- Published
- Oct 30, 2026
- Reading time
- 11 min read
Podcast Ad Delivery: Reconcile Downloads and Confirmed Plays
Reconcile podcast downloads, ad-delivery records and confirmed plays using current IAB guidance, matched evidence and a practical discrepancy worksheet.

In this guide (10 sections)
When a podcast host and sponsor report different delivery, first establish whether they counted the same thing. A qualified episode download, delivery of the advertising portion and a client-confirmed ad play are separate measurements. Match the episode, advertisement, reporting window and method before comparing counts; then connect the agreed quantity to the applicable commercial terms.
A public question about podcast sponsorship discrepancies asks which source governs a disagreement, whether it affects payment and what evidence resolves it. The poster's follow-up asks whether the parties actually bought downloads, ad delivery or listens. That is the right distinction to preserve. A reply's reported percentage gap and payment delay are anecdotes, not a benchmark or a promise about another campaign.
The useful result is an explainable difference and a precise unresolved question. Choosing the larger dashboard, averaging the totals or calling every missing play invalid traffic will not produce that result.
Use the current podcast framework for the relevant delivery path#
IAB Tech Lab's current podcast measurement page lists version 2.3, updated 29 September 2026. Its finalization notice confirms that the guidelines are final and hosted on GitHub. An older development mirror or public-comment PDF is not the current version simply because a search engine ranks it first.
The v2.3 release explanation covers audio and video podcasts distributed through open RSS or podcast applications using server-side file delivery. It explains why server-based ad delivery and client-confirmed exposure remain different. Check your actual distribution path before applying those definitions to another streaming product or a closed platform's view counter.
Record the provider's declared methodology and version. A published guideline does not establish that a particular service implements it, that its whole product is certified, or that the campaign's commercial agreement adopts it. Verify those questions separately.
Agree on what counts as podcast delivery
Downloads, ad delivery and confirmed plays are separate measurements. Discuss which delivery definition belongs in your order reconciliation.
Discuss a measurement questionA short inquiry to scope relevant options; no campaign access is needed.
Put the three measurements beside each other#
The current technical guidelines distinguish a filtered qualifying episode download, an ad delivered on the basis of validated delivery records, and a client-confirmed ad play supported by the player's signal. Ad delivery requires the relevant ad content to have been delivered, not merely an ad selected. Client confirmation requires client-side evidence, which may be unavailable.
Use these practical questions to interpret the report you receive:
| Reported object | Ask the provider | Avoid this inference |
|---|---|---|
| Episode download | Which episode requests qualified under this method? | Every counted download became a complete listen |
| Ad delivered | Which advertisement and delivered portion support this record? | Every episode download contained this campaign's ad |
| Confirmed ad play | What player event and reporting coverage support confirmation? | Missing confirmation proves the ad was never played |
| Audience estimate | Which identity method and time frame does it use? | A delivery count is a count of distinct people |
Keep the original labels alongside any normalized names in your worksheet. A field labeled impressions may need an explanation before you can place it in one of these rows. Do not silently redefine it to match the other party's vocabulary.
Start with one order line and one episode cohort#
Choose a manageable comparison. Identify the campaign and placement, the relevant show and episode identities, the advertisement version and the agreed service period. Preserve the original exports with their extraction times and timezones.
A recommended comparison contract contains:
- The delivery path and the source responsible for each report.
- Episode, campaign and ad identifiers, with an explicit mapping between systems.
- Dates, timezone, measurement window and report extraction time.
- The exact metric and its methodology version.
- Included geography, distribution sources and other filters.
- Counting exclusions and any duplicate-handling treatment.
- The evidence available for the advertising portion and client playback.
- The order's stated counting basis and the owner of unresolved terms.
Names can assist review, but do not join two episodes solely because their titles look alike. Preserve reissued files, replacement creatives and mapping changes with effective dates. If one reporting row covers several episodes, aggregate comparable delivery to that scope before joining it; do not copy the whole row into each episode.
