Spam-detection systems on every major platform learn from the same patterns. Pick a handle that triggers any of these and you'll get throttled, sent to spam folders, or quietly ignored.
Mistake 1: Real-name plus 8 random digits. "JohnSmith48273951" is the canonical bot pattern. The platform's spam filter learned that pattern from a million fake accounts. If you must include digits, keep it to 1 or 2 and make them meaningful (a year that means something, a lucky number).
Mistake 2: Two first names smashed together. "JenniferSarah", "MichaelDavid", "AmandaTaylor" — these read as auto-generated. Real humans have one first name. If you want a two-word handle, pair a noun with a verb or an adjective with a noun, not first-name with first-name.
Mistake 3: Underscore-padding. "x_Username_x", "_user_name_" — this was a teenage forum aesthetic that aged into a bot signal. One underscore in the middle is fine; bracketing underscores get flagged.
Mistake 4: All consonants. "Krzltng" or "Bvchsrm" looks like generated noise. Names need at least one vowel per 3 characters to read as human.
Mistake 5: Generic handle plus a long number. "user12847", "guest3892", "anon4421" — the platform reserved these patterns for default-generated accounts that never picked a real handle. Using one yourself signals "I'm a fresh sign-up that won't post much," and the algorithm treats you that way.
How PickName handles this: every style except "Tech" avoids long-digit suffixes by default. Generated names lean on adjective-plus-noun structure with at most a 2 or 3 digit suffix. You can also toggle "Cross-platform safe" mode, which restricts to the alphabet+digits+underscore band that survives every major platform.
Bonus: if you want to test a handle, search for it in Google with quotes ("johnsmith48273951"). If 70% of the hits are the same person on five different platforms, it's a real human. If the hits are all "this account has been suspended," learn the lesson.