Overview

This episode of *ハイパー起業ラジオ* (Hyper Entrepreneurship Radio) is a post-mortem analysis of the podcast's final season, centered on a detailed listener survey that received 384 responses. The hosts—IT critic Obara Kazuhiro and serial entrepreneur Kensuu—dissect every layer of their audience data, from survey design philosophy to NPS scores, free-text analysis, and platform metrics. The episode feels like a masterclass in data-driven content strategy, blending technical rigor with honest self-reflection about what made the show work and what they might have done differently.

---

0:00Survey Design: The Strategic Art of Question Ordering

The analysis begins with the survey's architecture, which Obara designed with deliberate care. The most striking choice was placing the Net Promoter Score (NPS) question—"How likely are you to recommend this podcast to a friend or colleague?"—at the very beginning, before any other questions. This was not arbitrary. Obara explained that asking about specific likes or dislikes first would introduce bias: if listeners first listed what they enjoyed, they might inflate their NPS score; if they listed criticisms first, the score would drop. By capturing the raw, unfiltered recommendation intent upfront, the survey preserved a truer measure of listener sentiment.

The demographic questions—age, occupation, and other attributes—were deliberately placed at the end. This is standard practice in survey design: personal questions can feel intrusive and cause respondents to drop out. By front-loading questions about content and value, the survey kept respondents engaged through the core material before asking for their personal information. Obara emphasized that the survey design itself is a mirror of the producer's priorities: in this case, recommendation intent and content evaluation came first, reflecting a deep respect for the listener's experience.

---

1:22Multi-Select vs. Best-One: Uncovering Broad Appeal and Killer Content

A key analytical technique in the survey was the dual use of two question formats for favorite segments and series: multi-select (check all that apply) and best-one (pick your single favorite). Obara explained that these reveal fundamentally different things. Multi-select answers show what is broadly liked—the "maximum common denominator" content that many people found pleasant or useful. In contrast, best-one answers reveal what *deeply resonated*—the "edge" or "killer content" that creates passionate fans and drives word-of-mouth promotion.

The results were illuminating. In multi-select, practical series like "Network Effects," "Recruit Edition," "Persona Edition," and "Pricing Edition" ranked high, along with early large-scale series. But in best-one selection, "Network Effects" became overwhelmingly dominant, and "Recruit Edition" also remained strong. Obara interpreted this as the lasting impact of early, deep, high-impact series. Meanwhile, shorter series like "Persona" and "Pricing" dropped in rank for best-one, and the relatively recent "Meta Edition" was popular in multi-select but not in best-one. This suggested that recency bias (the "new and shiny" effect) inflated its multi-select score, but it hadn't yet earned the deep loyalty of a best-one pick.

Most notably, the short series hosted by Kensuu—"Community Edition" and "Kougyou Edition"—performed exceptionally well in best-one selection. Obara attributed this to Kensuu's unique, experience-based perspectives: content that could only be heard from him, based on his personal history and failures. This reinforced the idea that irreplaceable, personality-driven content is what truly drives fan engagement and recommendation behavior.

---

3:32NPS and NPS Lift: Beyond the Overall Score

The podcast's overall NPS was 72.5, a world-class score (anything above 70 is considered exceptional). Obara expressed genuine gratitude even to those who gave low scores, noting that critical feedback is essential for growth. But the deeper insight came from NPS Lift analysis—comparing the NPS of specific subgroups (by age, favorite segment, etc.) against the overall average.

The "Network Effects" series had both high overall popularity and a positive NPS lift (+0.2 above average), an ideal pattern. The "Meta Edition" had a slightly negative lift (-0.2), suggesting it was well-liked but not strongly recommended. The most interesting finding was the "Management and Execution" series: its overall selection rate was low (not broadly popular), but its NPS lift was high. This is the classic profile of a "killer content" hidden gem—few people listened, but those who did became passionate advocates.

Age-based analysis revealed a surprising result: listeners aged 20–24 had an NPS lift of +2.3, far above average, while listeners over 50 had a negative lift. Obara hypothesized that younger listeners, not yet immersed in the "startup village" conventional wisdom, found the show's insider knowledge and fresh perspectives uniquely valuable. For them, the raw, non-textbook reality of the conversations was a revelation. The lower score among older listeners could stem from several factors: the content might not match their needs, or the show's abstract, jargon-heavy style might have been a barrier. Notably, when ChatGPT suggested adding explanations for technical terms to improve accessibility, Obara pushed back, arguing that the show's abstraction was part of its core appeal—dumbing it down could dilute its unique value.

---

7:45Free-Text Analysis: Capturing Listener Transformation

The survey's most powerful question was open-ended: "What insights or learnings did you gain from this podcast?" The response rate was remarkably high—219 out of 384 respondents answered, a strong signal that listeners felt they had experienced real change. Obara argued that the true value of any content lies not in information delivery but in the transformation it creates: what did the listener know or feel *before*, and what did they know or feel *after*?

The analysis revealed that Kensuu's concrete examples and his unique "gaze" (mono no mikata, or way of seeing things) were especially effective at grounding abstract business concepts in real-world situations. Listeners reported that the show didn't just inform them—it shifted their perspective, evolved their thinking, and gave them practical frameworks they could apply to their own work.

---

14:38Platform Data: Discovery Engines and Listening Habits

The survey also asked how listeners discovered the show. Kensuu's social media posts were the largest single source, but surprisingly, about 40% of listeners found the show through in-app recommendations on platforms like Spotify and Apple Podcasts. Obara noted this was higher than expected and highlighted a critical insight: platform algorithms are no longer just distribution tools—they are powerful discovery engines. For niche, specialized content, algorithm-based recommendations based on listening history may be more effective than traditional demographic targeting.

