AI platform tracking: what is included, and what is metered
They meter looks at data they must re-buy. We include looks at data we already own.
That is not a discount or a launch offer, it is a difference in how the two products are built. Showing you the exact wording an AI used about your brand means querying the engine again, and an engine query costs money every time. We keep every answer we sample as a stored record, so reading it back costs nothing and is simply part of the plan.
| Capability | SEOAIO | Ahrefs Brand Radar |
|---|---|---|
| Reading the exact wording an AI answer used about you | IncludedIncluded on every paid plan. We quote the verbatim sentence from the stored answer. | Charged separatelySold as a separate subscription on top of the base plan.Source: Ahrefs in-app upgrade modal and ahrefs.com/brand-radar, seen 2026-07-30Showing an answer requires them to query the engine again, which costs per look. We store every sampled answer as an immutable artifact, so re-reading it is free. |
| Price for AI-platform tracking | IncludedPart of the plan. No separate AI subscription, no per-platform fee. | Charged separately$199/mo per platform, or $699/mo for all platforms, charged on top of the base plan.Source: Ahrefs upgrade modal, seen 2026-07-30 by a customer on a $249/mo base plan |
| How many questions get asked | Different approachFewer, and they are YOUR questions: the prompt set is yours to edit, so every number answers something you chose to ask. | Different approach458M+ prompts a month, far more than we sample, derived from Google search demand and replayed at AI platforms.Source: ahrefs.com/brand-radar, "458 M+ Total monthly prompts", seen 2026-07-30Their breadth is real and we do not match it. The trade is fidelity: a big number about questions nobody in your market asks is not visibility, and a small number about the questions your buyers actually type is. |
| Sample size and confidence beside every figure | IncludedEvery share prints its sample size and a 95% confidence interval, and the composite refuses to render below the measured floor. | Not foundCould not verify a published sample size or interval on their public pages.Source: Reviewed ahrefs.com/brand-radar and ahrefs.com/custom-prompts, 2026-07-30 |
| Which specific pages the engines pull from on your prompts | IncludedIncluded: the exact third-party URLs, ranked by how often engines returned to each, with your own citation count beside them. | Different approachCited pages are shown; platform-level YouTube, TikTok and Reddit surfaces are in beta.Source: ahrefs.com/brand-radar feature list, seen 2026-07-30 |
| Claude coverage | IncludedSampled directly as one of the named engines. | Not foundNot among the tracked platforms listed on their product page.Source: ahrefs.com/brand-radar platform list, seen 2026-07-30 |
| Quota arithmetic stated in worst-case terms | IncludedEvery quota prints what it costs at maximum use, because a cap you cannot forecast is a price tag you cannot read. | Charged separatelyCustom prompts meter as one check per prompt, per platform, per location, per day, with 2,500 included at the top tier.Source: ahrefs.com/custom-prompts, seen 2026-07-30A compound unit makes a cap read as larger than it is: ten prompts across six platforms in three locations, checked daily, consumes 180 checks a day. |
Where they are genuinely ahead
Their prompt volume is real and we do not match it. They ask hundreds of millions of questions a month and can show you mentions you never thought to look for. If what you want is the widest possible net across a whole category, that is a genuine strength and this page is not going to pretend otherwise.
The trade is whose questions get asked. Their set is built from Google search demand and replayed at AI platforms. Ours is the set you write and edit, so every number answers a question you chose. A large number about questions your buyers never ask is not visibility; a smaller number about the ones they type is.
How to check this page
Every claim in the table carries its source and the date we read it. Prices and features change, and a comparison without a date is a comparison you cannot verify. If something here is out of date or wrong, tell us and we will correct it. Our own methodology, including the sample sizes and confidence intervals behind every figure we publish, is at how we measure, and the accuracy of our own detector is published at detector accuracy.
No account, no card. You get the whole result including the failures.