AI-Assisted Analysis Disclaimer
This entire project was built using Claude Code by Anthropic. The Python analysis scripts, HTML dashboard, and all data processing were developed with AI assistance.
This is not an exact representation of keyword tracking. The analysis uses simple substring matching and Named Entity Recognition which can produce both false positives (matching unintended content) and false negatives (missing relevant mentions). Context, sentiment, and semantic meaning are not analyzed.
The data should be considered indicative of trends rather than precise measurements. Always verify important findings by checking the source citations and listening to the original episodes.
On This Page
What is this?
The Podnews Weekly Report 2025 is an automated analysis of every episode of the Podnews Weekly podcast from January 2025 onwards. It provides data-driven insights into the most discussed topics, companies, people, and trends in the podcast industry throughout the year.
Podnews Weekly is hosted by James Cridland (Podnews Editor) and Sam Sethi (Podcast Business Journal). Each week they discuss the latest news, trends, and developments in the podcasting industry.
This project was created by The Podcast Setup Newsletter as a research tool for the podcasting community. It is not affiliated with or endorsed by Podnews Weekly.
How It Works
Transcript Download
Transcripts are automatically downloaded from the Podnews Weekly RSS feed at
rss.buzzsprout.com/1538779.rss.
The podcast provides SRT (SubRip) subtitle files for each episode through the Podcasting 2.0
podcast:transcript tag.
These files include timestamps for each spoken segment.
Text Parsing
Each SRT file is parsed to extract individual segments with their timestamps. The text is normalized to lowercase for keyword matching. Episode metadata (date, title) is extracted from filenames and matched to RSS feed entries to get episode URLs.
Keyword Analysis
A Python script scans each transcript segment for mentions of predefined keywords. This includes brands, platforms, hosting companies, podcast apps, notable people, and topic-specific terms for Podcasting 2.0, money/investment, video podcasting, and AI. Each match increments a counter for that term.
Citation Logging
Every keyword match is logged with its episode name, date, timestamp, and surrounding context (up to 200 characters). This creates a detailed audit trail showing exactly where and when each term was mentioned.
JSON Export
All data is compiled into a single JSON file (podnews_2025_report.json)
containing aggregated counts, citations, monthly trends, and episode metadata.
This file powers the entire dashboard.
Weekly Trend Calculation
For the top 15 entities in each category (hosting, apps, brands, people), the system calculates weekly mention counts. Each episode is assigned to an ISO week (e.g., 2025-W05) based on its publication date. This enables timeline visualization showing how mentions of specific entities rise and fall over time.
Visualization
The JSON data is loaded by a static HTML page and rendered using Chart.js. All charts are interactive - clicking a bar or data point navigates to the citations page showing the source evidence. The weekly trends page allows toggling individual entities on/off to compare specific trends.
Podcasting 2.0 Tags in Action
This report leverages several Podcasting 2.0 RSS namespace tags to enhance the episode listings with rich metadata. Podnews Weekly implements these open standards, allowing us to display more than just episode titles.
podcast:transcript
Transcripts
The foundation of this entire report. Each episode provides SRT transcript files via the
podcast:transcript tag. We download these
transcripts and analyze them for keyword mentions, enabling citation tracking with exact timestamps.
podcast:person
Guest Avatars
Episode guest information is pulled from the podcast:person tag.
This tag includes the person's name, role (host/guest), profile image, and link to their website or social profile.
In the episode list, you'll see guest avatars displayed as small square images with their first name below. Only guests are shown (not hosts, as James and Sam appear in every episode).
podcast:chapters
Chapter Artwork
The podcast:chapters tag provides a URL to a JSON file
containing chapter markers with timestamps, titles, URLs, and optional artwork for each chapter.
We fetch this chapter data and display chapter artwork thumbnails in the episode list. Each image represents a different segment of the show - interviews, news topics, tech updates, etc. Hover over any chapter image to see its title. Episodes with no chapter artwork won't show the "Chapters:" label.
