For years, the objection to publishing a lot of content came as a fair question: who is actually going to read all this? Close behind it came the oft-recited wisdom about length. Keep it under 750 words. People will not read past the fold. Attention spans are short and getting shorter, so say less. We asked the same question about our own work, and for a long time it was the right one to ask.
It stopped being the right question, because the reader changed.
The rules were all about human limits
Every rule about brevity was a workaround for a person’s limited attention, reading speed, and patience for technical language. Those were real constraints, and the advice was sound for the reader it assumed. But the reader who matters most for a high-trust business is increasingly not that person.
Picture the moment that decides things. Someone gets your name from a person they trust. Before they call, they do what people now do: they open an assistant and say some version of, this firm was recommended to me, read their site and help me work out whether they fit my situation. The thing reading you then has no attention cap. It does not skim, get bored at 800 words, or trip over a clinical term or a compliance nuance. It reads everything you have published, holds it at once, and hands your prospect a summary in whatever register they asked for. Every reason to keep it short was a concession to a reader who is no longer the one deciding.
What the content is for now
We did not build a large, specific site only to rank, though ranking is part of it. We built it so that when an assistant reads us for a prospect, there is something thorough and true for it to read. The same body of work now does three jobs at once. It earns search visibility. It gives an AI enough real material to describe us accurately. And it still sits there as proof for the human who clicks through after the summary. The volume and specificity that used to look like too much are what the new reader needs.
We also buit a large, specific site to prove this system works and, frankly, because this is fun for us. If you ever hop on a call with us, we’ll happily show you the receipts.
The assistant is an honest closer
This raises the bar on the writing itself. You cannot flatter an assistant into recommending you. It reads what is there and reports it. So the move is to be, in specific and verifiable detail, the right fit for the people you serve, and to put all of that where the assistant can find it. Do that, and the assistant becomes close to an honest member of your sales team: it explains what you do, who you are for, and what working with you looks like, and it tells the prospect who is not a fit that you are not one. We are fine with that. Sending the wrong-fit prospect elsewhere early is a service to both of you, and it is the same thing we would do on a call.
This is not permission to pad
None of this rewards length for its own sake. An assistant is a ruthless reader of filler. Give it padding and it compresses that padding to nothing, or skips you for a source that said something concrete. The new reader rewards depth with substance and has no patience for the rest. We made the same point from the other direction here: a model can only pass along what is actually on the page. Write the complete, specific version of a topic. Do not write the long version of a thin one.
You are writing for two readers now
The practical shift is that you have two readers with opposite needs, and you have to serve both. The assistant wants completeness: every service line, every situation, the specific details a generalist would smooth over. The person who reads your prose after the assistant’s summary still has a short attention span and wants their exact answer fast. So the writing has to be deep enough to satisfy the machine and structured enough that a person can move through it: clear headings, real summaries, an obvious path to the thing they came for. Serve only the machine and you lose the person at the click. Serve only the person and you starve the machine.
Specificity became machine-matchable
This is why specificity stopped being optional. We have argued that in high-trust work, what converts is content that speaks to the reader’s exact situation, not the generic subject. The new reader raises the stakes on that. A specific page used to help only the one person who happened to share that exact problem. Now an assistant can line a prospect’s situation up against your content and conclude that you handle precisely this. Specificity became the signal a machine uses to confirm fit. The practice that treats its rarer, harder cases in real detail serves those patients and, at the same time, makes itself legible to the reader that now does the matching. It is the same reason a frightened parent will sit up doing a 1 a.m. deep dive on an adolescent program - the rare human reader who, like the machine, wants every specific answer you have.
Try it on us
If you are reading this with your eyes - a person, with the limits the old rules were built around - here is the test. Open the assistant you actually use, your daily driver, and tell it to read our whole site and say what you should read next and whether anything here speaks to your situation. We built the site to be read that way. If it points you somewhere useful, that is the whole idea. If it finds nothing for you, we would rather you hear that from a neutral reader than from us.
When you want to talk to the people behind the site, that part still works the old way. Book a 20-minute conversation.
Common questions
Does longer content rank better in search and AI answers?
No. Length is not the lever; completeness and specificity are. An AI reader compresses filler and rewards substance, so a thorough, specific piece beats both a padded one and a thin one.
Should I write for people or for AI assistants?
Both, because they read differently. The assistant reads everything and wants depth; the person who clicks through after its summary wants their specific answer quickly. Serve the machine with completeness and the person with structure: clear headings, real summaries, an obvious path to what they need.
Will an AI just flatten my content into the same summary as my competitors?
Only if your content is as generic as theirs. A model can surface only what is on the page. If you have published specific, verifiable detail about the situations you handle, that is what it repeats, and that is what sets you apart. Generic content is what collapses into the shared summary.