you decide on the sounds used within the language- the consonants, the vowels
then you begin making words out of those allowed sounds
you assign meaning
as your lexicon grows you will eventually have enough to make example sentences, and you can start working out your conlangs grammatical structure
http://www.zompist.com/kit.html <—- this guy helps a lot
Think about the sort of people who would speak it. Those people make sounds (or gestures, even) that they then agree refers to a thing.
If they only have one mouth that can form one word at a time, then they prioritize the groups of word functions. This creates grammar, but because usually people don't sit down and think about the groups of words and classify them—they just go ahead and talk, and implicitly agree that so-and-so is the best way to say it—the rules of grammar might not be as strict. There might be exceptions or variations to an adjective coming before or after the noun that it describes.
Writing could begin from pictograms, or as representing the sounds. A friend of mine said that Greek scripture came out of a culture full of merchant traders who had to jot things down quickly, which was why the writing system is actually very practical.
Another friend of mine observed that a number of Southeast Asian languages didn't have gendered pronouns even though our colonialists brought in the patriarchy (and we let the patriarchy stay), because there used to be no property designations based on gender among pre-colonial people, so it didn't matter whether a thing belonged to "him" or "her".
So, the language might grow out of the culture. But nobody during this process thinks, "Ooh, we're so cultural! Let's put this quirk into our language!"
Then there's the way that words change meanings. There's a word in a Philippine dialect that I cannot recall, where a word that means "relaxation" actually means "swinging" as in somebody sitting in a tree branch and swinging their ankle. It's very onomatopoetic, even though it might now refer to spending the night out with friends who wouldn't make that visual sound effect and might not even be relaxed in the sense of "exhausted" or "serene"…a good parallel with English would be the phrase, "hanging out."
This could also affect naming conventions, just throwing that in there. I'm no anthropologist, but I like to think the names "Brenda" (sword) and "Hilda" (…sword hilt?) might speak to what the martial Norse people expected of their little girls.
I was wondering how does a language (And I mean a Conlang, not a Cypher Language) get made in the fist place.