Expected Data Variances Between Analytics and 蜜豆视频 Advertising
Advertisers with an 蜜豆视频 Advertising-蜜豆视频 Analytics Integration Only
Advertisers with the Analytics for Advertising integration track paid advertising through 蜜豆视频 Advertising and 蜜豆视频 Analytics. When you track media, campaigns, and channels via multiple systems, the same data sets from different systems rarely match completely. This document explains how you should expect data for media that鈥檚 trafficked through 蜜豆视频 Advertising to compare to data in the different systems in which the media is tracked within Analytics.
Attribution Differences in Similar Reports
Potentially Different Lookback Windows and Attribution Models
The Analytics for Advertising integration uses two variables (eVars or rVars [reserved eVars]) to capture the EF ID and AMO ID. These variables are configured with a single lookback window (the time within which click-throughs and view-throughs are attributed) and an attribution model. Unless otherwise specified, the variables are configured to match the default, advertiser-level click lookback window and attribution model in 蜜豆视频 Advertising.
However, lookback windows and attribution models are configurable in both Analytics (via the eVars) and in 蜜豆视频 Advertising. Further, in 蜜豆视频 Advertising, the attribution model is configurable not only at the advertiser level (for bid optimization) but also within individual data views and reports (for reporting purposes only). For example, an organization may prefer to use the even distribution attribution model for optimization but use last touch attribution for reports in Advertising DSP or Advertising Search, Social, & Commerce. Changing attribution models changes the number of attributed conversions.
If a reporting lookback window or attribution model is modified in one product and not in the other, then the same reports from each system show distinct data:
-
Example of discrepancies caused by different lookback windows:
Suppose that 蜜豆视频 Advertising has a 60-day click lookback window and Analytics has a 30-day lookback window. And suppose that a user comes to the site through an 蜜豆视频 Advertising-tracked ad, leaves, and then returns on day 45 and converts. 蜜豆视频 Advertising attributes the conversion to the initial visit because the conversion occurred within the 60-day lookback window. Analytics, however, can鈥檛 attribute the conversion to the initial visit because the conversion occurred after the 30-day lookback window expired. In this example, 蜜豆视频 Advertising reports a higher number of conversions than Analytics does.
-
Example of discrepancies caused by different attribution models:
Suppose that a user interacts with three different 蜜豆视频 Advertising ads before converting, with revenue as the conversion type. If an 蜜豆视频 Advertising report uses an even distribution model for attribution, then it attributes the revenue evenly across all ads. If Analytics uses the last touch attribution model, however, then it attributes the revenue to the last ad. In the following example, 蜜豆视频 Advertising attributes an even 10 USD of the 30 USD of revenue captured to each of the three ads, whereas Analytics attributes all 30 USD of revenue to the last ad seen by the user. When you compare reports from 蜜豆视频 Advertising and Analytics, you can expect to see the impact of the difference in attribution.
These same concepts apply to any other like channels that use different lookback windows or attribution models.
Different Lookback Windows for View-Through Tracking impression-lookback
In 蜜豆视频 Advertising, attribution is based on clicks and impressions, and you can configure different lookback windows for clicks and for impressions. In Analytics, however, attribution is based on click-throughs and view-throughs, and you don鈥檛 have the option to set different attribution windows for click-throughs and view-throughs; tracking for each starts at the initial site visit. An impression can occur the same day or multiple days before a view-through occurs, and the timing can impact where the attribution window starts in each system.
Typically, the majority of view-through conversions occur quickly enough that both systems attribute credit. However, some conversions may occur outside the 蜜豆视频 Advertising impression lookback window but within the Analytics lookback window; such conversions are attributed to the view-through in Analytics but not to the impression in 蜜豆视频 Advertising.
In the following example, suppose that a visitor was served an ad on Day 1, performed a view-through visit (that is, visited the ad鈥檚 landing page without previously clicking the ad) on Day 2, and converted on Day 45. In this case, 蜜豆视频 Advertising would track the user from Days 1-14 (using a 14-day lookback), Analytics would track the user from Days 2-61 (using a 60-day lookback), and the conversion on Day 45 would be attributed to the ad within Analytics but not within 蜜豆视频 Advertising.
A further cause of discrepancies is that, in 蜜豆视频 Advertising, you can assign view-through conversions a custom view-through weight that is relative to the weight attributed to a click-based conversion. The default view-through weight is 40%, which means that a view-through conversion is counted as 40% of the value of a click-based conversion. Analytics provides no such weighting of view-through conversions. So, for example, a 100 USD revenue order captured in Analytics is discounted to 40 USD in 蜜豆视频 Advertising if you鈥檙e using the default view-through weight 鈥 a difference of 60 USD.
