How a Meyppi audit works.
What you send, what we do with it, and what comes back.
- 01
You send us one campaign
A completed campaign, not one mid-flight: the media has to be published before there is anything to verify. You send the master creative files, one per contracted version, and the placement plan from the insertion order: publishers, shows, episodes, flight dates.
A typical starting package is three to five creatives across up to twenty publishers and forty episodes. Historical campaigns are fine; lookback is limited mainly by what remains publicly available in the feeds.
- 02
We collect the published media
We pull the published episodes from their RSS feeds and other sources we are authorised to use. If you hold recordings already, for example from a monitoring tool or a previous campaign, those go into the same pipeline.
Before anything is processed, each file is hashed and logged with its source (see the <a href="/security">security and data handling</a> page). From this point on, every claim in the final report can be traced back to a specific file with a specific hash from a specific source. That provenance is what makes the evidence worth having.
- 03
We fingerprint and match
Every episode and every creative version is fingerprinted: a constellation of spectral peak pairs that describes what the audio sounds like, computed identically every run, robust to the compression and loudness processing podcasts go through.
Matching the creative against the episode produces candidate alignments, and the alignment that holds up across the whole runtime is the match. A true match shows thousands of agreeing hash pairs at one consistent time offset. Interference, cross-talk and music beds do not produce that signature.
- 04
We classify every placement
Each expected placement gets one of six verdicts: Detected — Full, Detected — Partial, Version Mismatch, Not Detected, Insufficient Signal, or Manual Review Required (definitions on the sample report and methodology pages). The distinctions matter. A 24-second cut of a 30-second buy is not full delivery. Version 2 airing where version 1 was contracted is a version mismatch.
When several versions of a creative exist, the classifier ranks the evidence between siblings before attributing a version, so a version mismatch is reported only when the evidence supports it. Anything the pipeline cannot resolve with confidence is escalated to a person, and the report says that is what happened.
- 05
You receive the report
The deliverable is a report, not a login. A PDF you can forward, with an executive summary and a placement-by-placement table. CSV and JSON exports of the same data, so your finance or analytics team can work with it. And the raw verification events behind every row, so any single finding can be independently re-examined.
You will know where every rupee of that campaign actually landed, with the receipts attached.
What you need before starting
An audit can only be as good as the contract it checks against. Everything on this list is something you already have if the campaign ran.
- Master creative files for every contracted version (WAV, MP3 or M4A, the exact cut that was supposed to air)
- Insertion order or placement plan: publishers, shows, episodes, flight dates
- Campaign dates, and confirmation the campaign has finished airing
- If you have them: existing recordings or monitoring exports (optional)
- A contact who can answer questions about the contract when they come up
How long it takes
From receiving the creatives and media access, a typical single-campaign audit completes in five to ten working days. Processing itself is fast, over 1,000 audio-minutes per CPU-hour; the calendar time goes into source collection and into the manual review of flagged placements, which we do not rush.
Historical lookback usually adds days rather than weeks, bounded by what the feeds still serve. We tell you before starting if expected sources cannot be reached, because "we could not obtain the media" is a scope decision you should make, not a surprise in the report.
Audits cover pre-produced audio creatives in podcast and digital audio content, and pre-produced spots in video podcasts (matched on the audio track), where the published media is independently observable. FM radio campaign audits are running as a pilot in select cities: contact us. Dynamic ad insertion and host-read verification remain out of scope. If your campaign falls outside, we will say so before you pay anything.