Measuring what your podcast spend actually did

Why podcast attribution is hard

A podcast is downloaded, often over WiFi, often in advance, and listened to somewhere else entirely: a car, a run, a kitchen. There is no click. The listener hears about your product at 7am and searches for it from a work laptop at noon, or from a phone three weeks later. Any measurement system for this channel is reconstructing that broken chain after the fact.

This is why the channel's best practitioners are obsessive about one principle: every method undercounts, so never rely on one.

The spectrum of results

Before choosing instruments, be honest about what kind of result you're buying, because podcast advertising produces a spectrum and campaigns land all along it. At one end sits pure performance: a code redeemed at checkout, a pixel-matched conversion, growth math where every dollar reports back with a cost per acquisition. At the other end sits brand association, the slow transfer of a host's credibility to your name. Mailchimp was a scrappy email tool until its ad ran on Serial in 2014 and a mispronounced "MailKimp" became a national catchphrase. AG1 traveled from insider supplement to mainstream shorthand for a morning routine across years of Joe Rogan reads. Squarespace turned a decade of host-read offer codes into being the culture's default answer to "how do I make a website." None of those outcomes fit in a CPA report, and all of them were worth more than the media cost.

Most campaigns live between the poles, and the measurement stack is how you find out where yours lands. Each instrument weights a different stretch of the funnel: a pixel catches the visit that happens hours after the read, promo codes and vanity URLs count the bottom of the funnel where intent becomes an order, post-checkout surveys catch the buyers the click path missed and prove the ad worked even when attribution lost the trail, and brand lift studies measure the top, where awareness and association move before purchase ever does. Read together, they tell you where on the spectrum your reads are hitting, which is what decides whether you optimize the offer, the show list, or the message itself.

The five methods

1. Promo codes

A unique code per show ("use code SHOWNAME for 20% off"). Redemptions map cleanly to shows, setup is trivial, and hosts deliver codes naturally. The weakness is leakage: a large share of podcast-driven buyers never enter the code, and coupon sites can pollute the signal. Codes give you a reliable floor and per-show comparison, not a total.

2. Vanity URLs

A memorable per-show URL (yourbrand.com/showname) that redirects with tracking parameters. Same per-show logic as codes, same leakage problem, slightly worse recall. Useful as a second signal and as a natural call to action in the read.

3. Post-checkout surveys

One question at checkout or signup: "How did you hear about us?" This is the workhorse of podcast measurement. It catches the buyers codes miss, it's cheap, and at even modest volume it produces a stable percentage you can multiply against total new customers. Add "Which podcast?" as a follow-up and you get per-show signal too. Run it permanently.

4. Pixel-based attribution

Providers like Podscribe and Spotify Ad Analytics (which absorbed Podsights) log the IP and device fingerprint of each ad download, then match them against visitors and converters on your site. This captures the silent majority who never use a code, supports view-through windows, and scales across large campaigns. It requires dynamic insertion, involves probabilistic matching (household-level, not person-level), and needs enough volume to be statistically useful. From roughly $25,000 per quarter upward, it belongs in the stack.

5. Brand lift and incrementality

Survey-based lift studies compare awareness and intent between exposed and unexposed audiences; incrementality tests hold out matched markets or time periods to isolate the channel's causal effect. These answer the question the other four can't: "did the ads create demand, or just collect credit for it?" They're the right tool for large brand budgets and for settling renewal arguments at scale. Specialists like Signal Hill Insights run these for podcasting specifically, and their published studies are worth reading before you commission one.

Building your stack

Reading the results honestly

• Give it time. Podcast response arrives on a lag; a third or more of conversions can land weeks after the read. Judging a show at day ten mostly measures impatience.
• Multiply, don't add. If the survey says podcasts drove 8% of new customers and codes captured half of that, your true number is closer to the survey. Use codes for per-show ranking and surveys for the total.
• Watch branded search. A podcast flight that works shows up in branded search volume and direct traffic within days. It's a crude signal, and a persuasive one.
• Compare shows on cost per attributed action, not response volume. The small show with 40 conversions on $2,000 beats the big one with 200 conversions on $20,000.

Verification: did the ads even run?

An unglamorous truth: ads get skipped, misread, cut short, and buried at the wrong timestamps. On large campaigns, a few percent of purchased reads simply never air as sold. Air-checking (confirming each read ran, at the right position, with the right offer and disclosure) is basic hygiene, and it pays for itself: documented delivery problems become makegoods, which become free media. Agencies automate this with a mix of transcription tools and human review; if you buy direct, spot-check every show's first read and sample the rest.