AI Search Signal Watch

How AI visibility tools measure your brand: channels, denominators, and attribution

Browser-scraped consumer answers differ from API calls, metric denominators vary by tool, and an AI mention is not a recommendation or site visit.

A volume estimate is not a headcount

The useful question is where a number comes from. These definitions help you read a report without mistaking a model for observed demand.

Ahrefs Brand Radar

Demand estimate
AI adjusted volume multiplies a parent keyword’s Google search volume by the AI platform’s estimated usage ratio. SourceIt is modeled demand, not a count of people who typed that exact prompt. Ahrefs derives ratios from its aggregated referral traffic data.
Prompt selection and history
Prompts come from People Also Ask. If several keywords share a question, Ahrefs uses the highest-volume parent keyword. SourceAhrefs changed to AI adjusted volume on August 31, 2026; earlier exported volumes and affected metrics may not match current figures.
What a mention counts
A brand appearing several times in one AI answer still counts as one mention. SourceA mention count is not the number of people who saw it.
Found versus cited
Found in includes retrieved pages whether or not the answer cites them. A citation requires the page to be cited; multiple pages from one domain count once at domain level. SourceAI share of voice compares modeled impressions against tracked brands and depends on entity setup.

OtterlyAI

Coverage and demand
At prompt level, Brand Coverage is the share of collected AI responses that mention the brand. Intent Volume is an estimate of query frequency. SourceNeither metric is an observed count of all AI users. Keep the tracked prompts and engines fixed when comparing periods.
Cadence and inspectable answers
Prompts run daily, or weekly on Premium, across the workspace’s chosen engines. Prompt detail shows individual responses, citations and engine breakdowns. SourceThis is the documented collection workflow; no common repeatability or accuracy test has been run here.

Can you take the data with you?

Ahrefs publishes API access that our earlier comparison hadn’t confirmed. Keep the API allowance separate from the dataset you buy.

Ahrefs Brand Radar

API plan and units
API v3, MCP and Ahrefs Connect are available on Lite and higher. Lite includes 200,000 API/integration units per month and a 100-row request limit. SourceBrand Radar data access still depends on the purchased dataset. Custom-prompt API retrieval has the separate zero-unit rule below; other endpoints may consume units.
API data you can request
The Brand Radar API lists AI responses, cited pages, cited domains, mentions, citations and share-of-voice endpoints, including historical metric endpoints. SourceThis confirms documented endpoint coverage; it does not establish that a standalone $50 tracking purchase includes every discovery-index dataset.
The reader’s question

Will this bring you more business?

We can’t tell you that from these sources. A brand mention, a citation and a visit to your site are different things. The tools describe ways to track them; that doesn’t show how much extra business they bring you.

