The Invisible Reader: What 682,000 Hits Taught Me About Who's Actually Reading My Website
- Rich Washburn
- 15 hours ago
- 6 min read
I want to be upfront about something before I get into the numbers: this isn't a flex. I'm not going to tell you my human traffic, and it's not because I'm hiding something impressive — it's because the human numbers aren't the point. The point is a number I wasn't tracking, wasn't expecting, and honestly didn't think mattered until I actually looked at it.
Over the past twelve months, AI bots hit my website 682,157 times.
Not humans. Not search crawlers indexing pages for Google results. AI user bots — the kind that get triggered in real time when someone, somewhere, asks ChatGPT or Perplexity a question and the model goes looking for an answer. That's 682,157 separate moments where a machine showed up at my domain, looked around, and decided whether what it found was worth carrying into a conversation happening somewhere else entirely.
I've been building and thinking about AI systems for a long time now. I did not expect this. I'm a little humbled by it, honestly. And I think there's a decent chance you don't know this is happening on your own site either — even if you're deep into AI, even if you've built agents, even if you think you've got a handle on how this stuff works. This part snuck up on me.
Your Website Has a Second Audience You've Never Designed For
For as long as websites have existed, the audience was a known quantity: a human being, sitting behind a screen, scrolling. You designed for their eyes. Layout, typography, color palette, the "vibe" of the thing — all of that exists because a human is going to look at it and form an impression. But there's a second audience now, and it doesn't have eyes. It has infinite memory, no patience for typography, and an entirely different reason for showing up.
AI user bots don't care about your color palette. They're indifferent to design in a way that should honestly be a little insulting if you've spent any time on your site's aesthetics. What they're hunting for is the "meat" — the raw, machine-readable text they can extract, synthesize, and hand back to a human on the other end of a completely different conversation.
Here's the distinction that matters: a human reads your page because they chose to visit it. An AI bot reads your page because a human asked a question somewhere else — in ChatGPT, in Perplexity, in some other interface entirely — and the model went looking for source material to build its answer. Your site isn't the destination anymore. It's raw material for an answer that gets delivered somewhere you'll never see.
The Numbers, Not as a Brag But as a Wake-Up Call
Over the twelve-month window from August 2025 to July 2026, here's what happened on my site: Total AI bot hits: 682,157. Unique pages crawled grew from 8,431 to 11,427 — a 36% increase. These bots aren't just hammering my homepage. They're digging into the archives, the old posts, the corners of the site I forgot I'd written. Monthly crawl volume rose 25% year over year, peaking in March 2026 at 78,635 hits in a single month — which, by the way, lines up suspiciously well with a major model refresh cycle. When a big model gets retrained or updated, it seems to go back and re-ingest its sources.
The deep-dive number is the one that actually stopped me: 5.3 hits per unique page, on average. That's not a bot pinging a URL once to confirm it exists. That's a bot going back into the same piece of content multiple times — pulling text, metadata, structure, links — the way you'd study something you actually intend to use, not just glance at. And between June and July 2026, something interesting happened: total hit volume stayed almost flat, but unique pages reached rose by 5.4%. The bots weren't crawling more. They were crawling smarter — reaching further into content they hadn't touched before, like they were completing a map rather than repeating a search.
Perplexity, Come On
I use Perplexity constantly. It's one of my go-to tools when I need a sourced, synthesized answer fast. So finding out that Perplexity's bots are treating my own site as a source worth returning to, repeatedly, over a twelve-month period — that one got me. I'm sitting here using Perplexity as my research tool, and apparently Perplexity is, in some small way, using me as one of its. That's not ego talking. That's just a genuinely strange loop to sit with. The tool I ask questions to is also asking my website questions, in its own way, on behalf of someone else.
OpenAI's bots showed up too, and for the same underlying reason — not to index me the old-fashioned Google way, but to retrieve specific pages in real time when someone asks a question that my content might help answer.
Two Totally Different Reasons to Crawl a Site
There are two distinct reasons a bot shows up on your website, and they are not the same thing: Conventional search indexing — the old Google-style cataloging, building a map of pages for future search results. This has existed for decades and isn't new.
Conversational retrieval — this is the shift. A human asks an AI a question somewhere else entirely, and the AI's bot goes out, finds your page in real time, and pulls from it to construct the answer it hands back. Your page becomes an ingredient in an answer delivered in a completely different conversation, on a completely different platform, that you'll never see happen.
The audience is no longer just the person reading your page. It's also the machine deciding whether your page should become part of the answer.
Why This Actually Matters, Beyond the Numbers
Here's the part I think people miss, even people who are deep into AI and using it constantly in their own work: your website has quietly become source material for a system you don't control and can't directly see the output of. It's not a metaphor. It's happening at 682,000-hits-a-year scale, on one modestly sized personal site. Scale that thought out across every business site, every blog, every publication that isn't actively thinking about this, and you start to see the shape of something big.
Traditional metrics — time on page, bounce rate, session duration — are built for the human reader. They're meaningless to a bot. A bot doesn't bounce. It doesn't spend "time" on a page in any way that metric was designed to capture. If you're only tracking human-centric analytics, this entire audience is invisible to you. You could have hundreds of thousands of machine interactions happening on your site and never know it, because you were never looking at the right numbers. The metrics that actually matter for this audience are different: unique pages crawled, crawl depth, hits per page. Not because the old metrics stopped mattering for humans — they didn't — but because there's a second scoreboard running in parallel, and most people have never once looked at it.
And the practical implication is straightforward: if AI bots are hunting for "machine-readable source material and semantic structure," then clean writing, clear headings, and logical site architecture aren't just good practice for humans anymore. They're the thing that determines whether your ideas get chosen as the answer when someone, somewhere, asks an AI a question your content could actually help with.
The Part That's Actually Interesting
I'm not writing this to tell you to go check your own analytics, though you should — you'll probably be surprised by what you find. I'm writing this because it's a genuinely strange moment in how information moves through the world. A website used to be a place you sent people to. Increasingly, it's a place machines visit on other people's behalf, extract value from, and never tell anyone about — except in the sense that somewhere, right now, someone is getting an answer from ChatGPT or Perplexity that includes something I wrote, and neither of us knows it happened.
That's the invisible reader. Not a glitch, not background noise, not something to be alarmed by. Just a new and mostly unacknowledged fact about how content works now — one that I think most people, even the ones building and using this stuff every day, haven't actually looked at yet. Go check your own numbers. You might be as surprised as I was.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.