One of the most common questions we receive at AdParlor relates to the discrepancies between data our clients see on their Facebook or Twitter native accounts, what they see on AdParlor, and what they see on their 3rd party tracking software (e.g. via Atlas or Doubleclick).
AdParlor does not actually own any of the spend or tracking data; these belong to our publishers (i.e. Facebook and Twitter). We pull data in from each publisher into the AdParlor platform to empower our users with the advanced reporting and management tools not available on Facebook Ads Manager/Power Editor, or Twitter Ads Manager, and to allow easy management of multiple Facebook and Twitter accounts from a single login. This data synchronization occurs approximately every 15 minutes and is dependent on the rate at which Facebook/Twitter can send their updates.
The data discrepancies fall into three broad categories:
- Comparing apples with oranges (the most common)
- Synchronization differences between Facebook/Twitter and AdParlor
- Third party analytic software limitations
1. Comparing Apples with Oranges
The most common source of data discrepancies is when users compare two data sets, sampled from identical sources, but the samples have been selected using different criteria. This most commonly affects conversion-related metrics.
The three common selection parameters are as defined below.
Attribution Window Settings
The view and click attribution windows are simply reporting parameters used to determine when a conversion event can be credited to a particular Ad. For example, if I click on an Ad on Monday for a product but don't convert (e.g. buy the product) until 10 days later, whether or not the Ad gets credited with my conversion depends on the attribution windows selected. If I use a 7-day click attribution window, then my conversion falls outside the allotted timeframe and therefore does not count towards that Ad. If I use a 28-day click attribution window instead, then my conversion event falls within the click window and so my conversion gets counted for that Ad.
Hence, even though the data is identical (i.e. I was shown the Ad on day 1, I clicked on the Ad on day 1, and I converted on day 10), my reported numbers will be different between the two reports (generated using the two attribution window settings above).
Pro tip #1: The first thing to check is if the attribution windows are the same in both AdParlor and Facebook or Twitter when comparing your data.
Facebook allows users from the following time frames to determine view and click attribution criteria for conversions (default value is in bold):
- View: 0, 1, 7, 28 days
- Click: 1, 7, 28 days
Note: If you are measuring in-app (i.e. post-install) events using an MMP, note that MMPs can only report click-based conversions (i.e. not view-based conversions) so this could further contribute to discrepancies that you see.
Twitter actually sets attribution windows on a conversion pixel by pixel basis (this is set when you create a new pixel - but can be edited). For mobile in-app events, attribution windows are set through the MMP or MoPub.
Twitter differs from Facebook in that they only store one set of conversion data - for that specific attribution window, whereas Facebook stores all sets of conversion data - for all attribution windows - and allows the user to specify which set they want to view.
Impression vs Conversion Time (Conversion Attribution Settings)
Another potential source of discrepancies is the method used to choose which day the conversion event gets recorded on; by impression time or by conversion time:
- Impression Time: The conversion will be attributed to the day the Ad was served (impression) e.g. I saw the Ad 3 days ago but I installed the app today - the conversion gets attributed to stats for 3 days prior
- Conversion Time: The conversion will be attributed to the day the conversion actually happened e.g. I saw the Ad 3 days ago but I installed the app today - the conversion gets attributed to stats for today
By default, AdParlor reports conversions by "conversion time", whereas on Facebook the default is to report by "impression time". The impact of this is that when comparing day to day performance, the numbers will not match - but when looking at lifetime performance of the campaign, the numbers will match.
Partial vs. Complete Import
If you are comparing results across multiple Campaigns, Ad Sets or Ads, make sure that you are comparing the same entities, as well as the same number. The AdParlor system allows you to selectively import your Campaigns from your Facebook or Twitter accounts, so ensure that you are comparing the same entities across both reporting tools.
2. Synchronization differences between Facebook/Twitter and AdParlor
AdParlor synchronizes stats from Facebook and Twitter approximately every 15 minutes (the true cadence depends on the rate at which the publisher platforms can send updates for their data). If you are monitoring ongoing campaign performance in real-time, expect some minor discrepancies. If the campaign has completed, then your spend-related data should be identical between Facebook/Twitter and AdParlor.
3. Third party analytic software limitations
Third parties who track conversions, views or clicks of Ads served on domains not owned by them are susceptible to many sources of discrepancy. These tools typically depend on cookie-based tracking techniques as they do not have direct access to the same data that Facebook or Twitter do (what Ads were shown, when, and who saw them).
Tracking of conversion, view or click related data of Ads with cookie-based software is affected by:
- Users deleting cookies (thus platform is unable to track conversions)
- Users completing a conversion action in a different browser to the one they saw the Ad in (thus platform is unable to track conversions as cookies are browser-specific)
- Cookie disabling via Mozilla Firefox's Do Not Track functionality or Safari's default disabling of 3rd party cookies
- Browser extensions such as Ghostery which enable users to disable tracking
- Online 'spammers' or 'bots' that contribute to false clicks
- Third party software themselves may automatically remove duplicates of clicks before displaying your data
URL-based third party trackers are susceptible to incorrect interpretation of URL parameters or setup; these may also potentially be intercepted before reaching the analytical software (e.g. Google Analytics).
Still having problems? Contact your AdParlor representative with the following information:
- Publisher (Facebook/Twitter)
- Campaign ID(s), Ad Set ID(s) affected
- Attribution Window settings in Facebook/Twitter AND in AdParlor
- Data seen on Facebook/Twitter native (screenshots are useful)
- Data seen in AdParlor (screenshots are useful, or use the Shareable URL feature)
- Any additional information that could be helpful to your AdParlor representative for troubleshooting