What the metrics mean →
Profound
ChatGPTChatGPTGeminiGeminiAI OverviewsEnterprise: Perplexity, AI Mode, Claude +4
Trial: three engines. Enterprise: ten named capabilities, custom tracking and price.
Collection channel & cadence
ChatGPT, Gemini, Google AI Overviews (Page says Enterprise has more Answer Engines; exact Enterprise set not stated here) Source
Profound says it collects from consumer browser interfaces and runs tracked prompts daily. (Collection claims do not establish equivalence to all personalized sessions or a measured error bound.) Source
Profound Enterprise lists capability for ChatGPT, Perplexity, Google AI Mode, Gemini, DeepSeek, Anthropic Claude, Google AI Overviews, Exa Search, Muse Spark and Qwen. (Enterprise capability list, source lines127–149; tailored prompt tracking and custom price. Trial remains ChatGPT/Gemini/AI Overviews only. Collection channel and selected engine mix need confirmation.) Source
Metric definitions & denominators
Profound defines visibility using brand-containing responses as the denominator, and reports share of voice and sentiment separately. (Do not compare this percentage directly to an all-response denominator without reconciling exclusions; a mention is not necessarily a recommendation.) Source
Prompt, region & persona controls
The product supports manually uploaded/generated prompts and reports multi-region and multi-language monitoring. (Enterprise scope is not the trial: the saved price page limits trial users to a recommended, non-customizable prompt set.) Source
Citations, exports & traffic attribution
AEI offers raw-data CSV export. (Confirm exact fields, API access and plan entitlements for an operational integration.) Source
Profound’s pricing comparison lists Trial exports and API as None; Enterprise exports as CSV/JSON and API as Yes. (Trial/Enterprise table, source lines183–193. Enterprise price is custom. API inclusion does not establish REST protocol, endpoint details, request quotas or limits.) Source
Otterly.AI
ChatGPTChatGPTPerplexityPerplexityGeminiGeminiCopilotCopilotAIO, AI Mode, Claude (API)
All seven selectable using credits; cards retain a four-engine configuration. Claude costs three credits per answer.
Collection channel & cadence
4 of 7 engines (Specific included engine names not encoded in this fact) Source
Otterly describes collection through public AI web interfaces, with Claude tracked through an API. (This describes the channel, not an independent accuracy result or every personalized user session.) Source
Prompts can use country-specific context; Otterly acknowledges answers vary with session, location and account settings. (A neutral sample should not be presented as the answer every customer sees.) Source
Trakkr describes Otterly as a trend baseline and proposes a controlled evaluation; it discloses that it did not access an Otterly account or rerun prompts. (Competitor-authored assessment, not an independent performance audit. Its daily-only cadence discussion predates the November vendor update.) Source
The November 10 release announces all seven engines available for selection, with at least three per workspace and credit-dependent prompt capacity. (The release says plan prices stay the same and existing customers receive separate migration plans. The displayed four-engine configuration is not a universal entitlement limit.) Source
Otterly’s method page names ChatGPT’s website as a collection example and says Claude is tracked through an API. (Collection-method examples, not a complete engine roster or Lite entitlement list. The seven-engine release does not name all seven in this saved passage.) Source
Otterly’s pricing table names ChatGPT, Google AI Overviews, Perplexity, Microsoft Copilot, Google AI Mode, Google Gemini and Claude, and says every plan can track every engine using credits. (Saved pricing table engine section; top cards retain a4-of-7 configuration while the table and November release allow all seven selections. Credits and chosen cadence determine capacity; Claude uses three credits per answer and is collected via API per the method page. Do not treat the card’s four-engine selection as the complete roster.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controlsPrompt & persona controls unstated in reviewed pages
Citations, exports & traffic attribution
The dashboard exposes individual prompt results and the captured answer text. (Availability of text is not proof that all relevant citations or mentions were extracted correctly.) Source
Prompt history begins when tracking starts, with no earlier backfill; deleting a report need not delete saved-prompt history. (Report deletion is not the same as deleting the underlying prompt or terminating the account.) Source
