CM360 vs DV360: why two Google counts differ
Both products are Google's, both sit on the same campaign, and they still disagree. That is not a bug: they are counting different events, and each one's help pages say so.
Pixtru guides · updated
Two products, two jobs
Display & Video 360 is the DSP: it bids, wins and pays for the impression, and its count is the one you are billed on. Campaign Manager 360 is the advertiser's ad server: its tag delivers or tracks the creative, and its count is the one most advertisers treat as the independent record. They count at different points in the chain, so any impression lost between the two points is in DV360's number and not in CM360's.
How much difference is normal
Google's Campaign Manager 360 help calls discrepancies of up to 20% normal and will not investigate a dataset of fewer than 1,000 impressions. The Display & Video 360 help gives the same 20% and adds that figures can take 48 to 72 hours to settle.
Where the difference comes from
| Cause | What to check |
|---|---|
| Different counting points | Expected, and the floor of the gap. Latency and early exits live here. |
| Mismatched date ranges or time zones | Both reports on the same dates, in the same zone, generated rather than live. |
| Modified or wrapped tags | The CM360 tag should be trafficked as exported. A tag edited, wrapped by another server or double-wrapped counts differently. |
| Cache-busting done wrong | Without a unique value per request, a browser can reuse a cached response and the ad server never hears of the impression. |
| Duplicate tags | The same tag in two places counts one impression twice. |
| Different invalid-traffic filtering | Each product removes invalid traffic by its own rules, at its own time. |
| Click trackers used as impression trackers | A tracker set up for the wrong event inflates or deflates one side. |
Comparing them fairly
- Match the date range and time zone exactly, and use generated reports on both sides.
- Compare at the level both products share: campaign, creative and day, not a total.
- Look for a creative or placement where the gap is far outside the rest. That is where a tag was wrapped, duplicated or dropped.
- Only once the obvious causes are ruled out is the remaining gap the structural one — the impressions lost between delivery and render.
Time zones, the cheapest cause to rule out
A report's day is the account's day. When the advertiser's CM360 account and the DV360 partner or advertiser are set to different time zones, every daily figure straddles two days on the other side, and a daily comparison shows a gap that a weekly one does not. Before investigating anything else, check both accounts' time zones, and compare whole weeks until they match.
A worked comparison
One campaign, one week, compared by creative on the same dates and time zone:
| Creative | DV360 | CM360 | Gap |
|---|---|---|---|
| Leaderboard 728×90 | 410,000 | 381,300 | 7.0% |
| MPU 300×250 | 520,000 | 479,440 | 7.8% |
| Half page 300×600 | 190,000 | 176,700 | 7.0% |
| Billboard 970×250 | 80,000 | 148,800 | −86.0% |
The total — 1.2 million against 1.19 million — looks like a near-perfect match. It is two errors cancelling out. Three creatives show the ordinary 7–8% floor. The billboard shows CM360 counting almost twice what DV360 delivered: its tag was trafficked twice, once in the creative and once as a tracker, so every impression was counted two times. Fix the duplicate and the real total gap appears at about 7%. A total that matches is not evidence that nothing is wrong.
Which count to use for what
| Question | Count to use |
|---|---|
| What did we pay for? | DV360's, unless the insertion order names another Controlling Measurement. |
| Did the ad reach the page and render? | CM360's, or a verification tag's rendered count. |
| Could it be seen, and by a person? | A verification count: measurable, viewable, and after invalid-traffic filtering. |
| Where did each missing impression go? | None of them alone. It needs the counts joined impression by impression. |
What neither report tells you
Even when CM360 and DV360 agree, neither says how many of the rendered impressions were measurable, viewable, or shown to a known bot. That needs a verification count alongside them — and, to say where each impression went rather than how many are missing, an impression-by-impression join. DV360 discrepancies covers why that join is hard with DV360 in particular, and served, measurable, viewable the counts underneath.
Related
- DV360 discrepancies — against any third-party count, and what to send Google support.
- What gap is acceptable — where the 10% and 20% figures come from.
- Discrepancy calculator — the gap and its cost for your own numbers.