This question is from my google analytics log, i.e. someone got to my blog by search for this question. I’ll answer the question using catlangs (fake languages designed to be spoken by cats using a domestic house cat’s vocal apparatus, but a humans knack for languages, in a hypothetical world where everything is exactly the same as this one, except cats talk a catlang)
Pick a target audience.
Your goals will probably guide your decision about who your target audience is. Conlangs tend to draw the fans intended, not just generalist language hobbyists. Klingon got a science fiction fans that otherwise wouldn’t mess with languages. The remnant language revivalists get fans from descendants of the original language speakers. (Remnant languages are so poorly documented that revive essentially means construction almost from scratch) I think a well developed hypothetical catlang would attract a disproportionate number of cat fanciers.
Is your conlang under a glass case?
To use a language is to subtly change it, either intentionally or because it’s just hard for recreational language learners to follow any language specification. Some percent of people who study conlangs are mostly interested in the creation of conlangs, especially derivatives, rather than attempting to use a language.
If your conlang is under a glass case, i.e. you don’t really want people trying to write it, change it, etc, then you are promoting the entertaining reference grammar and dictionary you wrote. You aren’t promoting the *language* per se, at least not in the same way that the Alliance Francaise is promoting the use of French as a second language. Promoting a reference grammar for amusement is the same as promoting any item of literature, science fiction, or what ever category this falls in.
Do you have a business model?
Some conlangs do. For example, Globish is a condialect of English that isolates some easy subset of English that can be used among non-native speakers to get to the point of speaking and understanding faster than if they were dealing with full blown English. The business model is selling educational materials, translation services, etc, and all the principles of ordinary marketing apply.
Do you have a governance model in mind?
As for a catlang, I presume this is a free thing, so it’s more of a non-profit endeavor. Then clubs, such as the KLI would be relevant. The conlang clubs provide the means to route fans interested in something to other fans. Language normally aren’t used in a vacuum, unless you are writing the chemical conlang for dust mites– a reference grammar I’d like to read.
I admit, for most brand new conlangs, the idea of having a governance model is laughable, because what is there to govern with only 2 fans? Still, people like to join clubs and they like to get with the program. If there is no program, then, well, they mill about and leave to go read more Lolcat captions. Examples would be the Lojban community that really did make best efforts to follow up on the original research agenda, for example by constructing machine parsers, etc. The Esperantists really did follow up on the original social agenda, by holding conventions, traveling, marrying other Esperantists, etc.
A hypotethical catlang promoter would be well served by creating the Catlang Association and hosting an annual sing along…. Mraaaaaaaaaaaauuuuuuu! Mrau! Mrauuuuuuuuuuuuuuu! And of course, overseeing a cat breeding program that will select for cats of usual linguistic ability instead of selecting for cats with softer fur.
Technical Features are Overrated
Notice I haven’t said squat about working on the technical details of your conlang? I think people recognize well promoted conlangs. The human brain can cope with all sorts of communication systems, elegant, ugly, what ever. If you build a better conlang, there is no particular reason to think fans will flock to your website. A technically “good” (on what ever criteria) conlang may hold the interests of some fans, or server them better, but the technical qualities matter *after* you have an established fan base.
Pick a Empty Playing Field.
No one, AFAIK, has done a conlang based on the chemical communication signals of ants. No one has done a conlang based on the hypothetical communication skills of crows or ants. There isn’t even a portmanteau for such conlangs (they aren’t really auxlangs, artlangs, or loglans).
But there are too many players in the field of auxlangs, there are too many players in the field of artistic conlangs that are trying to achieve the same thing as Elvish, and so on. The world of conlangs is so big, that unless you really enjoy writing derivative conlangs and aren’t interested in conlang promotion, one shouldn’t let their conlang fall into a category dominated by one of the big three (Esperanto, Lojban, or Elvish)