However, the data also revealed a gap. Spotify showed roughly 20,000 monthly impressions (how many times the podcast's title and artwork appeared on users' screens), but only about 3,500 of those led to new listeners actually starting an episode. The conversion rate from impression to listen was low. Obara expressed regret that they hadn't optimized episode titles and descriptions more aggressively for platform algorithms, which could have captured more of that potential audience.

Other engagement metrics were extraordinary. Total followers across Spotify and Apple were about 23,000–24,000. Monthly total listening time was roughly 17,000 hours, translating to about 45 minutes per follower per month. Even more striking: Spotify data showed cumulative listening time per follower was about 10 hours. This deep, sustained engagement indicated a highly loyal core audience.

Listening patterns also stood out. Episode retention rates were very high—about 80% of listeners finished a 30-minute episode. This contrasts sharply with YouTube, where half the audience typically drops off in the first 90 seconds. Podcast listeners, especially for this show, tended to commit to full episodes once they started. Early dropout (from Episode 1 to Episode 2) was initially over 50% but improved to about 40% over time, likely due to better recording quality and growing word-of-mouth. Crucially, dropout after Episode 3 was very low, suggesting that once listeners formed a habit, they stayed.

The show also exhibited a strong "catch-up" pattern: about half of listeners listened in real-time, while the other half binged past episodes over months. Episode play counts often peaked not at release but about six months later, demonstrating the long-tail value of evergreen content. Even the very first episode, now two years old, still gets about 30 plays per day (900 per month).

---

21:33AI-Powered Analysis: ChatGPT Classifies Listener Transformation

In a final analytical layer, Obara used ChatGPT to process the 219 free-text responses about learnings and insights. The AI classified the comments into several major themes: "understanding/evolution/clarification" (e.g., "my resolution improved"), "understanding the entrepreneur's mindset and struggles," "application to my own work," "gaining new perspectives," and even "entertainment value" (surprisingly, many listeners said they found the show fun).

Going further, ChatGPT performed a "before-and-after" analysis. It identified words listeners used to describe their state *before* listening (e.g., "abstract," "hard to apply," "couldn't get a bird's-eye view," "superficial understanding") and words they used *after* (e.g., "resolution improved," "systematic understanding," "motivation for career increased," "gained a manager's perspective," "became applicable to real work"). By mapping these word pairs, the AI visualized the specific transformation pathways the podcast enabled.

Finally, ChatGPT segmented listeners into tiers by engagement and expertise: "light layer," "learning layer," "practical layer," and "strategic layer." It then identified which keywords resonated most with each tier. For example, motivation-related words worked best for light listeners (useful for introductory content), while perspective-related words resonated with learning and practical layers (useful for retention and deepening engagement). This provided a roadmap for nurturing listeners through different stages of their journey.

---

25:02Actionable Insights and the Producer's Self-Reflection

The analysis distilled into several strategic directions. First, amplify strengths: double down on what already works—the killer content (like the Network Effects series) and the irreplaceable, personality-driven segments (Kensuu's unique perspectives). Second, address weaknesses carefully: ChatGPT suggested adding jargon explanations for older listeners, but Obara questioned whether that would dilute the show's core appeal. He also pushed back on a ChatGPT suggestion to add explicit calls-to-action, asking, "Would listeners of Hyper Entrepreneurship Radio really want that?" The lesson: don't blindly follow optimization advice if it undermines your identity.

Third, optimize discovery: improve titles and descriptions for platform algorithms to convert more impressions into listens, and leverage high-NPS-lift content to fuel word-of-mouth. Fourth, focus on transformation: the before-and-after analysis showed that the show's deepest value was in changing how listeners think and work—continuing to deliver that experience is the key to long-term loyalty.

Obara's personal reflections added a human layer. He admitted that his love of data analysis sometimes pulled him toward short-term optimization, distracting from his original goal of creating content that would still be valuable in 10 years. He also regretted not prioritizing title optimization earlier. He expressed renewed appreciation for Kensuu's unique contributions and for editor Higuchi's meticulous work. Most importantly, he emphasized that the producer's own enjoyment is essential: "If the performers aren't having fun, the audience will sense it's fake."

---

Conclusion

This episode matters because it demonstrates how to turn a project's end into a strategic asset. By combining thoughtful survey design, layered quantitative analysis (NPS, NPS lift, platform metrics), and AI-powered qualitative coding, Obara and Kensuu extracted actionable lessons that any content creator can apply. The episode's lasting impression is the tension between optimization and authenticity: data can guide you, but it should never override your core identity. The most successful content, this analysis suggests, comes from understanding your audience deeply while staying true to what makes you irreplaceable.

---

要点

• Place NPS questions first in surveys to avoid bias from later questions; save demographic questions for last to reduce dropout.
• Use multi-select questions to find broad appeal and best-one questions to identify killer content that drives passionate advocacy.
• NPS Lift analysis reveals hidden gems: content with low overall popularity but high recommendation rates among those who consumed it.
• Platform algorithms (Spotify, Apple) are powerful discovery engines; optimize titles and descriptions to convert impressions into listens.
• The deepest listener value comes from transformation—capture before-and-after language to understand what your content truly changes.
• Segment your audience by engagement level and tailor content to each tier's resonant keywords (e.g., motivation for newcomers, perspective for advanced listeners).
• Don't blindly follow optimization suggestions if they risk diluting your core identity; balance data-driven improvement with authenticity.
• The producer's genuine enjoyment is non-negotiable—audiences detect inauthenticity, and passion sustains long-term content quality.