Which AI visibility tools do the AI engines actually cite?

How did we rank the AI visibility tools?

By counting the answers in which an engine linked each tool's domain, in one full audit of our own category.

• One Citeshare audit of citeshare.ai, category "AI visibility audit tool", run on 2026-09-13; the report is public at /example.
• 20 buyer questions × ChatGPT, Claude and Perplexity three times each, Gemini and Grok once = 220 answers, 0 failed. None of the questions names Citeshare; 11 name Otterly, Peec or Profound, the competitors confirmed at checkout.
• A tool is counted once per answer in which an engine linked its host (www stripped; a subdomain such as visible.seranking.com or help.otterly.ai counts as its own host), never for a bare mention without a link. The methodology defines "linked".
• Citeshare's own pages were linked in 0 of the 220 answers, and it is listed at that number.
• One category, one day, engine APIs with web search as a proxy for the consumer apps. We re-run after 2026-10-13 and update the tables.

Which AI visibility tools do the AI engines cite most?

Profound, Geoptie and Discovered Labs, and then a cluster of eight tools within five answers of each other. The table lists the 12 most-linked tool and vendor hosts from the audit (aggregators such as LinkedIn, G2 and Reddit are counted separately below), Peec (named in five of the questions but outside the top 12) and Citeshare. Ranks are by answers; other hosts may sit between the 12th row and Peec. Prices are the cheapest public tier on 2026-09-13; the crawler column is the pass · warn · fail count from the readability scan below.

Two prices come from third-party pages because the vendor publishes none: Profound's from a pricing aggregator, Peec's from a review dated August 2026. Treat those two as reported, not quoted. Every other price was read from the vendor's own pricing page on 2026-09-13.

What do the engines cite instead of the tools?

Aggregators and other vendors' comparison posts. LinkedIn and G2 were each linked in 23 of the 220 answers, Reddit in 21, Zapier in 10, the HubSpot blog in 9 and YouTube in 8; SourceForge and Capterra in 3 each. 14% of answers linked no product site at all.

Among vendor domains, the pages that won questions were rarely product pages. The most-linked site on "alternatives to Otterly", "alternatives to Peec" and "cheaper alternatives to Profound" was in each case a different vendor's "alternatives to X" post, and the rest of the vendor links go mostly to "best AI visibility tools" roundups and "X review" posts. Otterly and Profound each won the "is X worth it" question about itself; Peec did not. The exception is the cost question, which Semrush's own pricing page won. Geoptie's 11 questions, the widest spread in the table, come largely from its reviews of and alternatives lists for other tools. This matches the pattern in the guide: engines lift third-party comparison pages before first-party ones.

Do the tools pass their own crawler checks?

Mostly, and the misses are ordinary ones. We ran the free AI crawler check, the same 13 deterministic checks the audit runs, on all 17 sites on 2026-09-13. No site blocks any of the eight AI crawlers. Two pass all 13. Thirteen of 17 carry no freshness signal, five have no llms.txt and two no sitemap.xml at the root. Three sites hide a substantial share of their text until JavaScript runs, which the crawlers do not.

Four hosts here are not in the ranking: peec.ai, and three that were the most-linked site on one of the cost questions each: supermarketers.ai (a pipeline-generation service, no public price), aiclicks.io (an AI search monitoring tool, $59/mo for 30 prompts per its pricing page) and layer3labs.io (an AI-automation agency, no public price).

These checks measure whether a crawler can read the site, not how good the tool is, and a clean scan does not earn citations: the two 13-of-13 sites were linked in 0 answers and outside the top 12 respectively, while the most-linked site passes 9. Run the check on your own site at /tools/ai-crawler-check.

Where does Citeshare fit?

Citeshare is a one-off $19 audit, where every other tool in the ranking is a monthly subscription or an agency: you get one complete run of 20 buyer questions across five engines, the saved answers, and a fixes pack. It does not have a dashboard, daily tracking or alerts; if you need to watch a number move week by week, one of the subscriptions above is the right shape. The cost page sets out when a one-off is enough and when it is not.

Method notes and limits

• Sample size. 220 answers from 20 questions. With three runs per question on ChatGPT, Claude and Perplexity and one on Gemini and Grok, a per-tool count of 20 has a margin of several answers; the order of the middle eight rows is not settled.
• One category, one day. The questions are the ones a small-business buyer of an AI visibility tool asks. A different category, or the same one a month later, gives a different table. Engine citation mixes shift abruptly.
• The questions favour three names. 11 of 20 name Otterly, Peec or Profound, because those were the competitors confirmed for the audit. Their counts and the "alternatives to" pattern partly reflect that.
• APIs, not apps. Each engine is queried through its API with web search on, which is a proxy for the consumer apps: the direction carries over, the exact percentages do not. Models and backends are on the methodology page.
• Prices were read on 2026-09-13 and change without notice; the source link in each row is the place to check. Prices marked as third-party are reported, not quoted.
• Your own numbers. The free snapshot asks four engines five of your buyers' questions and shows who each one links to; the full audit is what produced this page.

Run the free snapshot