Key takeaways
- Profound ingests raw server and CDN logs (Vercel, Cloudflare, AWS CloudFront, Fastly, Netlify) and verifies bots against published IP ranges, giving it the most independently documented crawler-log pipeline of the three.
- Scrunch AI's Agent Experience Platform (AXP) works at the edge, classifying bots into retrieval, index, training, and agentic buckets, but the deepest features are locked behind a custom Enterprise quote and the company lacks public developer docs.
- Evertune is not really a crawler-log platform at all. Its pitch is direct API access to foundation models plus a demographically weighted panel of real users; bot analytics exist but are thin and barely documented.
- Sitecore acquired Scrunch on June 3, 2026, which is worth weighing if you care about product independence and roadmap continuity.
- If crawler-log depth is your main requirement rather than AI answer-tracking breadth, tools built around Agent Analytics style log ingestion, including Promptwatch, are worth evaluating alongside the big three.
Why crawler log depth even matters
Most GEO vendors will show you a chart of how often your brand gets mentioned in ChatGPT or Perplexity. That's useful, but it answers a different question than "is Claude-User actually fetching my pricing page, and did it get a 200 or a 403?" Log-level data tells you whether the retrieval bots that power live AI answers can even reach your content, which pages they favor, and where you're bleeding crawl budget on pages nobody in an LLM will ever cite.
This matters more than it used to. Claude's citation crawler went from roughly 30 visits a day in mid-December 2025 to several thousand a day by mid-April 2026, a jump of more than 100x in four months, according to Promptwatch's data on Claude citation crawler visits. If your monitoring tool only samples API outputs once a day, you'd miss that inflection entirely. If you're logging real traffic, you'd see it happen in near real time.
Meta is an even sharper example. Promptwatch's AI crawler traffic report shows OpenAI's share of verified AI crawler requests dropping from 94.8% in the week of June 8-14, 2026 to 79.8% by the week of August 31-September 6, 2026, as other providers (Anthropic, Google, Perplexity, Mistral) picked up share. A platform that only tracks one or two bot families will quietly go blind to that shift. The report's own advice is worth repeating here: compare your logs against this mix, and if a provider shows a large share in aggregate data but near zero in your own logs, you're probably blocked or unreachable for that provider.
Profound: server and CDN logs, verified against real IP ranges
Profound's crawler product is called Agent Analytics, and it's built differently from most competitors. Rather than inferring bot activity from front-end scraping alone, it ingests actual server-side and CDN logs through native integrations with Vercel, AWS CloudFront (via Amazon Data Firehose), Cloudflare (Workers or Logpush), Fastly, and Netlify. The Vercel setup is a one-click Log Drain; once it's wired up, logs start appearing in the dashboard within minutes according to Profound's own documentation.
Profound verifies bot identities against the crawler operators' officially published IP ranges rather than trusting whatever user agent string shows up, which matters because spoofed user agents are common. It also classifies visits into categories like indexing bots versus on-demand retrieval bots, and lets you filter by Bot Provider, Bot Name, Bot Type, and Page, with anomaly alerts on traffic spikes.
Profound
LightSite AI's 2026 technical-GEO ranking specifically credits Profound for "enterprise crawler observability," describing it as "verified server and CDN AI traffic analytics with bot-purpose classification." The same review notes that Profound (alongside WordLift) is one of the only vendors in its seven-platform comparison to publish genuine developer documentation, which matters if you're trying to verify technical claims rather than take a sales deck at face value.
Scrunch AI: edge-layer detection, but gated and under-documented
Scrunch's crawler-facing product, the Agent Experience Platform (AXP), sits at the CDN edge using Cloudflare Snippets, Akamai EdgeWorkers with a Forward Rewrite cloudlet, or AWS CloudFront Functions. It does two things at once: it detects AI agent traffic and serves it an AI-optimized HTML variant, with logging happening as a byproduct of that edge routing rather than from raw server logs.
A developer walkthrough of Scrunch's Cloudflare Worker implementation found it buckets user agents into four categories: retrieval (ChatGPT-User, Claude-User, Perplexity-User), index (OAI-SearchBot, Claude-SearchBot, PerplexityBot, Applebot), training (GPTBot, ClaudeBot, CCBot, Meta-ExternalAgent, Bytespider, Amazonbot, Diffbot), and agentic (headless browsers like Playwright, Puppeteer, Selenium). It also cross-checks user-agent claims against Cloudflare's verifiedBotCategory signal, flagging mismatches as likely spoofed traffic, similar in spirit to Promptwatch's own IP-matching approach.
That's a genuinely thoughtful classification system on paper. Geoptie's review calls Scrunch's named-crawler tracking "some of the best in the category for teams that think in logs and crawl budgets," pointing to the ability to see that ClaudeBot hit a pricing page eleven times in a week, or that PerplexityBot never reached a key landing page at all. But two things temper that enthusiasm. First, AXP and the named-crawler Agent Analytics are gated to Scrunch's Enterprise tier only, not the $250/month Core plan, so the depth you actually get depends entirely on what you negotiate. Second, LightSite's review flagged that it couldn't find a public developer documentation site for Scrunch at all, unlike Profound, which means most of these capabilities are verifiable only through vendor claims and third-party walkthroughs, not primary documentation.
