> For the complete documentation index, see [llms.txt](https://help.rankability.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.rankability.com/track/setting-up-keyword-tracking.md).

# Setting up keyword tracking

Create a Track report from what customers should find the client for, then choose platforms, queries, locations, and a monitoring cadence.

Create a Track report from what customers should find the client for, then choose the platforms, queries, locations, and monitoring cadence that answer the business question.

Before starting, confirm the client domain and brand, the target market, and which search surfaces matter. Tracking is a recurring measurement commitment, so avoid enabling platforms or locations you will not use.

## Create a report

Setup is a guided builder. Each page asks one thing, **Back** preserves what you already chose, and nothing is created until the final page.

1. Select a client, open **Track**, and choose **New project**.
2. **What do you want visibility for?** Enter the product, service, or category customers should find the client for. Use **Add more topics** to expand the field into a list; commas and line breaks both separate topics, and exact duplicates are dropped. Each topic becomes its own report with its own history, so a list creates several reports rather than one combined report.
3. **Where should Rankability check?** Choose the channels you want: AI answers, local results, classic search, and video.
4. Confirm each channel you chose. Every channel has a settings page followed by a confirmation page, so you review the platforms or coverage first and the wording second.
5. On **Confirm what we'll ask**, review the AI prompts. Rankability suggests up to ten per topic and selects five. The suggestions follow the wording of the topic you entered, so a topic about pricing, comparisons, reviews, or recovery produces questions about that angle rather than a generic list. Edit any prompt in place, add your own — pasting several lines at once works — and the count for each topic is shown as you go. Local, classic, and video search share one keyword list and one confirmation page, because they test the same customer phrase.
6. **How often should we check?** Choose **One time** for a single snapshot with no recurring checks, or **Recurring tracking** and then the cadence.
7. **Review your report settings.** This summarizes the business, each topic with its selected prompt count, the platforms, and the schedule, which reads **Once, starting now** or starts now and then repeats. Expand a topic to see its prompts, and use the edit control in any section to change it without leaving the page.
8. Select the create action, which names how many reports it will create.

Review your selected coverage and schedule before you create anything.

A completed setup without a completed scan does not yet contain visibility evidence.

## Choose platforms deliberately

Traditional, AI, local, and video platforms measure different kinds of visibility and have different workload impact. Location-aware AI platforms can multiply work across additional locations. Enable only the surfaces you intend to monitor.

Which AI platforms you can select depends on your base plan and any paid platform add-ons. Google Search, local, and video tracking are included on every paid plan. If an AI platform is locked, open **Billing** to add that platform or compare a plan that includes it. Access begins only after any required payment succeeds. See [Understanding billing and usage](/account-and-settings/understanding-billing-and-credits.md) for the enabled platforms and tracked prompt allowance on your subscription.

## Organize tracking

Use **Create cluster** on the project list to group related reports. A cluster is a way to organize reports, not a combined report: every report inside it keeps its own settings, schedule, scan state, and history.

Open a cluster's menu and choose **Add topics** to add work directly to it. Paste one topic or question per line — your exact wording is used, and each line becomes its own independent report. Repeated lines are skipped, as are topics already tracked with the same settings. Choose an existing report to take settings from, confirm the website, language, optional location, and schedule, then start; the action names how many topics it will add and scan. **Pick existing** moves reports you already have into the cluster instead.

A report's own menu can assign it to a cluster, move it to another, or remove it from one. Deleting a cluster leaves its reports in place as unclustered reports and keeps their history.

Turning auto-tracking off stops new scheduled scans for that keyword; you can still run a manual scan.

Use one persisted keyword record for each intended query/client scope. If you import or create in bulk, verify row-level language and location settings before starting scans. Similar keyword text can legitimately exist in different projects or markets, but accidental duplicates waste capacity and fragment history.

## Track a running scan

Each row shows its full topic or question rather than cutting it off, so a long AI prompt stays readable. You can select the text, and **Copy query** puts it on your clipboard without opening the report.

After you start a scan, Rankability returns you to the keyword list and shows the run's state on the project's row. A scan in progress shows **Processing**, and the list refreshes on its own while any run is still working — you do not need to keep a report page open.

A new report opens only after a scan has covered every platform you selected. Faster platforms can return before the rest, so a row can look busy for a while before it becomes available; that wait is the report being completed rather than a stalled run. Once a report has a completed scan it stays open, so adding a platform later does not take the existing report away while the new one is measured for the first time.

A run that does not finish shows **Stuck** or **Failed**, with the recorded reason available on the row. You can open a failed report to read a **Latest scan failed** notice with that reason, and it says so plainly when no results have been saved for the report yet; opening it does not start a new scan. When the run can be retried, select **Retry** on that row; the project returns to **Processing** and stays in the list. A scan that stalls is stopped automatically and recorded as failed rather than sitting on **Processing** indefinitely, so a row that stays busy for a long time is still working. You can also cancel a scan that is pending, running, or stuck; a scan that has already finished cannot be canceled.

## Read the first result

After a completed scan, review SPI and its category breakdown, then open the detailed traditional, AI answer, citation, local, or video results that apply to the project.

## Common issues

* **The usage impact is too high** — Reduce unnecessary platforms, locations, or grid points before approval. See [Supported tracking platforms](/track/supported-tracking-platforms.md).
* **No scan data appears** — Check that a run completed and that the selected platform, date, and location match it.
* **A platform says not tracked** — Enable it in the keyword settings and run a new scan; historical runs cannot contain a platform that was not selected.
* **The keyword already exists** — Open the existing tracked keyword and preserve its history instead of creating a duplicate.
* **Scheduled scans stopped** — Check auto-tracking, project state, connection health, subscription state, and any visible error before starting a manual replacement. A lapsed subscription stops automatic scans, and the skipped run records that reason on the row. Scheduled work is otherwise protected from on-demand usage limits.
* **The result differs from a manual search** — Location, device, personalization, time, and provider context can differ. Compare the persisted scan with matching settings.

## Related articles

* [Supported tracking platforms](/track/supported-tracking-platforms.md)
* [Understanding SPI](/track/understanding-spi.md)
* [Usage limits reference](/account-and-settings/credit-costs-reference.md)
* [Troubleshooting common issues](/troubleshooting/troubleshooting-common-issues.md)
* [Running benchmark reports](/track/running-benchmark-reports.md)