If you are responsible for AI visibility reporting, you need to understand how a tracker collects the answers your report will rely on. OppAlerts publishes this page and sells one of the products discussed. It covers how AI rank tracking works rather than which vendor to buy; for that, see AI search visibility tools.
An AI rank tracker sends a prompt to an assistant, records the answer, and reports whether your brand was named and which sources were cited. Four things decide whether the resulting report means anything.
1. What one unit of collection buys
Every vendor sells collection, and each names the unit differently. Ahrefs publishes the clearest definition: tracking one prompt, on one platform, in one location uses one check. OppAlerts counts a call the same way, as one tracked item on one provider on one day. Rankscale sells credits, and publishes the conversion: 50 prompts at 0.75 credits a run, weekly, is 162.5 credits a month. Peec multiplies tracked prompts by active models by tracking frequency. Semrush and SE Ranking sell prompts as a daily entitlement.
Before comparing any two, express your workload as collections a month: prompts multiplied by engines multiplied by runs. A hundred prompts on three engines checked twice a week is roughly 2,600 collections a month, whatever each vendor calls the unit.
A plan advertising a prompt count without an engine count and a frequency is not quoting the same thing.
2. How often collection runs, and when
Frequency is where the ranges diverge sharply. Rankscale schedules from hourly to monthly on every plan. Surfer refreshes 25 prompts weekly on its content plan and 100 daily on its analytics plan. Peec collects daily on its self-serve plans, with daily or weekly at Enterprise. Semrush and SE Ranking sell daily entitlements.
OppAlerts collects at most once a day. You tick which days of the week each item needs fresh data on, and the run happens the night before each ticked day. That suits reporting on a schedule and does not suit watching a prompt through a launch hour by hour, which is what Rankscale is built for.
Ask when collection happens as well as how often. A report read on Monday morning is only useful if the collection ran before it.
3. Sampling, and the question no vendor answers on its pricing page
AI assistants do not return the same answer to the same prompt every time. That is a property of the systems being measured, not a defect in any tracker, and it is explained in why repeated answers vary.
It creates one question that decides how much a tracking report is worth: how many times does the tool run each prompt per collection, and what does it do with the variation?
No competitor page read for this research states its sampling method. OppAlerts runs each prompt once per provider per scheduled collection and reports that one answer, not an aggregate. 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.
Ask every vendor on your shortlist three questions in writing:
- How many times is a single prompt executed per scheduled collection?
- If it is executed more than once, is the reported result one answer, a most-common answer, or a rate across runs?
- What happens when a collection fails, and is the failed attempt charged against the allowance?
The third question matters commercially because vendors account for failed work differently.
4. What history you keep, and whether it stays comparable
A tracking report earns its cost through comparison over time, and that only works if the question stays fixed. Change the prompt wording and the series restarts, because you are now measuring a different question. Keep the wording stable and record the date of any deliberate change.
History retention differs by vendor and by plan. Ahrefs sells 6 months, 2 years and 5 years of historical data by plan, with unlimited history at Enterprise. Profound lists all-time history even on its entry plan. Others do not state a retention period on the pages read, so ask what happens to your series if you downgrade.
Where AI Overview tracking sits
Google AI Overviews are collected differently by different products, and it affects cost.
Several vendors treat AI Overviews as another answer engine alongside ChatGPT and Perplexity, so tracking it consumes a prompt slot or a collection unit of its own.
In OppAlerts the AI Overview arrives as part of a Google keyword request. One request returns the organic results, the ads and the AI Overview together. The AI Overview citation and the conventional position for that keyword appear in the same run and cost the same single call. That is worth knowing if you track a large keyword set and want AI Overview coverage across all of it.
The distinction between AI Overviews and Google AI Mode also matters, because they are different result formats and behave differently. How Google AI Mode and AI Overviews differ covers it.
A checklist for comparing quotes
- Write down prompts, engines and runs per week, and multiply them.
- Ask each vendor what that workload costs on their meter, not what their plan is called.
- Ask the three sampling questions above and get the answers in writing.
- Ask what history is retained and what happens on downgrade.
- Check whether users are charged, since several vendors include unlimited seats and others cap licenses.
- Check whether failed collections consume the allowance.
A tracker that answers all six clearly is usually a better purchase than one with a longer engine list and vaguer arithmetic.
Before selecting a tracker, check collection, frequency and retained history against your reporting needs. You can then assess whether its records support the comparisons you intend to make.
See how OppAlerts collects tracked prompts, how tracking usage is counted, or how to evaluate an AI rank tracker.