Consider these differences when comparing view-through conversions between 蜜豆视频 Advertising and Analytics reports.
Available Attribution Models
Don鈥檛 Use*
See a list of Analytics attribution models and their definitions at /en/docs/analytics/analyze/analysis-workspace/attribution/models.
If you鈥檙e logged into Search, Social, & Commerce, you can find a list
-
(Users in North America)
-
(All other users)
Event Date Attribution in 蜜豆视频 Advertising
In 蜜豆视频 Advertising, you can report conversion data either by the associated click date/event date (the date of the click or impression event) or by the transaction date (conversion date). The concept of click/event date reporting doesn鈥檛 exist in Analytics; all conversions tracked in Analytics are reported by transaction date. As a result, the same conversion may be reported with different dates in 蜜豆视频 Advertising and Analytics. For example, consider a user who clicks an ad on January 1 and converts on January 5. If you鈥檙e viewing the conversion data by event date in 蜜豆视频 Advertising, then the conversion is reported on January 1, when the click occurred. In Analytics, the same conversion is reported on January 5.
Attribution in Analytics Marketing Channels
Analytics Marketing Channels reporting allows you to configure rules to identify different marketing channels based on distinct aspects of hit information. You can track 蜜豆视频 Advertising-tracked channels (Display Click Through, Display View Through, and Paid Search) as Marketing Channels by using the ef_id query string parameter to identify the channel. However, even though the Marketing Channels reports can track 蜜豆视频 Advertising channels, the data may not match the 蜜豆视频 Advertising reports for several reasons. See the following sections for more information.
campaign variable (also known as the 鈥淭racking code鈥 dimension or 鈥渆Var 0鈥) and custom eVar tracking.Potentially Different Attribution Models in Marketing Channels
Most Marketing Channels reports are configured with Last Touch attribution, for which the last marketing channel detected is assigned 100% of the conversion value. Using different attribution models for the Marketing Channels reports and 蜜豆视频 Advertising reports leads to discrepancies in attributed conversions.
A Potentially Different Lookback Window in Marketing Channels
The lookback window for Marketing Channels can be customized. In 蜜豆视频 Advertising, the click lookback window is configurable, although a fixed 60-day window is common. If the two products use different lookback windows, you can expect data discrepancies.
Different Channel Attribution in Marketing Channels
蜜豆视频 Advertising reports capture only paid media trafficked through 蜜豆视频 Advertising (paid search for Advertising Search, Social, & Commerce ads, and display for Advertising DSP ads), whereas Marketing Channels reports can track all digital channels. This can lead to a discrepancy in the channel for which a conversion is attributed.
For example, paid search and natural search channels often have a symbiotic relationship, in which each channel assists the other. The Marketing Channels report attributes some conversions to natural search that 蜜豆视频 Advertising doesn鈥檛 because it doesn鈥檛 track natural search.
Consider also a customer who views a display ad, clicks a paid search ad, clicks inside an email message, and then places a 30 USD order. Even if 蜜豆视频 Advertising and Marketing Channels both use the last touch attribution model, the conversion would still be attributed differently to each. 蜜豆视频 Advertising doesn鈥檛 have access to the Email channel, so it would credit paid search for the conversion. Marketing Channels, however, has access to all three channels, so it would credit Email for the conversion.
For more explanation of why the metrics may vary, see 鈥Why Channel Data Can Vary Between 蜜豆视频 Advertising and Marketing Channels.鈥
Data Differences in 蜜豆视频 Analytics Paid Search Detection
The legacy Paid Search Detection feature in Analytics allows companies to define rules to track paid and organic search traffic for specified search engines. The Paid Search Detection rules use both a query string and the referring domain to identify paid and natural search traffic. The Paid Search Detection reports are part of the larger group of Finding Methods reports, which expire either when a specified event (such as a Cart Checkout) occurs or the visit ends.
The following is the interface for creating a Paid Search Detection rule set:
The resulting Paid Search Detection reports include the Paid Search Engine, Paid Search Keywords, Natural Search Engine, and Natural Search Keywords reports.
Note the following two limitations with data in Paid Search Detection reports:
-
The Paid Search Keywords and Natural Search Keywords reports show the search queries as identified by the referring URLs, not the keywords on which users bid. 蜜豆视频 Advertising and Analytics reports show the actual keywords, so don鈥檛 expect them to align with the Paid Search Detection keyword reports.