Why This Matters
These Podcasting 2.0 tags demonstrate the power of open podcast standards. Because Podnews Weekly implements these tags in their RSS feed, we can build richer experiences without any special API access. Any podcast that implements these standards can benefit from similar enhanced displays in supporting apps. Learn more at podcastindex.org.
Charts Explained
Topic Trends Over Time
A line chart showing how often four major topics are discussed each month: Radio & Audio, Video Podcasting, Money & Investment, and Podcasting 2.0.
How it's calculated: For each month, we count the total number of times any keyword from each topic category appears across all episodes in that month. This shows actual mention frequency, not just whether a topic was discussed.
Top Podcast Hosting Companies
A horizontal bar chart of the most mentioned podcast hosting platforms.
How it's calculated: Each time a hosting company name appears in any transcript segment, it counts as one mention. Companies are ranked by total mentions. Click any bar to see every instance where that company was discussed.
Top Podcast Apps
A horizontal bar chart of the most mentioned podcast listening applications.
How it's calculated: Same as hosting companies - each mention of an app name counts once. Includes both mainstream apps (Spotify, Apple Podcasts) and Podcasting 2.0 apps (Fountain, Podverse).
All Brands & Platforms
A comprehensive chart of all mentioned brands, platforms, and companies in the podcast industry.
How it's calculated: Tracks a broader list of industry names including platforms, networks, tools, and services. Some overlap exists with hosting and apps charts since those are subsets of this broader category.
Most Mentioned People
People discussed on the show, automatically detected using AI.
How it's calculated: Uses spaCy's Named Entity Recognition (NER) to automatically detect person names in transcripts. This means any person mentioned gets tracked, not just a predefined list. Company/brand names are filtered out to avoid false positives. The hosts (James Cridland and Sam Sethi) are excluded from detection.
Why top 50? NER detects over 1,500 unique names across all episodes, but many are one-time mentions or less relevant. We display the top 50 most frequently mentioned people to focus on the key figures driving podcast industry discussions.
Topic Deep Dives
Four separate bar charts showing the most frequently used terms within each topic category.
How it's calculated: Within each topic (Podcasting 2.0, Money, Video, AI), individual keywords are ranked by frequency. This reveals which specific terms dominate each conversation - for example, whether "YouTube" or "clips" drives video discussions.
Weekly Trends
The Weekly Trends page provides a timeline view showing how mentions of specific entities change over time. Access it by clicking the Trends link next to any of the four main charts (Hosting, Apps, Brands, People).
How Weekly Trends Work
- ● Weekly Granularity: Data is aggregated by ISO week (e.g., 2025-W05 for the 5th week of 2025). Each data point represents the total mentions for that week.
- ● Top 15 Entities: For each category, we track the 15 most-mentioned entities across all episodes. Less frequent entities are not included to keep the charts readable.
- ● Interactive Legend: Click any entity name in the legend panel to toggle its visibility on/off. Use "All" and "None" buttons to quickly show or hide all entities.
- ● Category Tabs: Switch between Hosting, Apps, Brands, and People using the tabs at the top. Each category has its own set of tracked entities.
Use Cases
- • Track when a company or person became a hot topic in podcasting discussions
- • Compare competing platforms (e.g., Spotify vs Apple) over time
- • Identify seasonal patterns or news-driven spikes in mentions
- • See which hosting platforms are gaining or losing mindshare
- • Spot emerging trends before they become mainstream
Interpretation Notes
Weeks with no episode releases will show zero mentions for all entities. A sudden spike may indicate a major news event, acquisition, or controversy. Remember that more mentions don't necessarily mean positive coverage - always check the citations for context.
How Citations Work
Every data point in this report is backed by citations - direct references to the source material. This ensures transparency and allows you to verify any claim or explore topics in depth.
What's in a Citation?
