Attribution Methodology

Audiohook uses a straightforward methodology for attribution. By simply asking the question “did we play an ad directly to someone who completed a desired action” we aim to deliver highly accurate, dependable reporting on one of advertising’s most asked questions.

When playing an ad, audiohook has access to specific information identifying the device that the ad is being played on, of most importance here (but not limited to) is the IP address, the mobile advertising ID (MAID), and the user agent string. Using this information, we can match it against device information for events gathered from the pixel that is placed on the advertiser’s website. When there’s a match, we know that these two events are related, which is the first step.

Next, we have to check that the event did, in fact, take place after the impression ran. If somebody placed an order on Tuesday, but heard the ad on Wednesday, that order clearly wasn’t influenced by the advertisement and no credit is earned.

Audiohook supports an attribution window of up to 30 days of the impression being served (i.e. a lookback window of 30 days), and with access to the underlying information, it’s possible for users to utilize a custom lookback window of their own.

Additionally, because pixels fire repeatedly and could send the same data, Audiohook implements deduping of data to ensure we are not counting the same order multiple times. This is accounted for by only letting each impression match a single time, per event type. This means that one impression could, for example, account for 1 attributed pageview and 1 attributed order, but it will not count for 2 pageviews or conversions.

To verify Audiohook attribution data, customers receive granular information about an event and user. Included in our reports are the unique event ID provided by the advertiser, a raw $ amount, the time stamp (date & time of the event) and the user’s IP address in hashed MD5 file format. This data can be compared with the advertiser’s customer sales data from their CRM, Shopify, or other similarly used 1st party sales platforms to verify the legitimacy of the conversions.

These principals (IP/MAID matches, event happened no more than 30 days but after the impression, and only one attributed event per type per impression) are how Audiohook calculates attributions.

Transparency is important and simply receiving ‘true’ or ‘false’ does not help to see anything beyond surface level. By utilizing the above method, not only can you see the underlying data behind the conversion, but also allows for a greater level of verification, making sure that when an attribution is claimed, it can be backed up with data. Additionally, this empowers users to utilize the data that Audiohook delivers to make informed decisions about supply and targeting that has (positively or negatively) impacted campaign performance and make adjustments as necessary.

The notable limitation to this method is that multi-touch ads do not factor into our equation. It is entirely possible that someone heard/saw multiple ads from different sources that all contributed to their taking the desired action. This does not deter our attribution process from being a verifiable metric that excels at being a measurement of performance, especially when used relatively.