Trackers

Use Trackers to see whether the numbers are moving, and whether your work moved them.

Trackers overview

Every other analytics screen answers one question: what does the latest scan say. Trackers answers the other one — what changed since the last scan, and is it going the right way.

Three tabs, one per level:

• AI visibility — your brand and your competitors inside AI answers, scan over scan.
• Site — your own pages, from AI citations through Google search to real visits, day by day.
• Sources — what the assistants read to build their answers.

The tab you left on is the tab you come back to.

How to read a dynamics chart

The charts on the AI visibility and Sources tabs are drawn scan by scan, not day by day. A dot marks every scan, so two scans on the same day are two points — the second one also shows the time, since the date alone no longer tells them apart.

The scale is fixed: visibility and shares always run 0–100%, average position runs 1–10 with the axis flipped because lower is better, and sentiment uses the same five faces as the Sentiment screen. Nothing rescales when you change a filter, so a line that didn’t move looks like a line that didn’t move.

Hover a point for the values at that scan, best first, each with its change from the previous scan and the number of runs behind it.

Charts draw six lines at most — you and the strongest others. The legend folds the rest behind a +N more chip, and every line stays switchable from there. With dozens of competitors or hundreds of cited domains, the tables below the charts are the complete list; the charts stay readable. Every chart works the same way. In the legend, one click hides or shows a line and a double click leaves only that one — double-click it again to get the rest back. Clicking the Y axis itself switches it between the full 0–100 and a range fitted to the lines on screen, which is what you want when everything sits in the bottom few percent; that choice is remembered per metric on this device.

Two marks carry meaning:

• A shaded stretch between two scans — the set of tracked prompts changed in between: prompts added, removed, or moved to another topic. Both readings are still true on their own, but the difference between them mixes a real change with a changed measurement. The tooltip says what changed.
• A dashed vertical line — the first scan after you marked something published. That’s the earliest point where your work could show up.

The filters

Above everything sit the two controls that scope the whole tab: the topic filter, and how many scans to show — 5, 15, 30 or 60. At a weekly scan that’s roughly a month, a quarter, half a year, a year. Both apply to everything below them, the headline numbers included, and the scan depth follows you from tab to tab.

By topic is the one chart the topic filter doesn’t narrow. It exists to compare topics against each other, so it keeps every line and highlights the one you picked.

AI visibility tab

Four headline numbers first — visibility, average position, sentiment, and your gap to the strongest competitor — each with its move since the previous scan.

Then the same metric in three cuts, one chart each:

• You vs competitors — who is gaining and who is losing.
• By topic — which topics you are growing in.
• By assistant — where the visibility is actually moving.

Under the charts, two tables name it: topics with their change and who leads each one, and every tracked brand with its change. Click a topic to scope the whole tab to it; click a brand to open it.

Sentiment works in all three cuts, competitors included. Each phrase in an answer is scored for the brand it names, so in a comparison two brands no longer inherit each other’s verdict — and it is the same number the Sentiment screen reports.

Site tab

This tab works on its own with the AI citation data Supasaito already collects. Connecting Google Search Console and Google Analytics adds the search and traffic columns — see Integrations. Pages you publish from a content recommendation are marked, so you can tell your deliberate work from the rest of the site.

Four groups sit at the top:

• AI scans — how many of your pages were cited in AI answers, against the total. History builds with each scan; the first scan is marked on the line.
• On your site · visits — daily visits over the period.
• On your site · key events — conversions, with how many came from AI referrals.
• In Google search — clicks against impressions.

Where a source isn’t connected, the tile says so instead of showing zero.

Rows in the table are your pages, grouped by section of the site, with columns for Cited in AI, Visits, Key events, Index, Impressions and Clicks. Each row carries a short verdict that’s more useful than any single column:

• Cited + traffic — working.
• Cited, no traffic — the engines use it, people don’t arrive. Usually a positioning problem, not a technical one.
• AI visits, not cited — people arrive from assistants but the page isn’t a cited source.
• Not indexed — Google doesn’t have it, so a whole channel is missing.

Filter with All pages, Cited in AI or Needs attention, and switch the period with the day control. Open a row for the per-page detail, including confirmed AI referrals.

Good to know AI traffic is undercounted: many AI-assistant visits arrive with no referrer, so nothing marks them as coming from an assistant. Treat the AI referral numbers as a floor rather than a total. Where the data supports it, a page’s detail also shows an estimate of the unattributed lift — how much direct traffic rose after its first AI citation, with the method behind it in a tooltip.

Sources tab

Three charts, each showing the total and the per-assistant split on the same axis: how many distinct domains were cited at each scan, how much of that was your own site, and how much belonged to a tracked competitor. Different assistants read different corners of the web, which is why the split is on the chart rather than behind a switch.

Below them the domains themselves, with their citations at each of the last scans and a new or gone badge — the two facts worth acting on. Filter by ours, competitors or other. The external URLs you track by hand sit at the bottom of this tab.

How to see whether a publication worked

1. Publish the piece and record its URL on the recommendation — see Content briefs.
2. The page enters your tracked set and appears on the Site tab; the scan that follows gets a dashed line on the dynamics charts.
3. Open the page’s detail and read From publication to effect, which lays the stages out in the order they actually happen: Published → Indexed by Google → First impressions → Cited in AI.

Citations are the slowest stage. A page that’s indexed and getting impressions but isn’t cited yet is progressing normally, not failing.

Good to know A percentage built on a handful of runs swings on its own. The run count is in every tooltip — read it before treating a jump as a result.

On the free plan the AI visibility and Sources tabs show a blurred preview of what a paid plan puts there. With one scan and none scheduled after it there is no second point to compare against, so the figures behind the blur are an example, not yours. Regular scanning is what a paid plan adds.