adding the same phrase resumes tracking with the old series intact.

There is no hard delete, deliberately. Nothing in RankX AI destroys a measurement to free a slot.

Choosing what to track

The temptation is to track everything you rank for. Two reasons not to:

Every tracked keyword is a recurring cost, multiplied by cadence. A large set on a daily cadence is a bigger line than most people expect.

A tracked keyword you would not act on is a row you scroll past. The useful set is the queries where a position change would change what you do this month.

A reasonable shape: your commercial head terms, the comparison and alternative queries in your category, the ten or so long-tail queries you are closest to winning, and the queries where you already rank and cannot afford to slip.

Reading movement

Compare against the check before it, not against yesterday. On a three-day or seven-day cadence, consecutive days are the same measurement.

Position improves as the number falls. Obvious, and still the source of most misworded reports.

A keyword that leaves the top 20 has not gone to zero. It has gone to "not in the top 20", which is a bounded statement: the check looks twenty deep and no further.

Read a drop alongside the site timeline and Search Console. A position change with a matching site change is a story; a position change alone is a number.

Driving this from an assistant

get_rank_tracking reads positions, save_keywords adds up to twenty at a time with per-keyword results, update_keyword changes settings, untrack_keyword stops one, and run_rank_check runs one now. See the tool reference.

"Show me tracked keywords for my main website. Separate the ones never checked from the ones outside the top 20, and tell me which five moved most since the previous check."

Where to go next

• AI Overviews, captured on these same keywords.
• Keyword research, for finding what to track.
• Search Console, for what Google reports rather than what RankX AI measures.