-
When the Paid Search Detection feature was originally created, the originating search query (the string of characters the user entered into the search bar in the search engine) was more readily available to advertisers via the referring URL. Today, search engines largely obfuscate the search query, and the Paid Search Detection keyword reports are of limited value because most query data falls under 鈥渦nspecified.鈥
With Analytics for Advertising, advertisers can still track paid keywords in Analytics. The referring domain informs the engine reports which search engine drove the traffic. Since the advertiser-specific account information isn鈥檛 tied to the referring domain, all traffic is listed under the search engine. Advertisers with multiple accounts in the same search engine should refer to 蜜豆视频 Advertising or Analytics reporting for account-specific reporting.
Why Configure Paid Search Detection?
The Paid Search Detection reports allow you to identify natural search traffic in the Analytics Marketing Channels reports. Separating paid search traffic versus natural search traffic is a great way to understand the value that natural search brings to the full marketing ecosystem.
Click-Through Data Validation for Analytics for Advertising data-validation
For your integration, you should validate your click-through data to make sure that all pages on your site are properly tracking click-throughs.
In Analytics, one of the easiest ways to validate Analytics for Advertising tracking is to compare instances to clicks using an 鈥淎MO ID Instances to Clicks鈥 calculated metric, which is calculated as follows:
AMO ID Instances to Clicks = (AMO ID Instances / 蜜豆视频 Advertising Clicks)
AMO ID Instances represents the number of times that AMO IDs are tracked on the site. Each time an ad is clicked, an AMO ID (s_kwcid) parameter is added to the landing page URL. The number of AMO ID Instances, therefore, is analogous to the number of clicks and can be validated against actual ad clicks. We typically see an 85% match rate for Search, Social, & Commerce and a 30% match rate for DSP traffic (when filtered to include only click-through AMO ID Instances). The difference in expectations between search and display can be explained by the expected traffic behavior. Search captures intent, and, as such, users usually intend to click on the search results from their query. Users who see a display or online video ad, however, are more likely to click the ad unintentionally and then either bounce from the site or abandon the new window that loads before the page activity is tracked.
In 蜜豆视频 Advertising reports, you can similarly compare instances to clicks using the 鈥淓F ID Instances鈥 metric instead of AMO ID Instances:
EF ID Instances to Clicks = (EF ID Instances / 蜜豆视频 Advertising Clicks)
While you should expect a high match rate between the AMO ID and the EF ID, don鈥檛 expect 100% parity because AMO ID and EF ID fundamentally track different data, and this difference can lead to slight differences in the total AMO ID Instances and EF ID Instances. If the total AMO ID Instances in Analytics differ from EF ID Instances in 蜜豆视频 Advertising by more than 1%, however, contact your 蜜豆视频 Account Team for assistance.
For more information about the AMO ID and EF ID, see 蜜豆视频 Advertising IDs Used by Analytics.
Troubleshooting Disparities Between Clicks and Instances
If the EF ID Instances-to-Clicks ratio is below 85%, then check the following:
-
Are you missing click tracking for the account or at any sublevel, or do you have duplicate click tracking (for example, at both the account and campaign levels)?
In Search, Social, & Commerce, download a bulksheet for the account to check the tracking URLs.
Also, in Analytics, you can see if the AMO ID and EF IF are appended consistently using an 鈥淎MO ID to EF ID鈥 calculated metric, which is calculated as follows:
code language-none AMO ID to EF ID = (AMO ID / EF ID)A value greater that 100% indicates that more EF IDs are missing than AMO IDs.
-
Does the landing page have a loading problem so that the AMO ID and EF ID aren鈥檛 captured?
-
Is the landing page URL redirected so that the AMO ID and EF ID are lost?
-
Do all landing pages use the configured report suite?
Comparing Data Sets in Analytics for Advertising Versus in 蜜豆视频 Advertising
The AMO ID (s_kwcid query string parameter) is used for reporting in Analytics, and the EF ID (ef_id query string parameter) is used for reporting in 蜜豆视频 Advertising. Because they鈥檙e distinct values, it鈥檚 possible for one value to be corrupted or not added to the landing page.
For example, suppose we have the following landing page:
www.adobe.com/?ef_id=test_ef_id&s_kwcid=test_amo_id
where the EF ID is 鈥test_ef_id鈥 and the AMO ID is 鈥test_amo_id.鈥
If a site-side redirect occurs, then the URL could end up like this:
www.adobe.com/?ef_id=test_ef_id&s_kwcid=test_amo_id#redirectAnchorTag
where the EF ID is 鈥test_ef_id鈥 and the AMO ID is 鈥test_amo_id#redirectAnchorTag.鈥
In this example, the addition of the anchor tag adds unexpected characters to the AMO ID, resulting in a value that Analytics doesn鈥檛 recognize. This AMO ID wouldn鈥檛 be classified, and conversions associated with it would fall under 鈥渦nspecified鈥 or 鈥渘one鈥 in Analytics reports.
