Trends and history

The Tracking page in the sidebar is where you explore movement over time. It charts your visibility (how often AI answers mention your brand) across a date range you choose, so you can see whether your AI presence is improving, slipping or holding steady. The dashboard and Competitors pages show a visibility trend too; Tracking is where you choose what to compare. An old link to a Trends page opens Tracking's Changes tab.

Tracking has two tabs, Explore and Changes. Both use the same date range: a preset, from the last week to the last year, or your own start and end dates, up to a year long and ending today or earlier.

Explore: visibility over time

Visibility over time is a line chart. Compare, in the bar at the top, decides what each line is:

Compare	Each line is
Competitors	A brand: yours, which is always included, and the competitors you pick
Models	ChatGPT, Claude, Perplexity or Gemini
Terms	One of your prompts
Categories	A prompt category
Intents	A prompt intent

The chart shows up to eight lines at once. Each line has a card above the chart with its latest value and how far it has moved since its first value in the range. Click a card to hide or show its line, and use the Choose menu beside the cards to pick which lines are drawn.

Filter, in the same bar, narrows the responses by term, category, model or intent. Each filter in force shows as a chip you can remove, and Clear removes them all. When you compare competitors, every brand is measured on the same filtered responses.

Daily and Weekly switch the points. A weekly point averages that week's responses, and a partial week at either end only counts the days inside your range.

A shaded stretch of the chart, or a dotted stretch of a line, means there were no responses to measure there; a point at zero means there were responses, but none mentioned the brand. Hide days without data closes those stretches up. The pill above the chart says how much of your range has data and how many responses it rests on; click it for the full measurement note.

The Breakdown table under the chart gives each line's first and latest value, the change from first to last, its average and how many responses it's based on. Click a column heading to sort.

Tracking charts visibility only. Average position is on the dashboard scorecards and tone is on the Sentiment page; neither is charted over time.

Changes: what moved between two periods

Changes splits your date range into two equal periods, back to back, and compares visibility in the earlier one with the later one. Break the changes down by Categories, Terms or Models, and filter by prompt tags if you use them.

Each row shows the earlier and later visibility, the change in percentage points, and how many responses and matched prompt-and-model pairs it rests on. A row only appears when the same prompt was asked of the same model in both periods, so the comparison is like for like. It lists up to 20 rows, biggest moves first. The date range needs at least two days; if it has an odd number of days, the oldest day is left out so the two periods match.

On either tab, Copy link gives you a link to the exact view on screen, and Save this view… in the Views menu keeps it in your saved views, which only you can see.

How to read trend data

A few patterns to look for:

Steady climb across all models

Usually means your third-party signals are strengthening — citations accruing, content getting indexed, reputation compounding. Keep doing what you're doing.

Climb on one model, flat elsewhere

Often a sign that one model picked up a new piece of content (a press hit, a comparison post) that the others haven't yet. Worth investigating which content is driving it.

Sudden dip

Look for:

A scan that didn't complete (check Scans page).
A change you made to prompts (the new set may be testing different things).
A competitor that surged — they may be pushing you down.
A negative news story (check the Sentiment page for the same dates).

Volatile, jumpy lines

Common with small prompt sets. Say you're tracking 20 prompts: a single missed mention swings the average on its own. Add more prompts to stabilise the line.

How long is history kept?

The same on every plan: we keep all of your scan history for as long as your workspace exists, and the dashboard shows up to 365 days at a time.

Comparing periods

The Changes tab does this directly: it compares the two halves of your date range and lists the categories, prompts or models that moved most, with the change in points.

To eyeball two periods that aren't back to back, set the date range to cover both and read the Explore chart.