5 Analyses You Can Perform with Google Ad Manager Log Level Data

This article first appeared in AdMonsters. It has since been expanded with additional examples and insights for the Burt Intelligence Resources Center.

Unlock the full potential of Google Ad Manager’s log-level data with these five actionable analyses. Learn how to optimize your ad strategies and increase revenue using Data Transfer Files.

Google Ad Manager’s Data Transfer Files (DTFs), also known as GAM log-level data, capture every ad request, auction, impression, and related event, providing publishers with a granular view of advertising performance. DTFs aren’t a new offering; many tech-savvy publishers already use them.

However, in many conversations with publisher ad ops professionals, we often hear that while they want to utilize the data treasure stored in the DTF, they’re just not exactly sure what to do with it. Many industry publications and conference keynote speakers praise the value of log-level data, but few explain exactly what you should and could do with it.

So, as a quick guide for the perplexed, here are five ways you should be working with your DTFs:

1. Segment Analysis

With log-level granularity, you can see how targeting parameter combinations perform beyond what API or UI-based reporting can offer. You can measure how much a certain segment increases CPM, compare its performance by itself and in combination with other segments, and determine whether the lift is actually coming from another third-party ID that carries its own fee.

You can also examine whether certain segments work only on specific parts of your inventory or provide a boost across the site, as well as which other targeting parameters they perform well with.

This becomes especially useful when users match multiple audience segments at the same time. Built-in ad server reporting may show performance for each audience independently, even though a user rarely belongs to only one audience. Without deduplication, it can be difficult to determine which segment is actually responsible for the lift.

For one digital news publisher, Burt used log-level custom targeting keys to build deduplication logic that gave first-party segments precedence over third-party tags. Incoming key-values were parsed, impressions matching multiple segments were flagged, and a dedicated deduplication field was created for clean reporting.

The publisher then worked with its programmatic partner team to analyze which segments drove value in the open marketplace. By comparing CPMs and fill rates across deduplicated audiences, the team could focus audience development on the most valuable segments and package top performers into private marketplace deals.

2. Key-Value Pairs Analysis

In the DTF, all the key-values you have set up on your site are available in a deduplicated manner, so you do not get overlaps between combinations. This gives you the flexibility to combine them freely and see how different combinations perform.

For example, you can identify which combinations of positions and custom parameters lead to higher CPMs, see which targeting combinations Prebid vendors are bidding on, and understand how performance varies across content, geography, device, audience, placement, and ad unit.

The same granularity can also support forecasting. Google Ad Manager forecasts are not always well suited to publishers whose traffic changes sharply around breaking news or major events. A traffic spike may be real, but that does not mean it should become part of the baseline for the following month.

For one publisher, Burt analyzed historical logs to separate stable traffic from spikes. Traffic above an agreed threshold relative to the prior-day average was treated as anomalous and excluded from baseline projections. This gave the publisher a more predictable month-ahead forecast without assuming that every major news event would repeat.

Burt also connected the log data to the publisher’s order management system. Matching granular impression data against actual booked orders made it possible to identify potential overcommitment and examine sell-through across specific portions of inventory.

Because each log entry included the ad unit hierarchy, relevant targeting parameters, and URL, the publisher could drill down beyond a sitewide estimate and understand availability and contention across individual inventory segments.

3. Latency Checks

Given the granularity of the data, you can measure the latency of your bid process to ensure you aren’t leaving money on the table and creating a bad user experience. For example, this could allow you to test latency when adding new bidders or turning on Google’s Protected Audience API.

The built-in reporting from the ad server only measures the load time of creatives from the time the creative tag is triggered until the creative loads. Before that happens, header bidders, verification vendors, identity vendors, and other technologies may already have added latency to the page.

In one implementation, log timestamps enabled the publisher to analyze the path from the ad request through creative load completion. The team identified ad calls loading serially instead of in parallel and flagged slow tags. Re-architecting the page to fire more ad calls concurrently reduced latency by approximately 800 milliseconds, which improved the user experience and created more opportunity for bids and impressions during each visit.

4. Incremental Revenue Analysis

Compare your winning bids with other bids to determine potential efficiencies in your ad stack. Do you have a slew of bidders bidding within $0.01 on most auctions? Do all your vendors bid on the same auctions, and none on others? Well, all of these might be signs you should look over your ad tech stack and make it leaner.

Combining auction data with audience and targeting data can extend the analysis further. The publisher described above used the same log-level view to identify which first-party segments created higher CPMs and fill, helping the team distinguish audiences that added value from those that simply added complexity and informing which segments should be packaged into PMP deals.

5. Loss Reason Analysis

In the GAM UI/API reporting, you can get some basic metrics for loss reasons. However, to understand what really happened, you need to dig deeper and see all the targeting and other parameters that were set on the request. The only way to do this is by digging into the log data.

Data Done Right: Operationalizing Log-Level Data with Burt

So now that we’ve established how powerful and useful log-level data can be, how do you actually put it to work?

Google Ad Manager Data Transfer Files can generate billions of rows of data each month. Managing that volume requires more than storage. You need a platform that can ingest, organize, and query the data efficiently while making it accessible to the teams that rely on it.

Burt provides that foundation (see also: Why AI Needs Reconciled Revenue Data). The platform stores log-level data well beyond Google Ad Manager’s standard reporting windows and manages the ETL, schema management, indexing, mapping files, and conditional formulas needed to turn raw log files into usable business data. Burt also integrates with order management systems, Prebid analytics, and other advertising platforms, creating a single source of truth across the revenue organization.

Frequently used datasets can be pre-aggregated, allowing common analyses to run in seconds instead of repeatedly processing billions of raw records. That means teams spend less time waiting for queries and more time acting on insights.

Access matters as much as infrastructure. Revenue operations, ad operations, engineering, sales, and finance all rely on the same underlying data for different purposes. Burt gives each team self-service access while ensuring everyone works from consistent definitions and shared business logic. In one publisher implementation, this reduced manual handoffs and accelerated cross-functional projects from weeks to days.

The result is that publishers can focus on optimizing revenue, inventory, and operational performance instead of building and maintaining complex data infrastructure.

Ready to Unlock the Value of Your Log-Level Data?

Google Ad Manager’s log-level data is a treasure trove of information that can significantly enhance your advertising operations. By performing analyses such as those suggested above, publishers can gain a much deeper understanding of their inventory, audiences, and performance.

Burt helps publishers put that data to work. By transforming billions of log records into a trusted, accessible source of truth, Burt enables better forecasting, audience analysis, performance optimization, and revenue operations.

Interested in getting more from your Google Ad Manager Data Transfer Files? Get in touch to learn how Burt can help.