Podcast Attribution Rebuilt After Chartable Shut Down

Chartable's attribution products were wound down by Spotify in 2024, so a measurement plan that still names them is a plan nobody has opened in two years. What we install instead is Podscribe as the primary household graph, Spotify Ad Analytics for Spotify-served inventory, and a first-party conversion feed you own. Expect about 21 days between the pixel going live and a reading worth acting on, because household matching needs a pre-period baseline before it can say anything at all. The stack, in the order we deploy it: Server-side GTM container on a subdomain of your own domain, so the conversion side of the match is not the first thing an ad blocker removes. Podscribe pixel on cart, checkout and one mid-funnel event that is not a pageview. RSS prefix installed in the correct order, because prefix order decides which system counts first, and we have inherited two accounts where a stacked prefix quietly halved the numbers for a quarter before anyone noticed. Then the publisher handshake: every network serving your spots must fire the pixel from their own ad server. Megaphone, Art19, Acast and Podigee all support it. A network that will not is a network we stop buying. Promo codes and vanity URLs stay in the plan and stay in their place. They capture a floor, usually a small fraction of what actually converted, and they are a sanity check on the graph rather than the measurement itself. Consent, under German law specifically: Section 25 TDDDG, the law that replaced the TTDSG in May 2024, governs anything written to or read from a device, so the web pixel is gated through your CMP with TCF 2.2 and Google consent mode v2. The podcast side is a different question. IP-based household matching is not device storage, but an IP address is personal data, and your data protection officer has to sign a legal basis for it. We write that assessment with them before anything goes live. If they will not sign, say so on the first call: the honest alternative is matched-market geo testing, which we will build instead, and which needs a larger budget to reach significance. What we will not report: Attributed conversions that are branded search wearing a costume. Last-touch podcast credit. Publisher-supplied attribution of the publisher's own inventory. Downloads as a performance number. And no 30-day post-impression windows: audio decays quickly, so we read seven days below 150 euro average order value and fourteen above, written into the specification before the first campaign rather than chosen afterwards to flatter a result. What you get: A measurement specification your analytics team can argue with, line by line. Pixel, prefix and publisher handshakes live and verified with a test spot. A weekly reconciliation across Podscribe, Spotify Ad Analytics and your own GA4 and Shopify or Stripe data, with the discrepancy explained rather than averaged away. A holdout design that keeps fifteen percent of impressions dark. And a cut threshold agreed in writing: the number at which a show leaves the plan, decided before anyone is attached to it. Not for you if: You run fewer than about ten shows, or under 30,000 euro a month in audio. Below that the household match produces noise that people over-read with confidence. Also not for you if the holdout is politically impossible, or if the reason for measuring is to defend a plan already signed off. We have ended two engagements at exactly that point, and both times it was obvious in week one.