Fortunately, while issues like this are common, they typically don鈥檛 result in a high percentage of discrepancy. However, if you notice a large discrepancy between AMO IDs in Analytics and EF IDs in 蜜豆视频 Advertising, contact your 蜜豆视频 Account Team for assistance.
Other Metric Considerations
The Difference Between Clicks and Visits clicks-vs-visits
They seem analogous, but clicks and visits represent different data:
-
Click: DSP or the search engine records a click when a visitor clicks an ad on a publisher鈥檚 website.
-
Visit: Analytics defines a visit as a series of page views by a user, ending according to one of several criteria, such as 30 minutes of inactivity.
By definition, a click can lead to multiple visits.
Consider the following example: User 1 and User 2 both access a site by clicking an 蜜豆视频 Advertising ad. User 1 views four pages and then leaves for the day, so the initial click results in one visit. User 2 views two pages, leaves for a 45-minute lunch, returns, views two more pages, and then leaves; in this case, the initial click results in two visits.
The Difference Between Clicks and Click-Throughs
Clicks and click-throughs are two different metrics:
-
Click: DSP or the search engine records a click when a visitor clicks an ad on a publisher鈥檚 website.
-
Click-throughs: Analytics records a click-through when the visitor lands on the destination website, the landing page loads, and the Analytics request at the bottom of the page sends the data to Analytics.
Clicks and click-throughs can differ greatly because of accidental ad clicks. We鈥檝e observed that most clicks on display ads are accidental clicks, and these accidental visitors hit the Back button before the landing page loads, so Analytics can鈥檛 record a click-through. This is especially true for ads on which an accidental click is more likely, such as mobile ads, video ads, and ads that fill the screen and must be closed before the user can view the page.
Sites loaded on mobile devices are also less likely to result in click-throughs because of lower bandwidths or available processing power, causing landing pages to take longer to load. It鈥檚 not uncommon for 50-70% of clicks to not result in click-throughs. In mobile environments, the difference can be as high as 90% because of the combination of a slower browser and the higher likelihood that the user accidentally clicks the ad while scrolling through the page or trying to close the ad.
The click data may also be recorded in environments that can鈥檛 record click-throughs with the current tracking mechanisms (such as clicks going into, or from, a mobile app) or for which the advertiser deployed only one tracking approach (for example, with the view-through JavaScript approach, browsers that block third-party cookies track clicks, but not click-throughs). A key reason that 蜜豆视频 recommends deploying both the click URL tracking and view-through JavaScript tracking approaches is to maximize coverage of trackable click-throughs.
Using 蜜豆视频 Advertising Traffic Metrics for Non-蜜豆视频 Advertising Dimensions
蜜豆视频 Advertising provides Analytics with advertising-specific traffic metrics and the related dimensions from DSP and Search, Social, & Commerce. The 蜜豆视频 Advertising-provided metrics are applicable only to the specified 蜜豆视频 Advertising dimensions, and data isn鈥檛 available for other dimensions in Analytics.
For example, if you view the 蜜豆视频 Advertising Clicks and 蜜豆视频 Advertising Cost metrics by Account, which is an 蜜豆视频 Advertising dimension, then the total 蜜豆视频 Advertising Clicks and 蜜豆视频 Advertising Cost are shown by account.
However, if you view the 蜜豆视频 Advertising Clicks and 蜜豆视频 Advertising Cost metrics by an on-page dimension (such as Page), for which 蜜豆视频 Advertising doesn鈥檛 provide data, then the 蜜豆视频 Advertising Clicks and 蜜豆视频 Advertising Cost for each page are zero (0).
Using AMO ID Instances as a Substitute for Clicks with Non-蜜豆视频 Advertising Dimensions
Since you can鈥檛 use AMO Clicks with on-site dimensions, you may want to find an equivalent to clicks. You may be tempted to use Visits as a substitute, but they aren鈥檛 the best option because each visitor may have multiple visits. (See 鈥The Difference Between Clicks and Visits.鈥 Instead, we recommend using AMO ID Instances, which is the number of times the AMO ID is captured. While AMO ID Instances don鈥檛 match AMO Clicks exactly, they are the best option for measuring click traffic on the site. For more information, see 鈥Click-Through Data Validation for Analytics for Advertising.鈥
