Topic: Transcription

Maya and Tom introduce the season finale of How to Get Discovered by presenting a thought experiment involving two identical podcasts. Show A invests in discoverability through proper transcripts, custom domains, and optimized show notes, while Show B continues with audio-only distribution. The experiment tracks the divergent paths of these shows over a twelve-month period to illustrate the long-term impact of technical SEO.

Tom reflects on his past dismissiveness toward podcast SEO, admitting his eye-rolling was a defensive mechanism to avoid additional work. He clarifies his stance on listener loyalty, noting that while it is vital, it is not opposed to acquisition since every loyal listener begins as a stranger. He also acknowledges that his initial rejection of transcripts was based on outdated, poor-quality versions of the technology.

A raw wall of text from a machine transcription service lacks the headings and paragraph breaks necessary for effective search indexing. By breaking a transcript into sections with specific headings and timestamps, each segment receives its own unique URL or "addressable moment." This allows a user searching for a specific topic, such as freelance rate negotiation, to land directly on the relevant audio segment rather than a generic homepage.

Technical infrastructure elements like transcripts, structured pages, and custom domains serve as surfaces that allow a podcast to present itself as an authority. Shows that become "canonical" on specific topics are often referenced by journalists, researchers, and authors because they are findable through search during the research process.

The hosts conclude the episode by previewing next week's technical deep dive into transcript indexing and domain authority. The upcoming segment will explore why transcript pages are treated differently by search engines compared to standard show notes.

The conversation explores the future of podcast discovery through multilingual transcripts and AI chatbots like ChatGPT. By providing translated transcripts, English-language podcasts can surface in foreign-language search queries and chatbot responses globally. The host mentions PodHerd as a service that automates this process, predicting that translated content will become a standard industry practice within three years.

The back catalogue was processed through PodHerd to transcribe, structure, and index every episode into sectioned pages with timestamps. After three months, episodes that previously received zero search traffic began seeing consistent weekly listens. One forgotten interview about switching from agency to in-house work emerged as a major driver of new listeners, proving that existing audio holds untapped value if it is made findable.

Word-of-mouth recommendations in 2026 are described as requiring infrastructure beyond a simple MP3 feed to be effective. Listeners prefer sharing specific moments or exchanges via clips in group chats rather than assigning "homework" by sending full episode links with timestamps. Effective clip generation allows a listener to highlight a transcript passage and share a video clip with audio and attribution, making the recording portable.

AI chatbots currently rely on text-based data including articles, Reddit threads, and transcripts rather than directly ingesting audio files. For a podcast to exist within a chatbot's recommendation pool, it must have a presence on the open web through structured text. Unlike traditional search engines where a show might rank on a second page, being absent from a chatbot's concise answer results in total invisibility for the creator.

The quality of text provided to AI models is identified as a critical factor, as simple "walls of text" from raw machine transcripts are often ineffective. Modern transcription technology has improved significantly in areas like speaker separation, punctuation, and paragraph breaks, making the output more readable for both humans and AI. High-quality, structured text is presented as a necessity for moving beyond the "slab" of data that characterized early automated transcription efforts.

Hosts Maya and Tom introduce the episode "Whose House Are You Building?" for the podcast How to Get Discovered. The discussion focuses on the unglamorous mechanics of podcast searchability and why the specific URL where transcripts are hosted is a critical factor for long-term discoverability.

Search engines like Google assign trust scores, formerly known as domain authority, to websites based on age, reputable links, and consistent content. When transcripts are hosted on a platform's subdomain, the SEO benefits are diluted across thousands of other shows. Hosting transcripts on a personal domain allows every share and link to accrue lasting authority and ranking power for the creator's own property.

Modern tools have simplified the process of building searchable transcript pages, which previously required significant development time. Services like PodHerd can now ingest an RSS feed to automatically transcribe and structure episodes into searchable web pages. While these pages default to the service's domain, they provide immediate benefits for listener accessibility and indexing.

Standard podcast transcripts are often published as unstructured "blobs" of text that search engines cannot effectively parse or understand. By using tools like PodHerd to index and structure episodes with specific URLs and timestamps, creators can surface specific moments for search traffic. Data from an eight-month trial showed that properly indexed old episodes began receiving steady traffic and new subscribers years after their original release.

The premiere episode concludes with a preview of next week's topic regarding where to publish podcast transcripts. The upcoming discussion will cover the trade-offs between domain authority and platform authority, specifically focusing on whether creators should host content on their own URLs. The hosts also tease a technical debate regarding the use of CNAME records for podcast SEO.