Compare · Searchable
Skald vs Searchable: sampled AI answers and a bot list, or the whole log
Skald vs Searchable compares an AI visibility platform with a server-log tool. Searchable asks AI engines your prompts every day, writes content and audits pages, and counts the bots your CDN or app forwards. Skald reads every request in your server log, checks the crawlers of nine companies against their owners' addresses, and follows AI visits to your leads.
Searchable is an AI visibility platform: it asks up to ten AI engines the prompts you track and keeps their answers, and its LLM Analytics takes the requests your CDN, WordPress or app forwards and keeps the bots on its list. Skald reads your server's access log, sorts every request into five groups that add up to the log, and checks the crawlers of nine companies, OpenAI, Anthropic and Google among them, against their owners' published addresses. Every claim about Searchable below links to its own page or docs, read on 26 October 2026.
30-day free trial · no card · we set it up within one working day · free to self-host
out · asked of an AI
back · on your site
- asked
- answered
- fetched
- a fake
- clicked
- lead
A question your buyers ask is put to an AI assistant
- Searchable, by its pages
- Asked daily of ChatGPT, Google AI Overviews and Perplexity on every plan; seven more models, Claude among them, as paid add-ons
- Skald
- Asked of ChatGPT, Perplexity, Gemini or Claude through their APIs, with your own keys, 1 to 5 tries
A sample question and sample requests (made-up addresses). The Searchable lines follow its own pages and docs; the Skald lines are what Skald files.
Searchable, read 26 Oct 2026 · searchable.com/pricing · docs.searchable.com/using-searchable/visibility-tracking.md · searchable.com/features/aeo-insights · docs.searchable.com/setup/overview.md · docs.searchable.com/integrations/mcp.md · docs.searchable.com/integrations/google-analytics-4.md · searchable.com/features/traffic
[01] question by question
Skald vs Searchable, question by question
Nine questions, each Searchable answer taken from its own page or docs and linked. Where its pages did not say, we say so. See every comparison.
| The question | Searchableby its own pages | Skaldyour server log |
|---|---|---|
| What it is for | "Visibility & Analytics from AI Search — and the actions to drive growth"Searchable, 26 Oct 2026 · searchable.com | Every visit and every AI fetch, counted from your log and checked |
| What AI says about you | "ChatGPT, Google AI Overviews, and Perplexity are included on every plan"; Claude, Copilot, DeepSeek, Gemini, Google AI Mode, Grok and Meta Muse are paid add-ons, ten in allSearchable, 26 Oct 2026 · searchable.com/pricing | ChatGPT, Perplexity, Gemini and Claude through their APIs, with your own keys: up to 50 questions in each of up to 10 tracking profiles, 1 to 5 tries each |
| How it gets your traffic | It "points your hosting platform's request logs at our ingest endpoint": Vercel, Cloudflare, CloudFront, Netlify, Fastly, Akamai, Google Cloud, a WordPress plugin, or a REST callSearchable docs, 26 Oct 2026 · docs.searchable.com/setup/overview.md | Reads your web server's access log, or a CDN's; nothing in your code |
| Which requests it keeps | It "only records events whose user_agent matches a known AI crawler" and drops the restSearchable docs, 26 Oct 2026 · docs.searchable.com/setup/rest-api.md | Every request, in five groups that add up to the log |
| Is that crawler real? | Bots matched by user-agent pattern: its public list has 174 names and patterns, and nothing else per bot. We found no address check in its docsSearchable, 26 Oct 2026 · tracker.searchableanalytics.com/v1/bots.json | Checked for nine owners (OpenAI, Anthropic, Perplexity, Google, Bing, Apple, DuckDuckGo, Meta by its network, Yandex) against their published addresses or DNS; a fake is filed under bots as "claims to be GPTBot". Other crawlers are counted by the name they send, unchecked |
| Why an AI came | Bot categories "ai_training, ai_search, ai_assistant, ai_agent, and search_engine"Searchable docs, 26 Oct 2026 · docs.searchable.com/integrations/mcp.md | Three reasons kept apart: training crawl, live fetch (an assistant fetching for someone's question) and click-through (a person clicking from an answer); never added together |
| People sent by AI | It "identifies traffic from AI platforms by analyzing referral sources", read from GA4, or from its own first-party trackerSearchable, 26 Oct 2026 · searchable.com/features/trafficSearchable docs, 26 Oct 2026 · docs.searchable.com/integrations/mcp.md | From the log, by assistant, a floor as apps strip referrers; likely AI visits shown apart |
| From an AI visit to a lead | Conversions from GA4; its crawl-and-visit view joins pages over the same window, which its docs call "co-occurrence, not attribution"Searchable docs, 26 Oct 2026 · docs.searchable.com/integrations/google-analytics-4.mdSearchable docs, 26 Oct 2026 · docs.searchable.com/integrations/mcp.md | The pages that assistant fetched in the 15 minutes before the visit, then the form, booking or payment that followed |
| Price | For brands, Pro $125, Scale $400 and Enterprise $999 a month; for agencies, Launch $250, Growth $400 and Enterprise Agencies from $999; Custom by quote. A 14-day free trial on Pro, Scale, Launch and Growth. As listed on searchable.com on 26 Oct 2026Searchable, 26 Oct 2026 · searchable.com/pricingSearchable, 26 Oct 2026 · searchable.com/pricing?audience=agency | Free to run on your own server; hosted plans are priced per site, with 30 days free, on the Skald pricing page |
Every Searchable answer was read on searchable.com and docs.searchable.com on 26 October 2026; prices are its monthly ones, in US dollars. If we have one wrong, tell us.
