This report stitches together four product-level analyses of the NBA DTC Voice of the Fan survey, one per product, each rendered from a persisted data sidecar so every figure traces back to a source collection. The NBA App, NBA League Pass, and NBA ID carry two comparable waves and are read wave over wave; NBA.com is present for a single period and is reported as a baseline with no change read. Wave 1 was fielded in November and December 2024 and Wave 2 in March 2025, read from the response timestamps in the source exports, so the two waves sit about three months apart within the same season.
The two waves are independent samples: linking respondents on their persistent fan ID shows only about two percent overlap between Wave 1 and Wave 2 on the App, so significance is tested with independent-sample logic that the linkage now confirms rather than assumes, with movement treated as signal only when it clears an approximate 95% threshold and labeled directional when it falls just short. Theme movement is labeled directional rather than formally tested, since open-end coding does not support the same statistics as the rating items. The screening across many measures, themes, and cohorts was broad, so the narrative elevates only findings with both statistical credibility and substantive weight, and uses restrained language elsewhere.
Open-end metrics exclude the schema’s no-signal and catch-all values from substantive numerators, and each theme or driver share is computed on an explicitly defined eligible base rather than on all respondents. App Wave-1 themes were crosswalked into the consolidated Wave-2 App categories using the prompt-schema mapping; because that mapping is many-to-one, a Wave-2 bucket is not equivalent to any single Wave-1 theme, and App theme movements are read on that crosswalked basis. Per client convention, the body reports percentages and mean scores and avoids raw response counts; where a subgroup base is too thin to trust, the point is made qualitatively rather than with a count.
A few instrument and coverage notes bound the interpretation. Stream quality is reported separately from the core battery everywhere because it uses a different scale, an Excellent-to-Poor rating with an explicit not-applicable option for fans who do not stream, which also makes its eligible base shift between waves as more or fewer fans use it. The App’s unique-content and content-features items were verified against the source questionnaire and are correctly and distinctly worded, so their flatness reflects fan opinion rather than an instrument error. NBA ID uses an identity-and-loyalty battery that is conceptually, not directly, comparable to the video products, and its Wave 2 open-end base is too sparse for a confident theme read. Finally, Wave 2 was scoped by the client to the rating battery and open ends filtered by NPS group; the profile dimensions used for composition were not appended to this delivery but remain joinable by fan ID, so composition is read from the Wave 1 baseline and the by-cohort view is available to re-append for future waves.