82M company pages, 250M AI crawler visits: a site that climbed to global rank 130k while getting almost no real human visitors

On 2026 May 5, New York serial founder Gregory Pellitteri posted "Show HN: Is your brand invisible to AI? We built a free visibility score" on Hacker News. It got 1 points and 0 comments. The next day it went up on Product Hunt, collected 2 votes and ranked #250 for the day. Half a year later, his engagemii.com had scored 82.1M websites, and its global rank had climbed from well beyond the millionth place before August to 129k (Tranco, 11-05). But what pushed it up was mainly not people. It was crawlers from ChatGPT, Claude and Meta. This is a case about programmatic pages fed to AI crawlers: rankings can be driven up this way, but users and money don't follow.

A Show HN nobody clicked

The engagemii domain had been live for 2024 years. Back then it was an "AI brand discovery / conversational sales" platform, and its subdomains still show a founder-era price of "top 250 Shopify brands $250/year". In May 2026 it pivoted to AEO (answer engine optimization, getting brands mentioned in AI answers like ChatGPT). The first step was that unremarkable Show HN: enter your URL and get a free "AI visibility" score of 0–10.

Later in June he posted two data pieces on HN, "AI bots cite dentists more than Fortune 500 companies" and "two-thirds of websites are invisible to AI bots," each scoring 3 points. I couldn't find a single post on X mentioning it in the past half year, and no media covered it. From a human-channel perspective, it was nearly invisible.

What it does: scoring the whole world first

Other AEO tools work like this: you enter a brand, they ask ChatGPT, and they tell you how often you were mentioned. Engagemii works the other way around: it doesn't wait for you to show up. It scores every company first.

• The homepage's 11-07 text reads: tracking 82,129,732 websites, re-scoring about 170k companies daily, of which 17.4M verified US companies.
• Each company gets a public brand page. The names scrolling across the homepage include Southern Pines Bathtubs (bathtubs, score 3.6), Shuman's Auto Clinic (auto repair, score 3.9), Emy Ray Bakes (bakery, score 3.8), and Fat Cat Pottery (pottery, score 4.1). Most are small businesses, and their scores are generally failing.
• The scores are mostly static assessments: how readable the website is for AI (6 on-site sub-scores), plus off-site signals like backlinks, news, and social media (6 signals). It doesn't actually query the models. Only one "Live" panel on the homepage asks ChatGPT "what's the best to buy" every 30 minutes and shows who it recommends.
• The free side also includes: a "fix pack" generated for your own site (9 files such as llms.txt and structured data), a Chrome extension, a WordPress plugin (launched in July, with 583 total downloads on wordpress.org), and an MCP server.

How the money is meant to come in: selling data per record

It doesn't rely mainly on subscriptions. It sells access to this list of 82M records. Pricing on the data page (11-07):

The buyers they have in mind are marketing agencies ("every low score is a lead you can sell a fix to"), data companies, and investors. It looks more like BuiltWith or ZoomInfo, with an added AEO score. Nobody outside can see whether anyone has actually bought, or how much they've sold: no disclosed revenue, no customer case studies, no reviews.

Why the rank climbed: the readers are crawlers

Engagemii itself published a research piece in September, "250.1 million AI crawler reads." Checking against each company's published IP ranges, it counted how often AI crawlers read its site:

Compare that with its global Tranco rank: in Aug 20 it first entered the top 1M (868k), Aug 27 430k, Sep 10 205k; it fell back to 286k in early October, climbed again in mid-October, and reached its best of 121k in Nov 1. Crawler reads rose 5x in August, the same late-August stretch when the rank jumped into the list; in September ClaudeBot reads rose about 5x and Applebot about 10x, and the rank climbed again in a second wave. The timing lines up almost exactly.

As of 11-07, its homepage running total is 252M reads: Meta 98.1M, Claude 62.9M, Apple 26.9M, GPTBot 26.1M, Perplexity 14.1M. Google's regular search crawler accounts for only 334k.

Three further signs that these rankings are not driven by people:

• No real Chrome users show up for it. In Google's CrUX data, which buckets websites by actual Chrome users, engagemii.com is absent from the global top 8M in both 1 and September. For comparison, peec.ai, a similar AEO tool with nearly the same rank (Tranco 130k), did make that bucket in September.
• The pages are written for crawlers. An ordinary scraper opening one of its brand pages hits a Cloudflare human check, while AI crawlers are let through by IP range.
• Google has barely indexed it. 330k Googlebot visits across 82M pages suggest these pages bring little traffic from Google Search.

