> [!question] Lingering Questions > - Do AI assistants actually weight review and forum content the way described, and does it vary by model and category? Asserted without evidence, and it's the episode's most consequential claim. > - If amplifying customer voice becomes the route into AI shortlists, does it commoditize exactly as SEO did? The episode treats it as durable without addressing saturation. > [!tip] Article Ideas > - "Ask three assistants who they'd recommend" — a testable exercise piece for WordPress agencies, lifted directly from the episode's own homework assignment. > [!link] Cross Reference > - [[3. Resources/Permanent Notes/Funnel Position Predicts Which Content AI Search Extracts]] — direct tension worth holding in one note rather than two competing claims: that note is about frozen-model knowledge entrenching incumbents; this episode argues retrieval from customer voice *disadvantages* incumbents. Both concern what determines visibility once models mediate discovery, pointing opposite ways. > - [[3. Resources/Permanent Notes/Cheap Generation Raises the Total Maintenance Bill]] — "everybody's working harder than they did before cuz they're trying to manage the machine now" converges with this note's maintenance-burden argument, from a marketing vantage rather than an engineering one. > [!quote] Vocabulary > - **marketing hourglass** — Jantsch's own funnel framework: know, like, trust, try, buy, repeat, refer > - **voice of customer** — what customers say publicly (reviews, social, forums), argued here to be the actual input AI recommendation engines draw from > - **referability** — whether a value proposition is simple enough for a third party to explain on your behalf ## Metadata - Source: [[3. Resources/Literature Notes/The Marketing Channel AI Can't Commoditize]] - Category: Article/Podcast (Duct Tape Marketing solo episode) - Published: 2026-07-31 ## Author's thesis AI made competent marketing output cheap and undifferentiated, pushing effort down the funnel toward retention and referral — but the genuinely novel claim is that AI recommendation engines appear to compile shortlists from customer voice (reviews, social, Reddit) rather than site authority, converting testimonials from a conversion asset into a discoverability one. ## Key arguments 1. **Shortlist exclusion is a silent failure** — unlike a lost pitch, if an AI assistant doesn't name you, the prospect never arrives to be lost, so the channel degrades without anyone noticing. 2. **AI recommendation engines appear to rank by customer-voice corpus, not site authority** — the episode's most consequential and least-supported claim; if true, incumbency stops conferring automatic advantage. 3. **A referral requires the referrer to be able to explain you** — goodwill can't compensate for an unexplainable value proposition; most positioning work optimizes for the buyer and forgets the intermediary who has to carry the message. ## Claim atoms ### AI recommendation engines read customer voice, not site authority > [!quote] ID:rw_01kywpb11nngv6ez9pqmzsk9hk > They're actually diving deep into voice of customer. What customers are saying in reviews, what they're saying on social media sites, what they're saying in Reddit. *Where:* ~00:11:30. *Why:* the episode's most consequential claim — if correct, testimonials become a discoverability input rather than a persuasion asset, and the ranking basis shifts from authority to corpus. Unsupported; directly testable. ### Shortlist exclusion produces no observable signal > [!quote] ID:rw_01kywpb11nngv6ez9pqmzsk9hk > if AI doesn't include you in the short list, you know, somebody may never come to your website or check you out. *Where:* ~00:02:00. *Why:* unlike a lost pitch, exclusion generates no signal, so the channel can degrade silently and indefinitely. ### A recommendation requires the recommender to articulate the value > [!quote] ID:rw_01kywpb11nngv6ez9pqmzsk9hk > Your value proposition or what you are going to do when you are referred has to be easy for them to understand and therefore easy for them to explain. *Where:* ~00:07:30. *Why:* names the intermediary that positioning work usually forgets — an unexplainable value proposition blocks a referral regardless of goodwill. ## Permanent candidates - **AI recommendation engines rank by customer-voice corpus, not by site authority** — the strongest, most testable, and most consequential claim; direct tension with existing vault material on frozen-model incumbency worth holding in one note. ## Processing checklist - [ ] Run the three-assistant recommendation test for a WordPress agency category - [ ] Look for published research on how LLM assistants actually source recommendation content - [x] Connect into the relevant MOC — [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] ## Links - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — AI-mediated discovery reshaping marketing practice is core territory here. - [[3. Resources/Permanent Notes/Funnel Position Predicts Which Content AI Search Extracts]] — the opposing-direction claim this episode should be read against. - [[3. Resources/Permanent Notes/AI Recommendation Engines Rank by Customer-Voice Corpus Not Site Authority]] — the Permanent note this Literature note generated.