Open ends are where the richest insight in a study lives, and where the timeline goes to die. Reading hundreds or thousands of verbatims, building a code frame, and applying it consistently is slow, expensive work, which is why so many agencies ship it offshore. Add multiple markets and languages and the problem multiplies: now you're coordinating coding passes in several languages and trying to reconcile frames that don't quite line up. The result is a step that holds up the whole study and often gets under-resourced precisely because it's painful.
It's a good candidate for automation, as long as the automation keeps the nuance and keeps the researcher in control.
Why open-end coding is a bottleneck
Coding verbatims by hand is genuinely hard. Someone has to read everything, infer the themes, build a frame that captures them without exploding into a hundred near-duplicate codes, and then apply it consistently across every response. It's labor-intensive and judgment-heavy at the same time, which is the worst combination for throughput.
Across markets it gets worse. Each language often gets coded separately, by different people, into frames that then have to be harmonized so a global client can compare markets. That reconciliation is where time and consistency both leak. Two markets answering the same way can end up in different code buckets simply because two coders built two frames.
Auto-coded, editable, pooled across languages
The modern approach handles the heavy lifting automatically and then hands control back to you. Responses are auto-coded into a draft frame, the themes surfaced for you rather than built from scratch. The frame is fully editable, so you refine it, merge codes, split a theme, rename for the client's vocabulary, with your judgment shaping the final structure.
The part that matters most across markets: the coding is pooled across languages. The same thematic frame applies across every language in the study, so a respondent in one market and a respondent in another who say the same thing land in the same code, regardless of the language they said it in. You get cross-market comparability by construction, not by a painful reconciliation pass at the end.
The researcher still owns the frame
This is automation that accelerates the researcher rather than replacing the judgment. The machine does the first read and the consistent application across thousands of responses in many languages, the part that's mechanical and error-prone by hand. You do the part that requires expertise: deciding what the themes actually mean, how granular the frame should be, and how to present it so it answers the client's question. The draft is a starting point you control, not a black box you accept.
What it changes for the firm
Practically, the step that used to be a multi-week, multi-vendor bottleneck becomes part of the same fast turnaround as the rest of the study. You stop outsourcing coding and waiting on it; you stop reconciling mismatched frames across markets; and you keep the qualitative richness that makes open ends worth asking in the first place. The nuance survives, the comparability improves, and the timeline stops bending around the coding pass. Open ends go back to being the best part of the study instead of the slowest.