For the commercial calculation after the count is understood, use the invoice reconciliation workflow. It covers order versions, rates, accepted corrections and applied credits. This guide addresses the podcast evidence needed before that calculation.
Trace the evidence boundary that explains the gap#
Ask each provider for a sanitized example showing how its reported record relates to the episode and advertising segment. Keep the request focused: what was observed, what rule made it count, and which later event remains unobserved?
Do not ask a partner to expose an entire audience dataset to prove one aggregate. An authorized sample, methodology explanation and scoped reconciliation can be more useful than a large unstructured log dump. Keep personal information and sensitive identifiers inside the appropriate access-controlled process.
The technical guidelines require a validated download before an ad can qualify as delivered and distinguish episode-file delivery from the bytes covering the ad. Ask whether the discrepancy occurs before or after that validation. A full-ad delivery record is stronger evidence of that boundary than a generic episode request, but it still does not establish attention.
Use a short finding table:
| Observed mismatch | Next evidence request |
|---|---|
| Raw request count differs from qualified downloads | The method's inclusion and exclusion summary for the same scope |
| Qualified downloads differ from campaign ad delivery | Episode/ad mapping and evidence for the relevant ad portion |
| Ad delivery differs from confirmed play | Player-confirmation availability, eligible population and event receipts |
| Provider counts agree but invoice differs | Applicable rate, pricing unit, amendments and billed scope |
| Aggregate difference cannot be reproduced | A fixed export, timestamped scope and a smaller comparable sample |
These are investigation branches, not diagnoses. An ad near the end of an episode may motivate a coverage check, but its position alone does not explain a particular discrepancy. Preserve the unknown until the matched records support a finding.
Keep methodology changes separate from missing delivery#
When two sources disagree, ask whether they observe the same part of the delivery path. Then request each source's measurement-window policy and the effective version of its filtering rules. Those are comparison inputs, not controls to adjust until the figures match.
The v2.3 framework discusses prefix measurement, enclosure URLs and measurement windows. For an affected episode, ask whether a feed, file or measurement-path change occurred during the reporting period. Retain the old and new references privately and have the operator explain how the transition appears in each report.
A feed migration, a replaced episode file or a changed reporting configuration can therefore be an investigation lead. None is established as the cause merely because it happened near the count change. Compare a stable interval on each side and document the boundary you can actually verify.
If a provider revises a historical total, retain both snapshots. Record whether it corrected a mapping, changed a methodology, processed additional evidence or supplied no explanation. Replacing the earlier file without a revision note makes the disagreement harder to reproduce.
A platform playback dashboard answers a narrower question#
Apple Podcasts Analytics describes aggregated listening and viewing activity on Apple Podcasts, while identifying hosting/server data as a source of other metrics such as downloads. It also applies a visibility threshold: at least five unique listeners must play content within the selected period before show data appears.
A blank Apple dashboard is therefore not automatically a zero-download finding. Nor is an Apple-only play total an all-distributor podcast-delivery total. Preserve the platform, date range, available metric and any threshold message with the observation.
Do not assume a play count is the same as an ad-specific confirmation. Check whether the available event concerns the whole episode, an advertising segment or a specific milestone. If the campaign requires ad-level evidence and the available platform report is episode-level, record the mismatch.
Avoid combining host downloads and platform plays into a larger audience total. They may describe overlapping activity through different observation points. The same person can also generate several events. The metric comparison guide provides the broader discipline for maintaining these distinctions.
Work a hypothetical reconciliation without inventing missing listens#
Hypothetical example only. A reviewer selects one episode cohort with one relevant campaign ad opportunity per qualified download. All counts use the same period, mapping and deduplication basis. These are invented teaching figures, not customer measurements or a prescribed billing formula.
| Evidence stage | Count | Meaning within this example |
|---|---|---|
| Qualified episode downloads | 10,000 | Episode-delivery records passed the declared method |
| Ad delivery established | 9,200 | Records establish delivery of the relevant ad content |
| Ad delivery not established | 800 | The matched record does not establish that boundary |
| Ad deliveries eligible for client confirmation | 3,000 | The tested confirmation path covers this subset |
| Ad deliveries outside that confirmation coverage | 6,200 | Client confirmation cannot resolve these records |
| Client-confirmed plays within the eligible subset | 1,800 | The required confirmation was received |
| Eligible records without confirmation | 1,200 | The confirmation remains absent or unresolved |
The 800-count difference is 8% of the 10,000 qualified downloads. Investigate that episode-to-ad boundary. Do not label the 800 fraudulent or unpaid solely from this worksheet.