- ● Episode Title: The name of the podcast episode
- ● Date: When the episode was published
- ● Timestamp: The exact time (HH:MM:SS) when the term was mentioned
- ● Context: The surrounding text (up to 200 characters) showing what was said
- ● Episode Link: A direct link to listen to the episode on Buzzsprout
How to Access Citations
- 1. Click any bar in any chart on the dashboard
- 2. You'll be taken to a citations page showing all mentions
- 3. Citations are paginated (25 per page) for easier browsing
- 4. Click the external link icon to open the episode
- 5. Click the timestamp to copy it to your clipboard
Categories Tracked
Podcasting 2.0
The next generation of podcasting built on open standards. Includes Value for Value, boostagrams, streaming sats, Lightning payments, RSS namespace features (chapters, transcripts, live), and Podcasting 2.0 apps like Fountain and Podverse.
Money & Investment
Financial discussions in podcasting. Includes funding rounds, acquisitions, revenue, CPM rates, sponsorships, advertising spend, valuations, and industry economics.
Video Podcasting
The visual side of podcasting. Includes YouTube, video-first strategies, clips, shorts, reels, TikTok, and discussions about whether podcasts should have video.
Radio & Audio
Traditional radio and audio-first discussions. Includes BBC, NPR, SiriusXM, iHeartRadio, terrestrial radio, commercial radio, talk radio, and the broader audio landscape that intersects with podcasting.
Industry Database
The Industry Database section uses a hybrid approach: automatic Named Entity Recognition (NER) to detect organizations, plus curated keyword lists for specific categories.
Industry Organizations
Key industry bodies and research firms: IAB, Edison Research, Podcast Academy, Sounds Profitable, Nielsen, Podtrac, Chartable, and major media companies.
Networks & Shows
Major podcast networks and notable shows discussed: Wondery, The Ringer, Barstool, Crooked Media, plus frequently mentioned shows like Joe Rogan Experience, Call Her Daddy, and industry podcasts.
Technologies & Standards
Technical standards and tools: RSS feeds, GUID, podcast namespace tags, ActivityPub, Mastodon, Bitcoin/Lightning, transcription tools (Descript, Whisper), and streaming protocols.
Industry Events
Conferences and events: Podcast Movement, Evolutions, Podfest, The Podcast Show, Radio Days Europe, SXSW, and other industry gatherings.
Keyword Lists
The analysis uses predefined keyword lists. Here are the terms tracked in each category:
Podcasting 2.0 Terms (29 keywords)
Money & Investment Terms (31 keywords)
Video Podcasting Terms (17 keywords)
Radio & Audio Terms (30 keywords)
Hosting Companies (70+ keywords, sourced from Livewire.io)
Podcast Apps (26 keywords)
People Detection (AI-powered, 1,500+ names detected)
People are detected automatically using spaCy Named Entity Recognition (NER), not a predefined list. This allows the system to discover any person mentioned in the transcripts.
The NER model detects over 1,500 unique names across all episodes. We store the top 50 most frequently mentioned people in the report, as many names are one-time mentions or less significant to podcast industry discussions.
The following are excluded to avoid false positives: podcast hosts (James Cridland, Sam Sethi), company/brand names (Spotify, Libsyn, Riverside, etc.), social networks (Mastodon, Bluesky), and common standalone first names.
All Detected Names
For complete transparency, here is the full list of every person name detected by our NER system. Names are sorted by mention count. Click "Load Names" to fetch the data.
Methodology & Limitations
Important Caveats
- • This is automated keyword matching, not semantic understanding
- • Context and sentiment are not analyzed - a negative mention counts the same as positive
- • Only predefined keywords are tracked - new terms may be missed
- • Transcript accuracy depends on the source (auto-generated transcripts may have errors)
Mention Counting
Each time a keyword appears in a transcript segment, it counts as one mention. A single episode can have hundreds of mentions of the same term. This reflects how often something is discussed, not just whether it was mentioned.
Case Sensitivity
All matching is case-insensitive. "Spotify", "SPOTIFY", and "spotify" all count as mentions of Spotify.
