Why Subtitling Quality Directly Affects Viewer Retention

When streaming platforms first started commissioning subtitles at scale, many treated them as a compliance checkbox. That assumption turned out to be expensive. Subtitle quality is among the top predictors of watch-through rate for non-native-language content.

What "quality" means for subtitles

1. Linguistic accuracy

The translation must be faithful to the source — not word-for-word, but meaning-for-meaning. Idioms need to be adapted, humour needs to be culturally transposed, and technical terminology needs to match what the viewer expects.

2. Reading speed and line length

Broadcast standards typically cap subtitle reading speed at 17 characters per second for adult programming. Lines should not exceed 42 characters. Violate these, and viewers either pause constantly or give up.

3. Timing synchronisation

A subtitle that appears 300ms before or after the corresponding speech breaks the sense of presence that good subtitling creates.

The cost of bad subtitles

One of our clients — a streaming platform with a catalogue of Korean dramas distributed to European markets — ran an A/B test between machine-translated subtitles and our professionally produced versions. In the German market, 7-day viewer retention for new episodes improved by 23% with professional subtitles.

Machine translation + post-editing: a viable middle ground

For high-volume content, MT + post-editing is a good compromise. Machine translation handles the first pass; a human subtitler reviews timing, adapts cultural references, and corrects errors. Quality isn't identical to full human translation, but it's far superior to raw MT — and 40-60% less expensive.