Data

Why numbers change

Every reason a Ranqo figure can move, even for a week that is over, from new runs and filters to prompt edits, competitor names and failed queries.

A figure you read last week can read differently today, even for a week that is over. Most of the time a new run explains it. This page lists every reason a number can move, so you can tell a real change from a change in what is being counted.

A new run landed

Each brand runs once every 7 days. When a new run finishes, it enters your date range and the oldest run may leave it, so every range-based figure moves.

Different figures also compare different things:

Visibility page cards compare your range with the range of the same length before it, as a percentage of the earlier value.
Home's visibility chip compares the latest run with the most recent run that finished more than a week before today, or with your oldest run when none did, in points. It reads "vs last week" whatever the gap, and it does not appear once the latest run is itself more than a week old.
The Action Center's Visibility card compares the latest run with the one before.
Source Trends compares the latest run's citations with the previous run's.

Two changes on two pages can disagree without either being wrong. See One point per run.

Answers vary from run to run

AI engines do not give the same answer twice. A prompt that named you last week may not name you this week, with nothing changed on your side. A few points of movement between two runs is usually this. See Noise bands.

Long answers

Before an answer is read, Ranqo cleans it and reformats it into a consistent shape, and the reformatting sees only the first 8,000 characters of the cleaned answer. A brand or competitor named only after that point is not recorded. When the reformatted version cannot be used, Ranqo reads the cleaned answer instead, up to its first 12,000 characters.

The answer you open from a run or a prompt's history shows the reformatted version when there is one, so for a long answer it can end before the part where you were named.

The date range and your time zone

A carried value. When your range holds no run, the page shows the last completed run marked As of its date. Pick a range that holds a run and the figures change to that run's. See Carried values.
Your time zone. Ranges start at midnight where you are, so a run near the edge of a range can be inside it for you and outside it for a colleague elsewhere. See Time zones.

Filters

Every filter narrows the answers a figure is built from: engines, locations, categories, themes and Branded or Unbranded. Share of Voice changes with them too, because the competitors in the narrowed answers change.

Filters read each prompt as it is today. If you change a prompt's category or location, its past answers move to the new group in every filtered view, including for runs before the edit.

Changes to your prompts and engines

Adding prompts or engines. Runs after the change ask a different set, so figures from before and after cover different questions. The Changes button on the Visibility trend chart marks each run where prompts or engines were added or removed, under Tracking Updated.
Pausing a prompt. It keeps its history and sits out later runs, so a range that spans the pause mixes runs with it and runs without it.
Deleting a prompt. Its answers are deleted from every run, so past figures worked out from answers, such as those for a date range, change. To stop asking a prompt and keep its history, pause it instead. See Prompts.
Deleting a theme. Its prompts move to No Theme with their history, except the ones you tick to delete with it. Deleting prompts that have been measured removes their answers from past runs as well. See Themes.
Editing a prompt's wording. The prompt keeps its history, so runs before the edit answered the old wording.
Changing the engines a brand tracks. Tracked prompts that use the brand's engines switch to the new set; prompts with their own engines keep them. Visibility is weighted by engine over the engines that answered, so adding or removing one changes how the rest are weighted. See Visibility and Brand settings.

Competitor names

Ranqo groups the different spellings of one competitor under one name. When a new spelling is grouped, or when you separate a spelling or group it back, competitor rankings and Share of Voice are recalculated. Groupings that are applied when a page loads change past figures too. See Competitor identity.

Failed queries

When an engine fails to answer a query, Ranqo either reuses a recent answer to the same prompt on the same engine, or the query has no answer. Pages count these differently: a run's own scores count the reused answer and leave the failed query out, the Visibility page's cards leave out both, and the Prompts page's Tracking tab and a prompt's insights page count a failed query as an answer that does not name you. A run with failures can therefore read differently from one page to another. See When a query fails.

Noise bands

How much a figure moves between runs when nothing changed.

Score dropped?

A checklist for a visibility drop you did not expect.

A new run landed

Each brand runs once every 7 days. When a new run finishes, it enters your date range and the oldest run may leave it, so every range-based figure moves.

Different figures also compare different things:

Visibility page cards compare your range with the range of the same length before it, as a percentage of the earlier value.

Home's visibility chip compares the latest run with the most recent run that finished more than a week before today, or with your oldest run when none did, in points. It reads "vs last week" whatever the gap, and it does not appear once the latest run is itself more than a week old.

The Action Center's Visibility card compares the latest run with the one before.

Source Trends compares the latest run's citations with the previous run's.

Two changes on two pages can disagree without either being wrong. See One point per run.

Long answers

The answer you open from a run or a prompt's history shows the reformatted version when there is one, so for a long answer it can end before the part where you were named.

The date range and your time zone

A carried value. When your range holds no run, the page shows the last completed run marked As of its date. Pick a range that holds a run and the figures change to that run's. See Carried values.

Your time zone. Ranges start at midnight where you are, so a run near the edge of a range can be inside it for you and outside it for a colleague elsewhere. See Time zones.

Filters

Filters read each prompt as it is today. If you change a prompt's category or location, its past answers move to the new group in every filtered view, including for runs before the edit.

Changes to your prompts and engines

Adding prompts or engines. Runs after the change ask a different set, so figures from before and after cover different questions. The Changes button on the Visibility trend chart marks each run where prompts or engines were added or removed, under Tracking Updated.

Pausing a prompt. It keeps its history and sits out later runs, so a range that spans the pause mixes runs with it and runs without it.

Deleting a prompt. Its answers are deleted from every run, so past figures worked out from answers, such as those for a date range, change. To stop asking a prompt and keep its history, pause it instead. See Prompts.

Deleting a theme. Its prompts move to No Theme with their history, except the ones you tick to delete with it. Deleting prompts that have been measured removes their answers from past runs as well. See Themes.

Editing a prompt's wording. The prompt keeps its history, so runs before the edit answered the old wording.

Changing the engines a brand tracks. Tracked prompts that use the brand's engines switch to the new set; prompts with their own engines keep them. Visibility is weighted by engine over the engines that answered, so adding or removing one changes how the rest are weighted. See Visibility and Brand settings.

Competitor names

Failed queries