Honeylog Review (2026): See the AI Bots GA4 Misses

1What is your analytics missing in 2026?

This is not a niche problem any more. According to Cloudflare Radar figures compiled by SEOmator, 72.5% of AI-bot requests in July 2026 were for plain HTML pages — exactly the content you write. And the trade is lopsided: for every visitor an AI platform sends back, its crawlers fetch dozens to hundreds of pages. Traditional search engines sit at single digits.

Those ratios move fast. Anthropic’s was around 286,930:1 in January 2025, according to Cloudflare’s own analysis. It fell to 1,915:1 in July 2026 and 899:1 in August. OpenAI went the other way that summer, from 256:1 to 391:1. A global average cannot tell you which of these bots is reading your pages, though — that is the gap a log-based tool fills. Our explainer on Chrome’s local network prompt shows the same pattern from the browser side. A lot happens on the web that your usual dashboards never surface.

2What is Honeylog and how does it work?

Honeylog was founded on 1 June 2026 in Rome by the SecondBreakfast team, with Lorenzo Diaco as CEO, and launched on AppSumo on 10 September 2026. It fits the kind of small, independent software we look for in our hidden-gem software research: one focused job, done at the log level rather than in the browser.

What it tracks

According to the Honeylog FAQ, it identifies GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended, Bytespider (ByteDance), CCBot (Common Crawl), Meta-ExternalAgent, Applebot-Extended, Amazonbot and others by name and user agent. Because anyone can put “GPTBot” in a user-agent string, Honeylog checks each claim against the IP ranges that OpenAI, Anthropic, Perplexity, Google and others publish, and flags mismatches as spoofed. That matters: impersonated bots are a classic cover for scrapers, a risk we also saw in the Huntarr security incident, where trusting surface signals went badly wrong.

The WordPress route

If you run WordPress without a CDN, there is an official Honeylog plugin (version 1.0.1). It queues front-end requests plus .xml and .txt files locally and sends them to Honeylog every 60 seconds in batches of up to 100, using WP-Cron. Two caveats from its own documentation: it had no reviews yet on wordpress.org, and the developers say it is not the complete integration for WordPress VIP, because edge-cache hits there never run PHP. The same logic applies to any heavily cached site — a CDN log feed sees more than a plugin can. If you are weighing a move to a setup with a proper CDN in front, our shared-to-cloud hosting migration guide walks through it.

3What did early users find with Honeylog?

Marketer David Nelson bought Tier 1 and wrote up his first weeks in a detailed hands-on review. His favourite use case surprised him: watching Googlebot actually find his new articles, instead of assuming his sitemap did the job. ChatGPT-User, OAI-SearchBot, PerplexityBot, DuckAssistBot and Claude-User all appeared in his early data.

The most useful lesson in his write-up is about reading the numbers carefully. A headline of four OAI-SearchBot requests looked like OpenAI crawling his content — until he added URL and status-code filters and found all four were requests for robots.txt. Another URL showed 12 bot requests that, once filtered, were all classified as “Fake Human”.

“Much more useful than I expected, but there’s a learning curve.”— davenelson, verified Tier 1 buyer, 4/5 on AppSumo [4], 17 September 2026

The second public review tells the story from the other side. A Tier 3 buyer credited the founder with guiding the setup personally, even over a weekend, and moved his nameservers to Cloudflare for the first time to make it work. What he saw afterwards — bots, fake users and spammy IPs — convinced him his site needed changes.

“It felt like I’d been sleeping in a bed of snakes.”— verified Tier 3 buyer, 5/5 on AppSumo [5], 19 September 2026

Not everyone got that far quickly. In the public AppSumo questions [13], a user on the Dorik site builder asked whether Cloudflare was required. The founder explained that logs must arrive through a CDN or from the server itself, and offered to help by e-mail. For hosted builders without log access, that means adding Cloudflare in front first.

4How much does Honeylog cost?

Source: AppSumo listing, checked 23 September 2026. Tier 3 and up add API access and custom alerts. One reviewer reported Tier 1 at $49 a week earlier, so the deal price has changed since launch. The deal is for new Honeylog users only and must be activated within 60 days.

Outside AppSumo, the official pricing page lists Lite at €49 a month (€490 a year) and Business at €79 a month (€790 a year) at the 500,000-events step, with Enterprise on request. Lifetime deals only make sense if the company is still around in a few years — a fair question for any four-month-old start-up, and one we raised in our Accordio review as well.

5Honeylog pros and cons

6Should you allow or block AI crawlers?

Plenty of sites have already decided. GPTBot is now the agent most often named in robots.txt files: 827 top domains in Cloudflare Radar’s scan of 24 August 2026, ahead of ClaudeBot at 726. And 24.0% of all crawler requests worldwide got a 403 in August, up from 20.1% in January (SEOmator). The payoff from allowing bots is thin but real: a Stanford Graduate School of Business study summarised in the same report put click-through rates at 0.33% for AI chatbots and 0.74% for AI search engines, against 8.6% for Google Search.

A robots.txt that blocks training but keeps citations

Each AI company runs separate bots for separate jobs, so you can opt out of model training and still appear in AI answers. The bot names and roles below come from the Cloudflare Radar bot directory as summarised by SEOmator (28 August 2026). robots.txt is a request, not a wall — which is why checking the logs afterwards matters.

# Training crawlers: opt out
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: CCBot
Disallow: /
User-agent: Bytespider
Disallow: /
User-agent: Meta-ExternalAgent
Disallow: /
# Search and user-action bots: allow (citations and visits)
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: PerplexityBot
Allow: /

Adjust the list to your own business case — a publisher negotiating licences may decide differently from a shop that wants every mention it can get.

A block rule is only as good as your ability to check it. Honeylog’s daily robots.txt fetch and its bot-by-bot view let you see whether a crawler you disallowed still turns up — which is exactly the kind of evidence self-hosters value. If you run your own services, the same “measure before you trust” mindset runs through our Brows3 and Dawarich reviews.

7Honeylog review verdict: who is it for?

The early community sentiment is positive but measured: 4.5 out of 5 from two verified AppSumo buyers, and 3.6 out of 5 in the one independent review so far. Both reviewers are keeping the product. Both also say its value comes from letting data accumulate on your own site, which is what the 60-day refund window is for.

Good fit: publishers weighing AI licensing, SEO and GEO specialists who want to see crawler behaviour page by page, agencies reporting to clients, and technically minded owners of WordPress sites or anything behind Cloudflare. If you like tools such as Obtainium that give you control instead of hiding the plumbing, Honeylog has the same spirit.

If you want a tool that turns search data into fixes rather than showing raw logs, our Nuwtonic review covers an agentic SEO and GEO platform that works from your Search Console data.

Poor fit: anyone who wants a dashboard that tells them what to fix, and sites on platforms that give no log access and no CDN option. If your interest in AI is more about agents working for you than bots reading you, our jurniti review covers that side.