Anybody who has looked at search logs for a site knows the same thing: people do not type what you expect. They type fragments, they misspell, they use words from a different tradition, and they very rarely use the terminology that the people making the work use among themselves. The gap between those two vocabularies decides a great deal about who ever finds anything.

This is worth understanding without cynicism. It is not about gaming anything. It is about the ordinary problem that a work has a name its maker chose and readers are looking for it with words they already had, and somebody has to bridge that.

Fragments, not sentences

Most searches are two to four words. They omit articles, they omit context, and they assume the system will work it out. A reader looking for a specific kind of drawn scene types the two or three words that feel most distinctive to them, which are frequently not the words a creator would have chosen to describe the same thing.

This means long, careful titles are almost never matched directly. What gets matched is the surrounding text: descriptions, tags, and the ordinary words used in a caption. A work whose only text is an evocative title is close to invisible, however good the title is.

Spelling is handled unevenly

Large search engines correct spelling aggressively and have done for years, so a misspelled query usually still finds the right thing. Internal site search on smaller platforms often does not, and neither do many tag systems, which match exactly or not at all.

The practical consequence is that platform-native discovery is far more fragile than web search. On a site with strict tag matching, a reader using a variant spelling or a synonym finds nothing, concludes the work does not exist, and leaves. That is a discovery failure nobody involved ever sees.

Creator vocabulary versus reader vocabulary

Creators develop precise internal terms for genres, techniques and subgenres, and those terms circulate among people who make things. Readers arrive with the words they learned from wherever they first encountered the material, which is usually a completely different set with different history, as described in the piece on where the words came from.

Neither vocabulary is wrong. The mistake is assuming they overlap. A creator writing descriptions entirely in the internal vocabulary has effectively addressed other creators, which is a much smaller group than the one they were hoping to reach.

Writing descriptions people can match

The fix is unremarkable: describe the work in plain, ordinary words, including the ones you would not use among colleagues. Say what is in it, what kind of thing it is, and what a reader would call it. That does not require keyword stuffing, which reads badly and works poorly; it requires writing a sentence a stranger could match.

Consistency helps as much as vocabulary. Using the same terms across every work builds a body of text that search systems can associate with you, which is the same argument for consistent tagging made in the piece on how tags route readers.

What not to do

Padding descriptions with unrelated popular terms is the obvious temptation and it fails on both counts: platforms penalise it and readers who arrive on a mismatch leave immediately, which most systems treat as a negative signal. It also erodes trust with the audience that would otherwise have stayed.

The same applies to misspelling deliberately in the hope of catching typo traffic. Search engines already handle that, readers find it faintly embarrassing, and it makes an account look like a spam operation, which is not a reputation that pays off later.

Making work findable without tricks

Write plain descriptions, use the words readers use, be consistent, tag honestly, and give every work enough surrounding text to be matched by something other than its title. That is the whole method, and it is closer to writing well than to any technical practice.

Everything published here involves fictional, unmistakably adult characters, and the limits are set out in the reader agreement. Corrections are welcome through the letters page.