It measured under the ceiling but still distorted

Your meter is telling the truth about the wrong thing

A digital file does not store a waveform. It stores measurements of one, taken tens of thousands of times a second. Everything between those measurements is reconstructed later, by the converter in your interface or the decoder on someone's phone.

Most DAW meters show you the highest sample. That is not the highest point the waveform reaches.

Every sample (the dots) sits below the ceiling, so a sample meter reports no problem. The reconstructed waveform between them, which is what actually comes out of a converter, rises straight through it. The circled crest is the peak nobody measured.

This is not a rare edge case. When a peak falls between two samples, both samples can read well below the real thing, so a file that measures -1.0 dBFS on a sample meter can still go over 0 dB on playback.

That reconstructed level is the true peak, measured in dBTP. With the TP safety limiter on, HardStop's Ceiling is set in dBTP; with it off, the Ceiling reads dBFS and only holds the samples.

Why it matters more after you upload

Two things make it worse downstream:

Lossy encoding. MP3 and AAC do not reproduce your waveform exactly, and the conversion can add peaks of its own. A master sitting at 0.0 dBTP can come back from an encoder clipping.

Every platform re-encodes. Spotify, Apple Music and YouTube each run your file through their own chain. You do not get to test all of them, so the margin has to be in the file you send.

The fix

Set CEILING to -1.0 dBTP and turn on the TP SAFETY LIMITER. With it on, HardStop keeps the true peak under your ceiling, not just the samples, as long as Out Gain stays at 0 dB or lower and, in Pro, Mix stays at 100%. If you are pushing for a competitive level, -0.2 dBTP leaves less margin but works.

That is the loud-and-safe choice. Plenty of commercial releases run true peaks above 0 dB on purpose, for the most aggressive sound, and that is a valid call too. TP safety limiter and the true-peak ceiling covers both choices, what the limiter does, and its latency.

A three-step check before you export

Set Ceiling to -1.0 dBTP (or -0.2 if competitive) and turn TP SAFETY LIMITER on. That alone solves most of it.
Watch the true-peak meter, not the sample meter. HardStop Pro shows LUFS and true peak together, so you can see loudness and headroom in the same glance.
Export, then check the bounced file with a true-peak meter. If it reads above your ceiling, something after HardStop in the chain is adding level.

Still seeing overs you cannot explain? Email support with the sample rate you exported at and the meter reading you are seeing.

Still stuck? Email support with your error number.