CM360 vs DV360 Discrepancy Why Numbers Don’t Match ?

I still remember the first time one of my team members came to my desk, looking genuinely panicked.

They had two tabs open—DV360 in one, CM360 in the other—and the delivered impressions were worlds apart. They were convinced they had broken the setup.

As a leader, you don’t want a talented specialist doubting their own work. At that moment, I realized that as a manager or trainer, we often teach people how to pull reports, but we forget to teach them why the platforms see the world differently.

Nothing was broken; we were just trying to use two different maps to find the same destination.

If you’ve ever felt the pressure of a data discrepancy, this guide is for you. Here is why 1 is not equal to 1 in the Google Marketing Platform (GMP) ecosystem, and how to explain it to your team member.

Two Platforms, Two Different Perspectives

The fundamental reason for the gap is that DV360 and CM360 are built with entirely different intentions. Think of them as two different people at an auction.

DV360 is the Buyer: It looks at the world from the point of view of a bidder. It cares about bid wins, auction dynamics, and real-time optimization. It answers: “What did I buy?”

CM360 is the Auditor: It looks at the world from the point of view of measurement. It sits at the end of the chain to verify what was actually delivered and how it affected the user. It answers: “What was actually seen?”

The “Big 5” Causes of the Data Gap

When your numbers are off, it’s usually due to one of these five technical realities:

1. The “Bid Win” vs. “The Render”

DV360 logs an impression the moment it wins a bid and sends the ad to the publisher. CM360 only counts that impression after the ad successfully renders on the user’s screen. If a user closes the tab or scrolls away before the ad loads, DV360 counts it, but CM360 won’t.

2. Strictness of Filters (IVT)

CM360 is the “Source of Truth” for billing for a reason: it has much stricter Invalid Traffic (IVT), bot, and spam filters. DV360 might report “raw” numbers in its initial view, while CM360 aggressively strips out non-human activity.

3. The De-Duplication Logic

This is a big one. CM360 de-duplicates clicks and conversions across different placements and campaigns to ensure you aren’t double-counting the same user action.

DV360, focusing on its own line items, may report “raw” clicks that haven’t been scrubbed for duplicates yet.

4. Technical Barriers (Ad Blockers & Cookies)

CM360 relies heavily on Floodlight tags and tracking pixels. If a user has a high-security browser setting or an ad blocker, the CM360 tag might be blocked from “firing,” even though the DV360 ad server successfully pushed the creative.

5. Attribution & Lookback Windows

They often use different lenses. If your DV360 lookback window is set to 30 days but your CM360 Floodlight is set to 14, they will never align. Furthermore, CM360 sees the “bigger picture” of the whole media mix, whereas DV360 only sees what happened within its own ecosystem.

6. Setup Errors:

Incorrect linking, modified tags, or duplicate tags.

Incorrect Linking: For DV360 and CM360 to work together, they need to be “linked” at the Advertiser level. Think of this as a walkie-talkie connection between the DV360 and the CM360.

The Problem: If the linking is incorrect or broken, the DV360 is winning auctions, but they aren’t telling the CM360 what to look for.

The Result: The Bidder (DV360) reports thousands of buys, but CM360 ignores because they don’t have the “manifest” to verify them.

Simple Explanation: It’s like buying a gift online but giving the wrong delivery address. The store says “Shipped,” but your house says “Nothing arrived.”

Modified Tags: CM360 tags are pieces of code (labels) placed on the creative or the website. If someone accidentally changes a single character in that code, it’s “modified.”

The Problem: Imagine DV360 puts a label on a box that says “Package ID: 123.” But on the way, the label(tags) gets smudged or torn. When it reaches CM360, they can’t read the ID.

The Result: The CM360 sees a delivery, but because they can’t verify the “label(tag),” they refuse to log it in the official book.

Simple Explanation: It’s like trying to board a plane with a boarding pass where your name is misspelled. The airline knows you bought a ticket, but the gate agent won’t let you through because the data doesn’t match.

Duplicate Tags: This usually happens when a tag is accidentally pasted twice into a website’s code or a container (like Google Tag Manager).

The Problem: Every time a impression is reported, CM360 accidentally clicks their counter twice.

The Result: This is the only time CM360 numbers will be higher than DV360. The DV360 says, “I only sent one impression,” but the CM360 logbook says, “I saw two.”

Simple Explanation: It’s like having a double-vision glitch. You bought one coffee, but your bank statement shows you were charged twice because the card reader swiped twice.

When Should You Actually Worry About CM360 vs DV360 Discrepancy ?

In the agency world, we have a rule of thumb: The 10% Rule.

Under 10% discrepancy: This is “business as usual.” It’s the result of the technical differences mentioned above.

Over 10% discrepancy: This is a red flag. It usually points to a setup error, mismatched time zones, or a broken Floodlight tag.

Best Practices to avoid CM360 vs DV360 Discrepancy Reporting:

1. The 48-Hour Rule: Never compare data that is less than 48 hours old. Both platforms need time to “clean” the data and apply final filters.
2. Define Your Source of Truth: Decide early. Usually, CM360 is the source of truth for billing and conversions, while DV360 is the source of truth for optimization and bidding.
3. Check Your Time Zones: Ensure your CM360 advertiser and DV360 partner settings are synced to the same timezone to avoid “shifting” data
4. Check Tag Implementation: Ensure Floodlight tags are perfectly implemented and synced, especially for video.

Ensure your CM360 advertiser and DV360 partner settings are synced to the same timezone to avoid “shifting” data.