Tranco is a blend of several data sources, and several of them measure DNS and network-level requests. Heavy crawler traffic can push a site's rank up (inference; which source contributed most was not verified). The early-October dip and the mid-October rebound most likely reflect crawlers reading in batches (inference).

It turned being crawled into the selling point

This is the cleverest part. 82M pages attract AI crawlers → the crawler logs become data only it has ("which AI is reading whom") → that data is both sold per record (the "AI crawler reads" column at 2¢ per record) and written up as research pieces for HN, Medium, and LinkedIn ("Meta accounts for 39% of all AI reads", "ClaudeBot reads 2.4x as much as GPTBot"). More pages bring more crawlers, the data becomes more exclusive, and the whole thing promotes itself.

The problem is that no people are in this loop. It proves AI crawlers read programmatic pages in volume. It does not prove anyone signed up or paid because of it.

Who it's competing with

AEO is one of the hottest marketing-tech categories of 2026, and "check for free whether you're invisible in AI" is a hook big vendors already give away for free. Engagemii's only real difference is "pre-score every company, with one public page each." That is exactly why it ranks, not why users show up.

If it were you

Rating C: worth learning from, not worth chasing. What it produced is rankings, not a business: no visible human users, no visible revenue. Copy it and you will most likely end up with a site that crawlers read diligently and nobody pays for. The "free AEO check" entry point is already given away free by HubSpot, Semrush, and Ahrefs, while Profound and Peec have money and users.

Who it suits: a small team that already has a large list (every business in an industry, every app, every open-source project) and an existing channel to sell the data to agencies or data companies. Or someone wanting to build a new data source out of "AI crawler reads."

Who it doesn't suit: anyone aiming for Google organic traffic from programmatic pages (its Googlebot reads are only 330k), or anyone wanting to build an AEO subscription SaaS, a space where companies raising hundreds of millions of dollars already compete.

How to do it: don't copy "give every company in the world an AEO score." Borrow three things instead: ① pre-build and score every entity in one vertical (e.g., every dental clinic in one country, every law firm in one city), with pages written for humans that can get indexed by Google, not just for crawlers; ② turn your own site's crawler logs into content and data, since that's something only you have, and write it up as "which AI reads which industry," which travels better than generic marketing copy; ③ package the low-score lists as sales leads for agencies, sold per record. Validate with three to five paying agencies first, then scale.

How to get users: HN, Product Hunt, the WordPress plugin, and the Chrome extension it tried brought almost no people. Don't count on launch platforms. A more realistic route is selling leads directly to marketing agencies in the vertical, and letting the businesses being scored "claim" their pages (it has this mechanism, usage unknown).

Timing: no rush. If you do want this "pre-built pages + crawler logs" setup, do it before model providers start downgrading programmatic pages like these. There are no public signs of that yet.

What would show I got this wrong: ① Engagemii discloses paying customers or revenue, or raises funding, which would mean someone is buying per-record AEO data and it's worth a second look; ② engagemii.com appears in Chrome's real-user top 1M bucket, meaning real people are arriving; ③ the reverse: if its rank falls back beyond the million mark once AI crawler reads slow down, that confirms the ranking was crawler-driven.

One line for everyone building products: AI crawlers read every page you spread out, so they can push your global rank up. To see whether a site has really taken off, look at real users and payments, not rank alone.

Data and sources

• Global rank: Tranco daily ranks (2026-08-20 to 11-05); CrUX global buckets (2026-08, 09).
• AI crawler reads: engagemii.com/research/ai-crawler-reads-2026 (published 2026-09-19) and the homepage live counter (11-07). All self-reported, methodology not verified.
• Product and pricing: engagemii.com homepage, /aeo/data, /about (11-07).
• Launches: Hacker News 48026866 (2026-05-05,1 points); Product Hunt (2026-05-06,2 votes); WordPress plugin page (583 downloads total).
• Competitors: TechCrunch / Bloomberg (2026-09-15, Profound); Peec AI's annual revenue is company-reported; other ranks from Tranco (early November).