[02] what each one keeps
Two sheets laid over the same minute of requests
Say twelve requests reach your site in a minute: people, Googlebot, GPTBot, ChatGPT-User, two bots, and one scraper calling itself GPTBot. Lay each tool's sheet over the list and read what it marks.
198.51.100.7 "GET /pricing/" "Chrome/130"droppedpeople203.0.113.24 "GET /pricing/" "GPTBot/1.2"kept · GPTBotAI · training · on OpenAI's list198.51.100.140 "GET /docs/" "ChatGPT-User/1.0"kept · ChatGPT-UserAI · live fetch · on OpenAI's list192.0.2.10 "GET /blog/" "Googlebot/2.1"kept · Googlebotsearch engines · on Google's list198.51.100.23 "GET /blog/" from chatgpt.com "Safari/18"dropped · GA4 or its trackerpeople · sent by ChatGPT192.0.2.77 "GET /.env" "python-requests/2.32"droppedbots192.0.2.200 "GET /pricing/" "GPTBot/1.2"kept · GPTBotbots · claims to be GPTBot203.0.113.9 "GET /about/" "ClaudeBot/1.0"kept · ClaudeBotAI · training · on Anthropic's list198.51.100.61 "GET /contact/" "Firefox/131"droppedpeople192.0.2.31 "GET /docs/" "bingbot/2.0"kept · Bingbotsearch engines · on Bing's list192.0.2.90 "GET /wp-login.php" "curl/8.5"droppedbots198.51.100.88 "GET /pricing/" "Chrome/130 Android"droppedpeople
6 of 12 requests kept
Kept: the six whose user agent is on its bot list, Googlebot and Bingbot too, and the fake GPTBot filed as GPTBot. Dropped: four people and two bots. People reach Searchable through GA4 or its own tracker instead.
Searchable, read 26 Oct 2026 · docs.searchable.com/setup/overview.md · docs.searchable.com/setup/rest-api.md · tracker.searchableanalytics.com/v1/bots.json
The twelve requests are made up. Searchable's sheet follows its docs and its public bot list, which names Googlebot, Bingbot, GPTBot, ChatGPT-User and ClaudeBot, and not python-requests or curl.
[03] to be fair
Where Searchable does more than Skald
Searchable is built around AI answers and the work that follows them. Pick it for more engines, content and audits, connectors that need no server, and a bigger API.
More engines
Ten on its pricing page, Google AI Overviews and AI Mode among them, with Copilot, Grok, DeepSeek and Meta Muse. Skald asks four and does not track AI Overviews.
Content and audits
A content studio that writes articles (20 to 200 a month by plan) and page audits (200 to 3,000 a month). Skald writes no content.
Deeper answer analysis
Share of voice (your share of the brand mentions in answers), sentiment (how an answer speaks of you) and query fanout (the sub-queries an engine runs behind one prompt), plus AI shopping and the ads shown inside AI answers. Skald counts which answers name you, cite you and name competitors.
No server needed
Connectors for Vercel, Cloudflare (a Worker on any plan, the free one too), Netlify, CloudFront, Fastly, Akamai and Google Cloud. Skald needs a log it can read.
More bot names
174 user-agent names and patterns, published as a file. Skald names 40 AI crawlers from 21 companies, plus search engines and other bots.