API endpoints expose prompt/engine results, mentions, citations, sentiment and share of voice for automated reporting. (The API guide names Standard/Premium/Custom access and 2,000/5,000 monthly requests; the Lite pricing card has conflicting access wording.) Source
User Analytics counts AI-referred website visits/page views using a site snippet, on Standard, Premium and Custom Enterprise. (This is human referral observation, distinct from bot crawling. Consent affects capture; the page does not establish causal revenue impact.) Source
A November 10 release introduces weekly tracking on Premium and unlimited GEO audits, alongside new credit-pack and workspace controls. (A dated vendor announcement supports an offering change; account migration and specific workload capacity remain scope-dependent.) Source
The October 5 release retires raw Brand Mention Count from Brand Reports in favor of coverage, position and share of voice, while retaining the raw count through API/MCP. (Dashboard metric continuity changed without proving the raw observations disappeared. Keep metric definitions and export channels versioned.) Source
The November 10 connector release adds multi-engine/tag selection and a prompt-level dataset alongside the existing brand overview dataset. (Release-note claim; account entitlement and existing connector refresh behavior should be checked for the selected plan.) Source
The June 30 release states the free trial is now seven days. (The announcement supports a changed trial offering but does not specify the previous duration; leave that unknown.) Source
The September 16 changelog reports ChatGPT advertiser reporting temporarily paused after an upstream display change. (This is a dated availability event, not proof it is still paused today. A later resumption event was not established in this review.) Source
Peec AI
ChatGPTChatGPTGeminiGeminiPerplexityPerplexityAI Overviews, AI Mode +3Copilot (product docs)Copilot (product docs)
Brand plans select three active models; extra-model fees apply. Enterprise choices differ. Confirm Copilot plan/channel.
Collection channel & cadence
Choose 3 models (Model identities may differ by plan; not resolved in this fact) Source
Peec describes browser/UI collection and daily prompt runs, normally letting the consumer platform choose its model and search behavior for a logged-out session. (This describes a sampling context, not universal users. Pricing and connector fields also reference API channels, so channel should be tracked per selected model.) Source
Peec’s introduction says it runs prompts across ChatGPT, Gemini and Copilot daily. (Product documentation, not a promise that all three are included in every selected plan or collection channel; the pricing card lets buyers choose three models.) Source
Peec’s brand-plan table lists ChatGPT, Perplexity, Google AI Overview, AI Mode, Gemini, Naver AI Briefing, ERNIE and Bing AI Mode as available models. Enterprise additionally lists API model choices. (Containing plan table begins at line65. Starter/Pro/Advanced select three active models; Enterprise unlimited. Additional-model offer at lines245–253 names ChatGPT, AI Mode, AI Overviews, Bing AI Mode, Perplexity and Gemini, with separate plan-based fees. Product documentation separately names Copilot; do not infer it is in every three-model selection.) Source
Metric definitions & denominators
Peec separately reports brand visibility, position, sentiment, sources and links; source retrieval and explicit citation are distinct metrics. (Read the actual metric denominator and distinguish a brand named in an answer from a source linked or retrieved.) Source
Prompt, region & persona controls
Prompts can be grouped by topics/tags and examined by country/model; suggestions include intent and estimated volume information. (Country coverage and project limits are tier-dependent; estimated prompt volume is not a census of actual buyer demand.) Source
Citations, exports & traffic attribution
All Chats exposes prompt-level responses, timestamps, sources, links, mentions and citation counts. (Tag filters do not apply to every recent-chat view; inspect the selected data context before reproducing a report.) Source
The API guide explicitly includes Pro, Advanced, Enterprise and agency plans, while excluding Starter. (The saved detailed pricing comparison agrees; the abbreviated Enterprise card alone would misleadingly suggest Enterprise-only access.) Source
The AI Referrals API documents GA4-derived session, event, conversion and revenue fields. (Schema support is a capability claim, not a live measured result. Requires connected/configured data; do not use the example response numbers as customer outcomes.) Source
The price comparison lists AI referrals on all four displayed brand plans. (Availability of a dashboard and programmatic API availability are different; Starter does not include the API.) Source
