In one sentence: A token is the small unit of text, usually a short word or a piece of a longer word, that an AI language system counts as one step when it reads a page or writes an answer.
People read in words. AI language systems read in tokens, which are pieces of text a bit smaller than words. A common word like "roof" is usually one token on its own. A longer or rarer word gets split: something like "weatherproofing" may come apart into two or three pieces, and a part number or an unusual brand name can break into several. Spaces and punctuation get counted too.
None of this is visible to you. It happens inside the system, before it starts matching anything or writing a word. The reason the term reaches your world at all is that tokens are the unit everything else gets measured in: how much text a model can hold at once, how much of your page it works with, and what an AI service bills you for.
An everyday way to picture it: a delivery van is not measured in boxes, it is measured in cubic feet. Tokens are the cubic feet of AI text. Two pages with the same word count can take up different amounts of room depending on which words are in them.
Plain everyday English is the cheap kind. Common words tend to cost one token apiece. Trade jargon, model numbers, long web addresses, and anything with unusual punctuation are the expensive kind, because the system has to chop them into smaller pieces it recognizes. A spec sheet full of part numbers can take up more room than a longer stretch of ordinary sentences.
You are never going to sit down and count tokens. What you will feel is the ceiling. When an assistant answers a question about your trade, it is not holding your whole website. It is holding a limited amount of text: some of your page, some of other sources, the question itself, and its own instructions. Anything past that limit is not in the room.
That rewards a plain habit. Answer the question early, in ordinary sentences, on a page about that one thing. Three hundred words of throat clearing before the answer costs room and buries the part worth quoting. The same goes for the boilerplate paragraph repeated on every page, which eats space without adding anything a system has not already seen.
Tokens are also the meter on AI services. Vendors usually price per token and count both what goes in and what comes back, so a chat tool that pushes your entire site into every single request costs more to run than one that sends only the relevant part. That is worth asking about if you get quoted a monthly figure with usage billed on top.
The ceiling also explains a frustration owners run into. You wrote the answer, it is sitting on your site, and the assistant still summarizes somebody else. Being on the page is not the same as being in the room. The passage has to be findable, short enough to travel, and clear enough to hold up once it is lifted away from everything around it. The longer version of how AI answers get assembled, and what that means for the pages you own, sits on the AI search hub.
A context window is a token budget and nothing more: the most tokens a model can consider at one time. Content chunking is the practice of breaking a long page into passages that fit comfortably inside that budget. A prompt is paid for out of the same budget, which is why a long rambling question leaves less room for your page.
No. There is no token report to check and no setting to change on your site. The only place the number touches you directly is an invoice from an AI service that bills by usage. Everything else takes care of itself if your pages answer a question early and stay on one subject.
No. A keyword is a phrase a person searches for. A token is a mechanical slice of text with no meaning attached to it on its own. The word plumber can be a keyword and a token at the same time, but the two ideas are doing completely different jobs.
AI search and AIO · Context window · Content chunking · Large language model · Prompt · All glossary terms · Plain-English answers
Free consultation, plain-English advice. If you don't need us, we'll say so.
Book a free consultation → Or call/text directly: (407) 694-2055Tell us a little about the business and we will come back with an honest read: what we would fix first, what it costs, and whether you need us at all. Prefer to see work before you talk numbers? Get a free homepage mockup, built for your business, yours to keep either way.
Brandon reads every one of these himself. You will hear back shortly with an honest read on what we would do first, what it costs, and whether it is worth it for you.