Add to that Sitecore's acquisition of Scrunch on June 3, 2026, folding AXP into Sitecore's broader digital experience platform. That's not automatically a red flag, but enterprise buyers evaluating a multi-year log-data dependency should ask how Scrunch's roadmap and API stability will be affected by sitting inside a much larger DXP business.
Evertune: strong on API access, thin on crawler logs
Evertune's whole pitch is different from the other two. It provides direct API access to foundation models (ChatGPT, Claude, Gemini, Meta, DeepSeek), which reveals what AI knows about your brand before any search augmentation happens, combined with tracking of consumer AI apps and a roughly 25-million-person demographically weighted panel called EverPanel.
That's a legitimate and differentiated approach to the "what does AI think about my brand" problem, especially as agentic shopping built on base models becomes a bigger deal. But if your specific question is crawler log depth, Evertune isn't really in the conversation. One 2026 review states flatly that "Evertune measures model outputs; no AI-bot or LLM crawler log analytics is documented." A separate comparison notes Evertune does include some site audit and bot analytics features for tracking crawl behavior and technical page signals, but it's thinner and less independently verifiable than what Profound or Scrunch publish. LightSite's technical ranking places Evertune last of the seven platforms it reviewed, describing its technical depth as limited to "site audits and bot analytics alongside broader AI search intelligence," and grouping it with vendors that publish marketing pages rather than developer docs.
Side-by-side comparison
| Platform | Log collection method | Bot verification | Where deep crawler data unlocks | Entry price |
|---|---|---|---|---|
| Profound | Server/CDN log ingestion (Vercel, Cloudflare, AWS, Fastly, Netlify) | Verified against published IP ranges | Available even on lower tiers | $99/mo (Starter, ChatGPT only) |
| Scrunch AI | Edge-layer detection via Cloudflare Snippets, Akamai EdgeWorkers, CloudFront Functions | Cross-checked against Cloudflare verifiedBotCategory | Gated to custom Enterprise tier | $250/mo (Core, 4 engines) |
| Evertune | Basic bot analytics, thinly documented | Not publicly detailed | Not a core focus; Enterprise adds broader data | $800/mo (Pro) |
How LightSite AI's independent ranking stacks up
LightSite AI's September 2026 technical-GEO ranking, which discloses its own methodology rather than relying on vendor briefings alone, places Scrunch second overall for "enterprise agent experience" (AI-specific content delivery at the CDN layer plus agent referral measurement), Profound third for "enterprise crawler observability" (verified server/CDN traffic analytics with bot-purpose classification), and Evertune seventh for having the weakest technical and crawler depth of the group. Its summary line is a useful shorthand: "If your priority is enterprise edge delivery, Scrunch may be the better fit. If your priority is server-log analytics, Profound is extremely [strong]."
That framing is honest about the tradeoff. Scrunch optimizes for controlling what AI agents see at the edge; Profound optimizes for observing, in raw log form, what's actually happening. Neither approach is strictly "deeper" in every sense, but if the question is specifically about crawler log data, meaning actual request-level records you can query, filter, and alert on, Profound's architecture is the more directly comparable to what enterprise SEO teams already expect from server-log analysis tools like Screaming Frog Log File Analyser or OnCrawl.
Where a platform like Promptwatch fits in
If you're comparing these three specifically because you want granular AI crawler visibility, it's worth widening the search a bit. Promptwatch runs its own Agent Analytics style layer, ingesting crawler logs via Cloudflare, AWS CloudFront, Fastly, Vercel, Netlify, Akamai, Google Cloud CDN, or a custom HTTP endpoint, and tracks 400+ AI crawlers including ChatGPTBot, ClaudeBot, PerplexityBot, GoogleOther, Google-Agent, and Meta AI, showing not just that a crawler visited but whether that visit led to a citation, with a crawl-to-citation rate calculated per page.
What separates it from the three platforms above is that the log data feeds directly into action: content gap analysis, automated Content Agents that write and publish to your CMS, and a prioritized Unified Actions list built from your actual visibility and crawler data, not a generic checklist. It's rated 4.7/5 on G2 across 1,840+ customers including Duolingo, Yelp, and Typeform, and its dataset (4.5B+ citations and 26.5B+ data points analyzed) has been covered by the Wall Street Journal and Axios. For teams that want log depth plus a clear path from "here's what's broken" to "here's the fix, already published," it's a reasonable fourth option to put in the bake-off. You can browse more GEO-specific platforms in the directory at bestgeosoftware.com if you want a wider field before narrowing down.
Practical recommendation
If crawler log depth, in the literal server-log sense, is your top decision criterion and you want a vendor with public technical documentation you can actually read before signing a contract, Profound is the strongest of the three as of late 2026. If you care more about controlling what AI agents see at the edge and are comfortable negotiating an Enterprise deal to unlock the deepest features, Scrunch AI's AXP is worth a serious look, with the caveat that Sitecore now owns the roadmap. Evertune is a strong choice if your priority is measuring foundational model knowledge and consumer-panel sentiment, but don't buy it expecting crawler-log parity with the other two; that's simply not what it's built for.
Whichever way you lean, ask each vendor for a live demo of their actual log dashboard, not a slide, and cross-reference what they show you against your own server logs for a week. The gap between marketing claims and what shows up in a real dashboard is usually where the honest answer lives.