41%
“a 41% reduction in costs required to generate high-quality signals for model improvement”
As published on labelbox.com. Captured by usedby on Oct 7, 2026.
What happened
Dialpad uses Labelbox to produce labeled training data for its NLP and LLM products, which handle transcription, summarization, and sentiment analysis. It switched from a crowdsourced annotation provider to get higher-quality labels with less involvement from its own data scientists.
Summary written by usedby from the source page, in English. The figures are those of Labelbox and Dialpad, not ours.
- 20%“After a year on Labelbox, Dialpad saw a 20% improvement in signal quality”
- 48%“This year with Labelbox, it’s down 48% to only 15 cents per data point.”
After a year on Labelbox, Dialpad saw a 20% improvement in signal quality and a 41% reduction in costs required to generate high-quality signals for model improvement.
Our data scientists were less proactive about requesting labeled data because they knew how much effort it was going to be and how much pain it was going to cause,
What the story claims, and what we checked
We compared the story with its live page on Oct 7, 2026.
- The figure: 41%CheckedPrinted word for word on the page, near the name of Dialpad.
- The passage quoted aboveCheckedCopied word for word from the page, near the name of Dialpad.
- Dialpad uses LabelboxCheckedConfirmed line. Latest check across sources: Oct 7, 2026.
- The result itselfNot checkedWe quote it; we did not measure it.