Scrunch AI
ChatGPTChatGPTPerplexityPerplexityGoogle AIOCopilotCopilot
Core plan (4 LLMs)
Collection channel & cadence
4 LLMs; ChatGPT, Perplexity, Google AIO, Copilot (Vendor page labels these; no independent coverage test) Source
Scrunch describes platform-specific browser automation and official APIs, checked against a separately collected response dataset. (The comparison dataset is a vendor-described validation process; no independent quality audit or error estimate was reviewed.) Source
Metric definitions & denominators
Scrunch separates brand presence, brand position, sentiment and source citations. (Presence or positive sentiment is not the same as the answer recommending the brand to a buyer.) Source
Prompt, region & persona controls
Optional, editable personas are injected into assigned prompts; multiple personas produce prompt variants. (This deliberately changes the input context. Compare persona-driven runs separately from neutral/logged-out query baselines.) Source
Citations, exports & traffic attribution
AI Referrals uses GA4 for AI-referred sessions, conversion rate and purchase revenue; Agent Traffic separately monitors crawler requests via hosting/CDN integration. (Revenue and conversion reporting require GA4 configuration. Bot visits, referred humans and causally incremental sales are different quantities.) Source
Semrush
Prompt DB: ChatGPT, Gemini, AIO, AI ModeBrand Perf: ChatGPT, Perplexity, Gemini, AI ModePrompt Track: ChatGPT Search, Gemini, AIO, AI Mode
Engine lists differ by report. The Starter price excerpt doesn’t name them.
Collection channel & cadence
Not stated for this tier in reviewed excerpt (Absence in excerpt is not evidence feature is unavailable) Source
Semrush describes separate datasets for broad prompt discovery, weekly Brand Performance and daily custom Prompt Tracking. (Collection method, audience and cadence depend on the report. The prompt-database section says daily rolling updates while its summary says monthly; retain this inconsistency.) Source
Semrush KB lists ChatGPT, Gemini, Google AI Overviews, and AI Mode in Prompt Database; ChatGPT, Perplexity, Gemini, and AI Mode in Brand Performance; and ChatGPT Search, Gemini, AI Overviews, and AI Mode in Prompt Tracking. (Engine coverage varies by Semrush report/database and plan tier rather than one uniform engine set across all modules.) Source
Metric definitions & denominators
The broad AI Visibility score combines topic coverage and mention consistency; topic volume is modeled from data, not a measured count for every exact prompt. (Its 0–100 score is not interchangeable with a simple mention percentage from another vendor.) Source
Custom Prompt Tracking separates mentions and cited-domain positions; its guide says estimated traffic and share of voice are not available in that report. (This is a report-specific limitation; do not infer the entire Semrush suite has no traffic analytics.) Source
Prompt, region & persona controls
Prompt Tracking accepts manual or TXT/CSV prompts and location/language targeting with daily tracking. (Documentation names ChatGPT Search, AI Mode and Gemini for setup; the broader data page also names AI Overviews. Confirm report-specific coverage instead of pooling product-wide logos.) Source
Citations, exports & traffic attribution
Brand Performance exposes a full-answer panel and cited-domain context; historical weekly snapshots can be selected. (The scope is this report and its stored sample, not every consumer answer or an independently validated market census.) Source
Ahrefs Brand Radar
ChatGPTChatGPTPerplexityPerplexityGeminiGeminiCopilotCopilotAI Overviews, AI ModeGrok new collection paused
Custom tracking; discovery-index packages have separate scope.
Collection channel & cadence
Brand Radar describes a large query-derived discovery index alongside custom prompt tracking. (Index breadth is not evidence that a custom buyer query was sampled or a mention is a recommendation.) Source
Ahrefs custom prompt tracking lists Google AI Overviews, Google AI Mode, ChatGPT, Perplexity, Gemini and Copilot. (Custom-tracking help page, source lines 25–37. This is not a claim that every discovery-index package includes all engines. New Grok collection is paused separately.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controls
Select question, assistant, location and daily/weekly/monthly refresh. (Lower cadence uses fewer checks; multi-platform monitoring multiplies usage.) Source
Citations, exports & traffic attribution
The product page separates website referral analytics from AI crawler analytics. (Visits and bot activity are different from causal revenue uplift; installation is required.) Source
