Honestly, I'd divide it into five categories, two of which, if the dev team knows what they're doing, shouldn't even exist outside of pre-release playtest builds.
Tier 0: God tier. So good that its very existence makes anyone below this tier no longer viable. Their matchups against everything else are so one-sided that skill is much less of a factor than in other matchups. Should be nerfed to Tier 1 during playtesting. If you want a game that's bat**** insane, bump things up to here (See: Brawl Minus)
Tier 1: Very viable. You probably won't need a secondary.
Tier 2: Viable, but you might need a secondary to cover your bad matchups
Tier 3: Not very viable in tournament play. You will definitely need a secondary to cover bad matchups, but they might shine in alternate rulesets.
Tier 4: Utter garbage. Unless you're a top professional, you're not even going to win in casual play with this character. Much like Tier 0, their matchups are so one-sided that skill is barely a factor, but the matchups are one-sided against them as opposed to in their favor. Should be buffed up to at least Tier 3, I'd even say up to Tier 2, during playtesting.
Stuff like Brawl's tier list is what happens when you fail to rework anything that ends up falling into Tier 0 or Tier 4. If these tiers are present in the final release, then it means that the dev team has either failed to adequately playtest and balance accordingly, or that they were unable to make the balance fixes necessary before they reached their deadline. I'd even go as far as to say that a dev team should work to at least minimize the number of Tier 3 characters, and devs like David Sirlin even suggest skewing the roster towards Tier 1 for one-on-one fighters.