Research Service Design Strategy

A long-standing client posed a unique challenge: Can we use unstructured VOC feedback (captured via their site survey) to identify CX drivers of NPS? The analytics team knew this was no simple ask. Traditional driver analysis requires formal surveys where customers provide numeric ratings of their experience across 20 or more pre-defined dimensions. In contrast, the client’s survey was much easier for customers to complete and placed no a priori constraints on how CX was measured. But those virtues came with a cost – their VOC data consisted of one open-ended comment and NPS rating per customer. Our solution to this challenge would involve the creation of an entirely new NPS service.


Fortune 500 Company


NPS and CX-Strategy Prioritization




  • Extract CX drivers of NPS from unstructured customer comments
  • Inform broader strategy by providing a principled means for prioritizing CX improvements


  • Sentier analysts used historical VOC data to identify a stable, emergent ensemble of CX thematic “tags”.
  • We then applied that ensemble to newly collected VOC data by hand-tagging comments with one or more themes and associated sentiments
  • The research team developed a novel quantitative model to accurately capture how likelihood-to-recommend depends on both CX thematic content and underlying sentiment
  • The analysis and model were instantiated within a client-friendly, tunable ROI “simulator" (delivered via Excel). This simulator empowered the client to freely explore and test out a wide range of hypothetical improvement efforts and to drive CX strategy in terms of feasibility and predicted return (i.e. NPS improvement)