The help page reports new Grok collection paused after a policy change. (A current availability statement with no confirmed pause start/resumption date; not a measured outage history.) Source
SE Visible / SE Ranking
ChatGPTChatGPTGeminiGeminiAI Mode, AI OverviewsClaude / Copilot: docs conflict
The FAQ says Claude and Copilot are coming soon; other pages list them.
Collection channel & cadence
SE Visible describes browser collection, a Claude API exception and stored unedited responses. (Its FAQ still calls Claude/Copilot coming soon while other sections list them. Confirm selected-plan availability.) Source
SE Visible names ChatGPT, Gemini, Google AI Mode and AI Overviews in both plans and FAQ. Plan blocks also list Copilot and higher-tier Claude, while the FAQ calls Claude and Copilot coming soon. (Source lines 741, 759, 778, 820–831. ChatGPT, Gemini, AI Mode and AI Overviews are documented; Copilot and Claude availability remains a simultaneous documentation conflict.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controlsPrompt & persona controls unstated in reviewed pages
Citations, exports & traffic attribution
SE Visible says API access is included on all plans, with prompt results, metrics and cited URLs. (This statement concerns SE Visible, not pricing or entitlements of other SE Ranking APIs.) Source
The same captured page lists Copilot/Claude in plan blocks while its FAQ calls them coming soon. (Treat this as simultaneous documentation conflict; no actual launch or withdrawal date is established.) Source
Rankscale
ChatGPTChatGPTPerplexityPerplexityGeminiGeminiCopilotCopilotAI Mode, AIO, Claude +10
17+ engines. Browser and API runs can use different credits.
Collection channel & cadence
Rankscale distinguishes browser-interface engines from direct API engines. (Credit cost depends on engine/method; browser collection is not an independent accuracy certificate.) Source
Rankscale lists 17+ supported engines (including ChatGPT, Perplexity, Gemini, Google AI Mode, Google AI Overviews, Claude, DeepSeek, Mistral, Grok, Copilot, and Meta Muse Spark) with no per-engine upsell tiers. (Vendor homepage claim; credit cost and availability vary by engine and plan tier.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controlsPrompt & persona controls unstated in reviewed pages
Citations, exports & traffic attribution
Pro+ advertises REST API access with 3,000 credits at USD 229 monthly. (Pro advertises dashboards/MCP; do not equate those with the Pro+ REST API.) Source
AthenaHQ
ChatGPTChatGPTPerplexityPerplexityGeminiGeminiCopilotCopilotAI Mode, AIO, Claude, Grok
Starter and Enterprise grid names eight engines within its eleven-model claim.
Collection channel & cadence
AthenaHQ’s pricing comparison names ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, Claude, Copilot and Grok under Monitoring for Starter and Enterprise. (Source lines 281–298. Eight named engines within a broader eleven-model claim; model versions and collection channels remain unknown.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controls
Starter lists country-level support and eleven-model coverage; multi-region and persona targeting appear in Enterprise. (Do not include Enterprise persona/region features in the Starter price.) Source
Citations, exports & traffic attribution
Starter lists CSV export; API access is a paid add-on rather than included. (Enterprise lists API inclusion. Add-on prices are not established by this capture.) Source
AthenaHQ lists GA4 and Search Console integrations for connecting visibility and site traffic. (An integration claim does not establish causal attribution or a measured result.) Source
Surfer AI Tracker
ChatGPTChatGPTPerplexityPerplexityGeminiGeminiAI Mode, AI Overviews
Standalone AI Search Analytics offer
Collection channel & cadence
The standalone offer names ChatGPT, Perplexity, AI Mode, AI Overviews and Gemini. (Named plan coverage does not establish every personalized session or an engine version.) Source
Surfer describes repeat/self-consistency checks and cross-model comparison to reduce response noise. (Vendor method claim; repeat counts, error bounds and independent accuracy are not established here.) Source
Metric definitions & denominatorsDenominator / score definition unstated in reviewed pages
Prompt, region & persona controls
Prompt Management allows setup suggestions plus manual prompts; deleting the entire project requires support. (Prompt-generation and project reset are different capabilities; retain the plan-specific limits.) Source
Citations, exports & traffic attribution
Current standalone and bundled offers have different prompt quotas and refresh rates. (Treat product-scope differences separately from proven historical price changes.) Source

← Return to 10-tool overview