Client-confirmation coverage is 3,000 / 9,200, approximately 32.61% of established ad deliveries. Confirmation within that eligible subset is 1,800 / 3,000, or 60%. Both percentages need their denominators beside them.
The 1,200 unconfirmed eligible records and 6,200 records outside coverage answer different questions. Subtracting 1,800 from 9,200 and calling the remaining 7,400 unplayed ads would conceal that difference. It would also turn missing evidence into a playback conclusion.
Do not sum every row: several rows are subsets of earlier ones. The client-confirmation classification partitions the 9,200 established deliveries into 1,800 confirmed, 1,200 eligible but unconfirmed and 6,200 outside coverage. It does not partition unique people or determine an invoice adjustment.
Run an acceptance check that preserves the real measurement path#
Use an authorized test plan with known episode and ad versions. State which observation you expect from each system and who can retrieve it. Keep test activity identifiable in your own record and respect the provider's supported testing arrangements.
Observe the sequence rather than manufacturing a beacon or repeatedly downloading files to make a target count rise. Save the relevant serving record, later provider report and available client event. A test confirms the tested path; it does not establish historical playback across apps that never supplied confirmation.
If a mapping is corrected, repeat the same comparison with the corrected mapping and retain the original result. If the problem is unsupported client evidence, no mapping change can create the missing historical player signals. The appropriate outcome may be a disclosed limitation and a revised evidence requirement for a future campaign.
Keep detection and financial resolution separate. A corrected report can support discussion of the applicable terms, but a reported discrepancy does not itself establish a refund, a makegood or a payment obligation.
Finish with a usable evidence packet#
The next review should contain the agreed counting basis, the exact episode/ad mapping, original export versions, methodology references, comparable totals and the first unresolved boundary. Include a compact exception list with an owner and next action for each item.
Useful conclusions include a verified scope mismatch, a corrected mapping, an explained methodological difference or an unresolved delivery/confirmation gap. An honest unresolved finding is more useful than a guessed listen count.
For supported campaigns already sending relevant events to AdProof, retain a campaign/date/timezone-scoped export alongside separately obtained hosting, ad-serving and player evidence. Confirm the actual event coverage before comparing it with podcast-specific metrics. The anonymized streaming-publisher case illustrates delivery reconciliation, not a podcast-host connector or IAB certification. Discuss your available evidence to establish which received campaign records can support the review.
Frequently asked questions#
Does a podcast download prove the sponsor's ad was heard?
No. Establish the relevant ad-delivery record and any available client confirmation separately. A qualifying episode download does not by itself establish complete ad playback or attention.
Which report should determine what the sponsor pays?
Use the applicable agreement's defined quantity and counting source, and have the responsible commercial owner resolve ambiguity. Technical guidelines explain metrics; they do not establish the terms adopted by every buyer and seller.
Does missing client confirmation mean the ad was not played?
No. First establish whether that delivery path could supply confirmation. Keep eligible-but-unconfirmed records separate from records outside coverage, and investigate each under its own evidence.
Does IAB compliance make two providers' totals identical?
Do not assume exact agreement. Compare the same scope and disclosed methods, verify the specific service's compliance claims and investigate the actual difference. Do not use certification as a substitute for the campaign's evidence or counting agreement.
Agree on what counts as podcast delivery
Downloads, ad delivery and confirmed plays are separate measurements. Discuss which delivery definition belongs in your order reconciliation.
Discuss a measurement questionA short inquiry to scope relevant options; no campaign access is needed.
