# AI visibility software: what the scores measure Compare AI visibility sampling and export access before choosing a monitoring tool. Canonical page: https://quillwell.com/guides/ai-visibility-software/ Updated 2026-10-30 Publisher: Quillwell Guides (https://quillwell.com/guides/about/). Editorial standards: https://quillwell.com/guides/editorial/ ## The short answer - Visibility scores summarize sampled answers, so the prompt panel and denominator determine what they mean. - Compare collection methods and score definitions before putting products side by side. - If you need to audit individual answers, start with documented raw exports such as [OtterlyAI’s response exports](https://help.otterly.ai/can-i-export-my-data-and-reports). Check the fields on your plan. - A brand mention and a cited page are distinct observations; neither proves a purchase. ## Define the observation panel Set the prompt panel before comparing prices. A wider geography or a faster schedule can change both coverage and cost. [Peec pricing](https://peec.ai/pricing) calculates answer volume by multiplying tracked prompts by active models and tracking frequency. Its self-service tiers list 50, 150 and 350 prompts with three active models. Suppose you track one prompt in four countries. [Otterly’s prompt limits](https://help.otterly.ai/amount-prompts-limit) count that as four prompts, so price the geographic panel you actually need. [Otterly’s collection explanation](https://help.otterly.ai/how-otterlyai-collects-data) describes daily monitoring through public interfaces, except Claude via API. It also notes variation by session and location. [Scrunch’s collection description](https://scrunch.com/faqs/what-methods-does-scrunch-use-to-collect-data-from-ai-platforms) covers browser and API methods. New prompts run daily for fourteen days, then every 72 hours by default, with on-demand refresh. It does not map each engine to a collection method. [Surfer pricing](https://surfer.ai/pricing/) lists AI Search Analytics at $82 per month for 50 prompts refreshed daily on annual billing. Standard costs $99 per month on annual billing, with 25 prompts refreshed weekly and 360 documents. Use the same wording and coverage in each demonstration. Ask how missing runs affect the denominator. ## Read the score behind the label Compare the denominator before the headline score. A changed prompt panel can move the number even when existing answers stay the same. [Ahrefs explains](https://help.ahrefs.com/en/articles/16249245-reasons-for-sudden-spikes-in-brand-mentions) that prompt database expansions can produce simultaneous mention spikes across brands. Chart annotations can distinguish index changes from engine changes. [Otterly’s KPI definitions](https://help.otterly.ai/brand-report-kpi-definition) deduplicate repeated brand mentions within one execution. Coverage omits untriggered AI Overviews, so reconcile a sample against the raw export. [Surfer’s metric explanation](https://docs.surferseo.com/en/articles/12330653-ai-tracker-overview-dashboard) says Visibility Score incorporates frequency and prominence, while Mention Rate ignores position. It does not provide an exact visibility formula or missing-run denominator. [Peec’s metrics](https://peec.ai/ai-instructions) describe a 0–100 sentiment scale; [Otterly’s definitions](https://help.otterly.ai/brand-report-kpi-definition) use net sentiment. Read the definitions before putting those figures on one chart. [Otterly’s KPI definitions](https://help.otterly.ai/brand-report-kpi-definition) calculate Likelihood to Buy from average position. This position-derived proxy does not measure purchase probability. Keep the prompt set and scoring definition beside every report. Ask for worked examples of missing answers and repeated mentions, alongside the count of completed observations. For example, a brand appearing in five of ten completed answers has a different share from five of twenty. If failed runs enter one denominator and leave another, the percentages answer different questions. Keep the count of completed observations alongside the score and the report. ## Check exports before you commit Choose an export that lets you inspect individual observations and keep them through a switch. Ask for a sample before committing. [Otterly’s export guide](https://help.otterly.ai/can-i-export-my-data-and-reports) offers raw responses in CSV or JSON on all plans, plus CSVs for citations and prompts. It separates brand presence in an answer from presence on the cited page. [Peec’s export guide](https://help.peec.ai/article/decision-lab-where-can-i-export-my-data-and-can-i-get-full-urls-5) describes chart CSVs and full-project chat CSVs. It also describes URL-level API reports and live Looker reporting. [Peec’s current API help article](https://help.peec.ai/article/how-much-does-the-api-cost-y3u6zh4lby7h9sksvyyzn2mi) lists access on all agency plans and Pro, Advanced and Enterprise brand plans; Starter is excluded. Other [Peec material](https://peec.ai/ai-instructions) conflicts, so confirm the order entitlement. [Scrunch’s Responses API](https://scrunch.com/faqs/what-types-of-data-can-i-access-through-the-scrunch-api) returns raw text and citations alongside prompt metadata. The API advertises a rolling 90-day history; that window alone does not prove deletion of older stored data. [Otterly’s cancellation article](https://help.otterly.ai/cancel-subscription) says tracked engines and historical data are deleted after cancellation. It leaves deletion timing and any recovery window unclear. [Peec’s terms](https://peec.ai/legal/terms-of-use/v2026-10-22) state a 30-day customer-data export period after termination, followed by deletion subject to retention obligations. Annual plans require 30-day termination notice; confirm the applicable order and entity. Open a representative export before buying. Check whether it contains the full response and the citation URLs needed to reproduce a report. ## Measurement and export worksheet Use one worksheet per vendor with the same proposed panel. A sample export makes missing evidence visible before purchase.