Substring Matching
Keywords are matched as substrings. This means "apple" will match in "apple podcasts" and "pineapple". We've tried to minimize false positives by using specific terms, but some noise may exist.
Person Detection & Filtering
People are detected using spaCy's NER model, which identifies person names in text. Over 1,500 unique names are detected, but we display only the top 50 most mentioned. James Cridland and Sam Sethi are excluded since they speak in every episode. Company names that NER sometimes misclassifies as people (like "Libsyn" or "Riverside"), social networks, and standalone first names are also filtered out.
Monthly Aggregation
The trends chart aggregates data by calendar month. Months with more episodes will naturally have higher counts. The number of episodes per month can vary based on the podcast's release schedule.
Exclusions & Filtering
To reduce noise and improve the quality of results, certain terms are intentionally excluded from the pre-built charts (Top People, Top Brands, etc.). Note: The Search Trends and Transcript Search features search the raw transcript data directly and do NOT apply these exclusions.
Excluded from "Most Mentioned People"
The hosts are excluded since they speak in every episode and would dominate the charts:
- • James Cridland - Podnews Editor, co-host
- • Sam Sethi - Podcast Business Journal, co-host
Additionally, company/brand names that NER sometimes misclassifies as people are filtered out (e.g., "Spotify", "Libsyn", "Riverside"), along with common conversational words ("thanks", "hello", etc.).
Excluded from Organization Detection
The following are excluded from automatic organization detection to prevent false positives:
- • The show itself: Podnews, Podnews Weekly, Podnews Daily
- • Major platforms: Spotify, Apple, YouTube, Amazon, Google, Facebook, TikTok, Netflix, Meta, Twitter, Instagram (tracked in dedicated categories instead)
- • Generic tech terms: RSS, API, SEO, URL, HTML, CSS, JSON, XML, iOS, Android
- • Too short/ambiguous: UK, US, EU, AI
Filtering Rules for Names
Names detected by NER are filtered using these rules to reduce false positives:
- • Names shorter than 3 characters are excluded
- • All-lowercase names are excluded (usually not real names)
- • Names containing numbers are excluded
- • Names ending with company suffixes like "cast", "pod", "fm", "audio", "media" are excluded
- • Names containing place indicators (e.g., "Ohio", "London") are excluded
- • Names with more than 4 words are excluded (likely not a real name)
- • Standalone first names under 10 characters are excluded (too ambiguous)
Name Consolidation
Some organizations have multiple name variations that are consolidated for accurate counting:
- • "iab", "interactive advertising bureau" → IAB
- • "npr", "national public radio" → NPR
- • "iheart", "iheartmedia", "iheartradio" → iHeart
- • "siriusxm", "sirius xm" → SiriusXM
- • "pod news", "pod news weekly review" → Podnews Weekly
Search Trends vs. Pre-built Charts
The Search Trends feature (on the main dashboard) and Transcript Search query the raw transcript database directly. They do NOT apply any of the exclusions above. This means you can search for "James Cridland" or "Sam Sethi" and see their mentions over time, even though they're excluded from the pre-built "Most Mentioned People" chart.
Technical Details
Technology Stack
Files
- index.html - Main dashboard
- trends.html - Weekly entity trends timeline
- citations.html - Citation detail pages
- about.html - This documentation
- podnews_2025_report.json - Summary data (fast load)
- podnews_2025_report_search.json - Search/citations data (lazy load)
- analyze_transcripts.py - Analysis script
- download_transcripts.py - Transcript downloader
Data Size & Performance
Data is split into two JSON files for optimal loading performance:
- • Summary file (~0.2MB): Contains aggregated counts, trends, and metadata. Loads immediately on page load.
- • Search file (~23MB): Contains full citations and transcripts. Lazy-loaded only when you use transcript search.
This split architecture ensures the dashboard loads instantly while still providing full search capabilities when needed.