How thorough?

Step 3 of the scan wizard, and the one that decides how long you wait. The subtitle states the trade honestly: "More thorough = slower, but more confident every match is an exact copy."

Thorough is pre-selected and badged Recommended.

The three levels

The times shown on the option rows are Trawler's own figures for a 250 GB scan.

Level On screen Time shown
Quick "Match on filename + size + a quick check. Good for music, images, videos." ~30 sec for 250 GB
Balanced "Quick, plus a closer check on every possible match. Catches documents and mixed content reliably." ~2 min for 250 GB
Thorough "Reads every byte of every possible match. Every match is guaranteed identical. Slowest." ~12 min for 250 GB

Your own times will differ — a slow drive, a network location, or a CPU usage cap set in Settings → Scanning & performance all change the answer, and so does Efficiency mode, which is meant to.

What each one actually does

Every level starts the same way, and this is why even a 250 GB scan is quick: a file whose size is unique is ruled out immediately, without ever being opened. Most files on a normal drive are eliminated here, for free. Only files that share a size with something else go any further.

After that the levels differ by how much of those remaining files gets read:

Step What it reads Run at
FAST CHECK The first 4 KB and the last 4 KB. Files under 8 KB are read in full here — there is nothing to sample. Quick, Balanced, Thorough
CLOSER CHECK First and last block again, plus extra blocks spaced through the middle of the file. Files under 64 KB have no middle worth sampling and are compared on the two ends. Balanced, Thorough
FULL CHECK Every byte. Thorough only

Each step only looks at files the previous one couldn't separate, so the work shrinks as it goes. That is the point the wizard's Why are big jobs fast? panel is making:

"Trawler reads the first and last block of each file before reading the rest — so a 100 GB scan finishes in seconds, not hours. Choose the depth that fits your files — you can rerun with a different setting anytime."

Which one to pick

Thorough is the honest default. It is the only level where "these are the same file" is a fact rather than a very strong indication.

Quick is fine for images, music and video, because those formats are dense and varied — two different clips almost never share a size and the same bytes at both ends. It is a poor choice for documents, virtual disk images, or anything with a long shared header and a large body of padding, where two files can look identical at the ends and differ in the middle.

Balanced is the middle: it samples through the body as well as the ends, which is what catches the document case. It still does not read everything.

Changing your mind afterwards

You do not have to re-run the whole thing.

On the Duplicates tab, Verify exact copies appears whenever the results came from anything other than a Thorough run. It re-checks the groups against the full file contents, and its tooltip says what it is for:

"Matches so far were confirmed by sampling. Re-check the remaining groups against the full file contents. Your ticks are kept."

Nothing already read is read again — only the files that are still tied get opened. The same is true of running the scan again at a deeper setting: the work already done is kept.

Related