[Impl0x]

This is a difficult question because there are multiple motivations for making a new constructed language. For instance, Esperanto is designed to be a universal second language. It emphasizes ease of learning through very regular grammar, deriving most of its words from pre-existing roots, and a simple-but-powerful word construction. A language like Lojban, on the other hand, tries to eliminate linguistic ambiguity by basing its grammar on predicate logic.

So to clarify, are you asking how we'd go about designing a language that's optimized for universality? Or some other metric?

[chatmasta]

Let's say the goal is to optimize universal ease of adoption, and maximize virality.

Or, conversely, maybe you want to create a corporation where all employees must communicate in a custom language.

[Impl0x]

There are a few "easy" answers for ease of adoption, some of which I think have been mentioned by other replies. A phonetic alphabet with only one possible sound per letter, the ability to identify a word's part of speech given only the word, and an exception-free verb conjugation scheme are all common approaches for making a language easy to learn. Some constructed languages have a simple rule for determining which syllable in a word gets emphasis.

Also worth considering is making the language as "familiar" as possible to a new learner. You could decide what phonemes your language uses by looking at the most frequently used phonemes in the most popular languages. You might decide to go a level higher and base your language's word for universal concepts (food, day, sleep) based on roots that are common across many languages.

Another way to make a language easy is to reduce the number of words a person needs to know in order to use it in the real world. I'm personally a big fan of Esperanto's word-building system, which uses around 30 affixes (in addition to the suffixes that define each word's part of speech) that can be attached to a root word to modify or enhance its meaning. There's not a limit to how many affixes you can have, the word just has to make sense. Every time you learn a new root word, you get a huge number of new words for free by knowing affixes. A language that has such a system can get by with a reduced number of root words. For example: "laboro" is the word for the word "work" or "job" in Esperanto. By adding only one affix you can arrive at words like "labor-ej-o" (workplace), "labor-ist-o" (a laborer by trade), "labor-ant-o" (a person performing labor), and "labor-ind-o" (a job worth doing).

However, it's not really enough to just make a language easy. I'm glad you brought up virality; drawing people in is almost certainly the more difficult part. Once you have your ideal language, you need to build a culture around it. This is the same thing as when people judge the quality/utility of a programming language by its "ecosystem". Unfortunately this is largely out of the language creator's control. There's just too much work involved for a single person, or even a small group of people. There needs to be an abundance of learning material, lots of translated works and a growing body of original works (across as many genres/media/levels of proficiency as possible), and networks for speakers/learners to communicate and collaborate.