Built for agencies
Agency plans with white-label reports and Looker Studio (Google's report builder) on each, unlimited seats on its listed plans, and unlimited projects from Enterprise up. Skald's client reports carry your agency's name, colour and logo, by mail and as a PDF.
A bigger API
A REST API of over 90 operations with write keys, and an MCP server, the way AI agents query a tool, signed in with OAuth. Skald's data API and its MCP server (20 tools) are read-only.
Less personal data kept
Its WordPress plugin zeroes the last part of each address before sending, and its log report never returns an address or user agent. Skald keeps each visit's address in full, shortened or not at all, as your admin sets it; the raw log stays as long as your web server keeps it.
Searchable, read 26 Oct 2026 · searchable.com/pricing · searchable.com/features/aeo-insights · docs.searchable.com/integrations/mcp.md · docs.searchable.com/setup/overview.md · tracker.searchableanalytics.com/v1/bots.json · docs.searchable.com/api-reference/openapi.json · docs.searchable.com/setup/wordpress.md
[04] use both
Ask what AI says, then see what it did on your site
Searchable tells you what AI engines say about you. Skald reads the log you already have, so it shows what they, and the people they sent, did on your site, from the days still on disk.
Ask
Track your prompts in Searchable, or in Skald with your own keys.
Check
Skald checks the crawlers of nine companies against their owners' addresses, and shows what each one fetched, its errors and what it cost your server.
Follow
When someone clicks through from an answer, Skald keeps the pages that assistant fetched just before, then the form or booking that person sends.
[05] choose by the question
Searchable, Skald, or both
Pick by whether you want AI answers and content work, or a checked count of every request in your log.
- "What do ten AI engines say about us, and what should we write?" Searchable.
- "Our site is on Vercel or Netlify and we cannot read a log." Searchable for the bots; Skald's script tag for people.
- "Is that GPTBot real, and do our groups add up to the log?" Skald.
- "Which AI visit became a lead?" Skald.
What it does not do
- Skald asks four assistants (ChatGPT, Perplexity, Gemini and Claude); it does not track Google AI Overviews, AI Mode, Copilot, Grok or DeepSeek.
- Skald writes no content and gives no share-of-voice or sentiment score.
- AI answers in Skald are a sample through each provider's API, like any prompt tracker's.
- On Vercel's Hobby plan, or Netlify below Enterprise, no log drain is offered, so Skald sees people only, through its script tag.
keep reading
More on Skald vs Searchable
questions
Skald vs Searchable FAQ
Straight answers, including what Skald does not do.
What does Searchable do?
By its own pages, read on 26 October 2026: it tracks how a brand shows up in the answers of up to ten AI engines, with share of voice (your share of the brand mentions), sentiment and cited sources, and adds a content studio, page audits and an agent. Its LLM Analytics counts AI bots from the requests your CDN, WordPress or app forwards, and it reads people who came from AI through Google Analytics 4 or its own tracker.
Does Searchable read server logs?
For bots, yes. Its docs point your host's request logs, a Worker, a plugin or a REST call at its ingest, and its REST guide says it "only records events whose user_agent matches a known AI crawler". People come from GA4 or its own tracker instead. Skald reads every request in your log, people included.
Does Searchable check that a GPTBot is real?
Not that we could find. Its docs describe matching bots by user agent, and we found no address check in them. For nine crawler owners (OpenAI, Anthropic, Perplexity, Google, Bing, Apple, DuckDuckGo, Meta and Yandex) Skald checks the address against the list, DNS or network the owner publishes, and files a fake under bots as "claims to be GPTBot". Other crawlers are counted by the name they send. How AI crawler checks work.
Is Skald a prompt tracker?
Partly. Skald asks ChatGPT, Perplexity, Gemini and Claude through their APIs, with your own keys: up to 50 questions a tracking profile, up to 10 profiles a site, 1 to 5 tries each. Its core is the server log. For more engines or content tools, a platform like Searchable does more. Prompt tracking vs server logs.
Can I run Skald on my own server?
Yes. Searchable is a hosted service. Skald runs as a Docker container on a server you choose, free and source-available under the Elastic License 2.0, and reads that server's log or the logs your CDN sends it.
Can I use both?
Yes. Searchable takes the bot requests your CDN or app forwards; Skald reads your server's log. Each is set up on its own, and neither needs the other.
$ tail -f access.log | skald
See what AI does with your site. Free for 30 days.
No card. We set up hosted Skald for you within one working day, or you self-host it free. Every line of your log, sorted and added up.