If you configure tracking or explain its reports, you need to know how a scheduled item becomes a collected result and a usage charge. A tracked prompt is a question sent to an AI assistant on a schedule you set. A tracked keyword is a search performed on a search engine the same way. This page describes where they are sent, when they run, and how usage is counted.
Where each one goes
A tracked prompt can go to three providers: ChatGPT, Gemini and Google AI Mode. You choose which of them each prompt is sent to. It is sent word for word, with no campaign text, persona or other context added to it.
A tracked keyword goes to two search engines. Google returns the organic results, the ads and the AI Overview in one request. Bing returns its organic results and ads.
Google AI Mode is tracked as a prompt because it returns an answer and cited sources without organic results. Its input field expects a question rather than a keyword.
Claude and Perplexity are not available for tracking and cannot be added on request.
When collection runs
You choose the days you want fresh data on, using seven checkboxes, one per day of the week. You never choose a time.
Collection runs the night before each ticked day. A tick on Monday is a Sunday night run, so the day's data is in place before the working day starts.
The reason for running the night before is that reports are read in the morning. A collection starting at breakfast would not be finished when the report is opened, because a prompt takes appreciably longer to collect than a search.
How usage is counted
One credit is one call: one tracked item, on one provider, on one day. The arithmetic is a multiplication:
providers chosen × days ticked = credits per week for that item
Two worked examples using the same weekly allowance:
- 1,000 prompts on all three AI providers, two days a week, is 6,000 credits. Adding 1,000 keywords on both search engines, two days a week, is another 4,000, for 10,000 in the week.
- 200 prompts on three providers, five days a week, is 3,000 credits. The same 200 prompts checked once a week is 600.
Because frequency multiplies, it is usually frequency rather than the number of tracked items that determines whether an allowance is enough. A large prompt set checked weekly costs less than a small one checked daily.
There is no separate quota for prompts and keywords. They draw on one weekly allowance, and you decide how to divide it.
Each scheduled prompt/provider combination produces one execution and one sample; results are not averaged across repeated runs. Credits are checked when the weekly assignment is saved rather than spent call by call, so a failed collection has no separate charge or refund.
What a tracked item belongs to
A tracked prompt or keyword is an account-level object. It is not owned by a campaign. It can be associated with any number of campaigns, or with none.
Brand and competitor names come from the campaign, and the labels are applied when the report is drawn rather than when the answer is collected. The practical effect is that two campaigns can label the same stored answer differently, and adding a competitor to a campaign changes how existing history reads without any new collection.
Items and campaigns can each be active or archived. Archiving stops collection and keeps the history.
What is extracted from an answer
The collected result already carries the answer text, the sources shown, the web results used, the follow-up queries the service reported, and the brand entities it identified. Nothing re-reads the answer afterward to find brand names.
Each report shows what the service returned for that execution. It does not reinterpret the answer afterward.
Read follow-up queries as what the service reported it would search. They are not proof of a private execution inside that service.
Keeping a series comparable
The value of tracking is comparison over time, and that only holds if the question stays fixed. Change the wording of a prompt and the series restarts, because a different question is now being asked. Keep the wording stable, and record the date when you deliberately change one.
Answers from AI assistants vary between runs even when the prompt does not change. That is a property of the systems being measured. It is why a single run is weak evidence and a series is stronger, and it is covered in why repeated answers vary.
Use these collection rules when setting the tracking schedule and explaining the resulting series. You can plan usage and keep comparisons tied to a consistent setup.
See AI rank tracking, how tracking usage is counted, or how to select a prompt set.