We agree on the goal, the methods and the reading window with the brand before a campaign starts, write them down, and report against that plan. A podcast test read without a baseline, or judged a week after the first episode, will usually understate what the channel did, and a test judged on one method will undercount the buyers that method cannot see.
The methods we use together
Each show gets its own code and vanity URL, and alongside them we run a post-purchase survey and, where the shows can support it, pixel attribution, because no single method sees the whole picture. Codes and URLs give a direct count and let us compare shows, though by Right Side Up’s estimate, in a guide updated in February 2024, they catch only some 20% to 30% of the sales podcast advertising brings in. A survey asking new customers where they heard about the brand counts the buyers who searched for it and never used a code. Pixel attribution matches visits to the brand’s site with households that downloaded the episode, though it is weaker for listening that happens away from home.
Read side by side, the methods cover one another’s gaps, and where they disagree the difference usually says something about how the brand’s customers listen and buy. Our article on measuring podcast advertising sets out what each one can and cannot show, with the sources.
The baseline and the window
Before launch we run the survey for at least 30 days and record the brand’s usual levels of new customers, direct traffic and searches for its name, so the campaign can be read against what would have happened anyway.
After launch we give the results time to arrive. The same Right Side Up guide (updated February 2024) reckons an episode goes on picking up listens for something like 19 to 21 days once released, and in Magellan AI’s benchmark report for the second quarter of 2026 (September 2026), response was still climbing at day 30. We check halfway through a flight to catch anything clearly wrong, and read the full results once the last episode has had at least a month to be heard.
What the report says
The report gives one view per show: what it cost, what each method credits it with, and our estimate of the cost per new customer, with every assumption written down. We say where the methods disagree and why we weighted them as we did, so a brand’s finance team can follow the reasoning and change an assumption if they see it differently. We also note anything that moved outside direct tracking, such as branded search or retail sales around a flight, and keep it separate from the cost per customer unless the brand has agreed on how to count it.
Each show ends with a recommendation to renew it, change something and test again, or drop it.
What you receive
- A measurement plan agreed before launch, covering the goal, the baseline, the methods and the reading window.
- A check halfway through each flight for anything clearly wrong.
- A full report after the reading window for each show, with cost, what each method credits it with, and an estimated cost per new customer with the assumptions written down.
- A recommendation for each show to renew, adjust or drop.