ABSTRACT The avalanche of available unstructured text data makes it ever more challenging for innovation practitioners (and academics) to extract meaningful insights from such data. Topic modeling can support these efforts and help spur innovation. The current study reviewed 1099 innovation management articles to identify and compare the most frequently used probabilistic topic modeling approaches for innovation. In an effort to contextualize the suitability of these approaches, we develop a framework to organize existing topic modeling applications along the different innovation stages (i.e., idea generation, development, and commercialization) and innovation research in general. By zooming in on the three innovation stages, the authors showcase how topic modeling can spur innovation within each stage and highlight the future potential of the specific approaches. To further assist in capturing the various dynamics in complex unstructured text datasets, we illustratively apply a tailored topic modeling configuration to 1444 Journal of Product Innovation Management articles (1984–2023) to identify emerging, stable, and mature topics, as well as looking at their respective impact. This demonstration could serve as a starting point or blueprint for innovation practitioners and researchers seeking to combine the advantages of several topic modeling approaches. We conclude by offering a future outlook, including a forward‐looking research agenda. Taken together, our study offers guidance to and equips innovation practitioners and academics to design distinctive topic modeling procedures to best serve their intended purposes. If deployed appropriately, topic modeling helps users extract a wealth of unique, unprecedented insights from a continuously expanding source of data.
Stored only when both papers are in the corpus — 7 citations exist upstream.