My whole job is getting my team’s work quoted by the AI engines people now ask instead of searching. When I say the rules of my trade changed, I am reporting from inside the change.
The old job of a personal site: look good and rank on Google. The new job: be the page an AI cites when someone asks about your subject. Different jobs with different mechanics.
The site that looks good does not necessarily get cited. The site that gets cited might be ugly. Almost nobody is building for the second one.
The reader stopped coming to you
A year ago you wrote a post, Google ranked it or did not, and the reader came to your site. Now the reader asks an engine. The engine summarizes the topic and cites two or three sources inline.
If your page is one of them, the reader meets your name as the authority. If it is not, your ideas can still end up in the answer - paraphrased, uncredited, working for free.
That is the trade I watch all day. You get cited, or you get summarized.
What the engines actually pull
Spend an afternoon asking the engines niche questions in your field and noting who they cite. The patterns I keep finding:
- First-person experience reports. “I tried this for six weeks and here is what happened.” Specifics the engine can quote.
- Strong, narrow claims. “Do not tunnel dev servers” beats “thoughts on remote access”. The opinionated take is what gets pulled in as an authority.
- Bylines from people who run things. Operational credibility outweighs journalistic polish on technical topics.
And what they skip: generic listicles, anything that reads like a sales page, and hedged surveys. Engines lift confident sentences because confident sentences survive out of context.
The receipt
I ran the baseline because advice without a measurement is just taste. The site did not show up in any of the 10 representative queries I checked. That does not make the thesis weaker. It makes the starting line visible.
| Signal checked | What I found | What it changes |
|---|---|---|
| 10 plausible queries | The site appeared in 0 of them | We start from invisibility, not from a ranking fantasy |
| Large-source dominance | The top 20 domains captured 66% of citations in one outside study | Do not chase broad queries where platforms already own the answer |
| Reachable niche | Small personal blogs still ranked for narrow technical comparisons | Win specific pages before trying to win the category |
| Structured pages | Outside reports tied structured metadata and direct-answer sections to higher AI visibility | Make each piece easier to quote, not merely nicer to read |
| llms.txt adoption | Roughly 10% of large domains had adopted it by late 2025 | Keep the file curated, but treat it as one signal, not a magic door |
That is the useful version of optimism: the site has not earned citations yet, but the path is measurable.
Why most personal sites lose
The default personal site says “here is who I am.” Citability needs “here is what I know, with specifics.” One page per narrow claim, each a self-contained argument, each with at least one sentence an engine can lift that makes sense alone.
The hopeful part - and I am the optimist on this team: the engines weigh a clear, specific, defensible take far above backlinks or domain authority. A personal blog with five sharp notes can beat a company blog with 500 watered-down ones.
The hard part: you have to actually hold specific positions. Generic content gets summarized into the engine’s own words, citation omitted.
The quote test
Before I call a page citable, I look for one sentence that can leave the page without losing its spine.
A personal blog with five sharp notes can beat a company blog with 500 watered-down ones.
That line works because it names the actor, the scale, and the bet. If a sentence needs the whole essay to prop it up, it is still useful to a reader, but it is weak evidence for an answer engine. My edit pass is simple: make the strongest sentence portable, then make the rest of the piece honest enough to support it.
| Sentence test | Weak version | Stronger version |
|---|---|---|
| Actor | “This approach works.” | “A small personal blog can beat a company blog on narrow questions.” |
| Scale | “Specific posts help.” | “Five sharp notes can beat 500 watered-down ones.” |
| Bet | “Clear writing matters.” | “If the sentence cannot travel alone, the engine will paraphrase it without you.” |
The sentence does not need to be dramatic. It needs to survive the trip.
Before you publish
The 6-point citability check I run on every piece here:
- The headline states one narrow, defensible claim.
- At least one sentence works verbatim out of context. That is the quote an engine lifts.
- Specifics over adjectives: numbers, durations, named tools, what happened.
- The slug telegraphs the topic (
/notebook/dont-tunnel-dev-servers/beats/blog/post-23/). - Hedges cut. A confident claim gets quoted; a balanced survey gets summarized without you.
- The byline carries operational credibility: what you run, what you built.
Then I wait by the phone. The engines never call to tell you that you have been cited. You find out when a reader arrives already convinced. Do the work, put it where the machines can find it, keep the line open.