Period comparison

The period comparison puts two freely chosen periods side by side, period A and period B. For each row you see the value of both periods and the change. This is useful, for example, for a follow-up measurement after a change or for the monthly report to a client.

The comparison works with responses that have already been measured. It does not start a new query and costs no credits.

Where to find it

Open your AI-Visibility project and click Period comparison in the page header. You get back with Back to project.

Choose the periods

For both periods you set From and To. The defaults are the last two measurements: the day of the last run against the day of the measurement before it. If there are not yet two runs, the last 7 days are compared with the 7 days before.

The dates are part of the address bar. You can therefore bookmark the page or send the link to colleagues who are allowed to see the project.

These are rejected:

• periods that overlap. Choose two periods without a shared day.
• a period that ends before it begins
• a period longer than 366 days

Comparison basis

Under Comparison basis you choose which responses count:

• All responses: everything measured in the respective period. This is the default.
• Same basis (unchanged questions only): only questions that were not changed up to the end of the later period, for both periods. If you rewrote or replaced prompts in between, this is the fairer choice.

What is compared

The table has four sections:

If you have not added any competitors, the last section is called Frequently mentioned brands and shows the brands that appear most often in the responses.

For each row you see:

• the value in period A and in period B
• the Delta: B minus A, for rates in percentage points (pp)
• Count A / B and Count delta: how often something occurred in absolute terms, for example mentions
• Sample A / B: how many responses are behind the value. For Avg. position and Sentiment it is the number of mentions of your brand that are averaged

Notices above the table

• Periods of different length: only the percentage points are comparable then. A count delta is not shown.
• Different frequency: if one period was measured daily and the other weekly, for example, a notice points this out. Pay attention to the sample per row in that case.
• Empty period: if there are no responses in a period, it shows »not collected«, never 0 %.
• Citations not measurable: for periods in which a platform did not yet deliver its citations in a measurable form, »not measurable« is shown instead of a zero. The row of that platform is what counts then.

Measures in the comparison

If you have recorded measures, they appear in the Measures in these periods section: with their position (in period A, in period B, between A and B) and the change afterwards.

Export

With Export CSV or Export Excel you download the table, with the same rows and the status of each row. The CSV file uses the semicolon as separator and the comma as decimal separator.

Over the API

GET /v1/tracking-projects/{id}/period-comparison returns the same comparison, with the four dates and the basis as parameters. If the periods overlap, the endpoint responds with 422. The response also contains the measures in the periods.

Read more

• Record measures and measure the change
• AI Visibility Tracking: the module at a glance