How to Use CM360 for Campaign Attribution: A Practical Guide

Most brands judge programmatic by the last click. Then the monthly review comes around, OTT and YouTube show a handful of conversions each, retargeting shows hundreds, and someone suggests moving the video budget into retargeting.

That conclusion is usually wrong, and Campaign Manager 360 (CM360) is one of the better tools for showing why. It won't settle the argument on its own. Set up properly, though, it shows which channels start journeys, which ones close them, and how the credit moves when you change the rules.

What does CM360 do for campaign attribution?

CM360 records the impressions and clicks your ads receive and links them to the conversions your Floodlight tags record on your site or app. Attribution is the set of rules that decides which of those ad interactions gets credit for each conversion.

Floodlight is Google's conversion tag. You place it on the pages or events that matter, such as a loan application submit or a test ride booking, and each conversion it records can be traced back to ad interactions inside a lookback window. The lookback window is simply how many days back an impression or click still counts.

The real strength of CM360 is that it is an ad server. If your ads run through DV360, another DSP and a few direct publisher bookings, CM360 can serve and count all of them. That gives you one view of the path to conversion, instead of three platforms each claiming the same sale.

Do you need CM360, or is DV360 enough?

If you buy everything through DV360, you may not need CM360 at all. DV360 can manage Floodlight on its own through Floodlight groups, and you can build attribution models there too. Our guide on tracking conversions in DV360 covers that setup.

CM360 earns its place when your media is spread across more than one buying route. It is a separate Google Marketing Platform product with its own access and costs, so decide based on what you actually buy.

How do you set up CM360 for attribution?

Setup matters more than the model. A clever model on top of messy tags gives you confident, wrong answers.

1. List the conversions that matter. Pick three to six actions that show real intent. For a two-wheeler brand that might be test ride bookings and dealer enquiries. For an NBFC, loan application starts and submits. Keep remarketing pixels and page-view tags out of this list.
2. Create Floodlight activities and choose how they count. CM360 can count every conversion, one per user per day, or one per session. Google's help suggests counting every conversion for purchases and one per user for leads, so a person who refreshes a thank-you page doesn't become five leads.
3. Set lookback windows. Click and impression windows are set separately, in days, up to a maximum of 90. Placements inherit the advertiser's windows, and an override can't be longer than the advertiser's value. Match the window to the buying cycle: a ₹600 skincare order and a ₹12 lakh car should not share one window.
4. Link DV360 at the start. You share the Floodlight configuration from CM360 with your DV360 partner, then select it when you create the DV360 advertiser. Google's help is clear that this link has to be made when the DV360 advertiser is created and can't be undone later. The steps are in Google's guide to connecting a Campaign Manager 360 advertiser to DV360.
5. Turn on enhanced attribution and link GA4 if you use it. Enhanced attribution adds a click identifier to your ad traffic, which the GA4 link needs. Once linked, GA4 conversions can appear in CM360 as Floodlight activities. Our walkthrough on connecting DV360 with GA4 covers the Analytics side.
6. Test before you spend. Trigger each conversion yourself, check it lands in Floodlight reports with the right count and value, and fix any implementation alerts CM360 shows.

Which attribution model should you use?

Start by understanding the default. Standard Floodlight reporting gives full credit to the last click before a conversion, and if there was no click, to the last impression. Clicks always beat impressions. That is why a retargeting banner clicked on the way to checkout takes the credit from a YouTube ad watched a week earlier.

In the Attribution Modeling Tool, found under Reporting & Attribution in CM360, you can compare this default with other models without changing your live reports.

A few rules from Google's help are worth knowing before you build anything. Each advertiser can have five models: the system Floodlight model and four you create. You can create only one data-driven model. It takes a day or two to train, and editing it can trigger retraining that takes up to nine days. Models can't be edited in place, so you remove one and create it again.

Then there is the primary model. This is the model DV360 uses for bid optimisation and conversion reporting on line items tied to that Floodlight. A data-driven model can be primary only when Floodlight is managed in CM360. If you switch the primary model on a running line item, conversion tracking changes from that date and bidding adjusts gradually over about 30 days. So don't switch it in the middle of a festive burst.

How do you read CM360 attribution reports?

Three views do most of the work.

Path to Conversion reports, built in Report Builder, show the series of impressions and clicks each converter saw before converting, with a conversion window of up to 90 days. Choose the pivoted version when you want one row per conversion. It is much easier to read in a spreadsheet.

Assisted Conversions, in Instant reporting, count the clicks and views that came before the final touchpoint. The assisted conversion ratio divides assisted conversions by last-touch conversions. A ratio above one means that channel works more as an assist than a closer. These metrics only work with the last-touch model.

The model comparison in the Attribution Modeling Tool shows how credit moves between channels when you change models. Look for the big swings. They tell you which channels your current reporting under-values or over-values.

Worked example: OTT, YouTube and display for test ride bookings

Here is an illustrative example. The numbers are made up to show the method, not results from any campaign.

A two-wheeler brand runs a four-week campaign of ₹40 lakh: ₹15 lakh on OTT through Programmatic Guaranteed deals, ₹12 lakh on YouTube through DV360, ₹8 lakh on programmatic display and ₹5 lakh on retargeting. Floodlight records 1,000 test ride bookings.

The formula is spend divided by attributed bookings. Under the default model, OTT looks almost twenty times as expensive as retargeting. Under position based, the gap shrinks to about four and a half times.

Neither column is the truth. Retargeting mostly reaches people who already visited the site, often because they saw the OTT or YouTube ad first. And an ad on a connected TV screen almost never gets clicked, so click-first models give it very little credit.

So don't cut OTT because of the first column. Check the assisted conversion ratio for OTT and YouTube, and if the budget allows, test the question properly with a holdout group or a brand lift study. Attribution tells you how credit is shared. It does not prove what would have happened without the ad. Our guide to measuring incremental reach shows a simple way to check whether a channel adds new people.

What mistakes should you avoid?

• Counting remarketing or page-view tags as conversions. They swell the totals and give the model noise to learn from.
• Setting long impression windows on cheap display. With a 90-day view window, an impression served three months ago can still claim a conversion, which flatters programmatic.
• Expecting CM360 and GA4 to match. They use different tags, windows and models, and GA4 reports only support click-through measurement for CM360 traffic.
• Changing the primary model mid-flight. Bidding takes weeks to settle.
• Forgetting the link rule. A DV360 advertiser created without the CM360 link stays unlinked.
• Missing platform changes. Google's CM360 help says that from October 2026, Floodlight uses Google signed-in data signals to help recover cross-device web conversions that previously went unattributed. Compare periods on either side of that change with care.

How does CM360 fit with the rest of your reporting?

Use each tool for the job it does best. DV360 reports tell you what was delivered: impressions, reach, frequency, viewability and cost. CM360 tells you how conversions were reached across every buy. GA4 tells you what people did on your site after they arrived.

When we run campaigns through our DV360 Managed Service or a Campaign-Based Engagement, delivery numbers such as spend, impressions, unique reach, frequency, CPM and video completion sit in our DV360 Reports Section, pulled from DV360 through its APIs and updated daily. Conversion paths still come from Floodlight. Our guide to measuring programmatic ROI brings the two together, and building DV360 campaign reports covers the delivery side.

Next step

If you are setting up a new DV360 advertiser, decide on CM360 first, because the link can only be made at the start. Our DV360 advertiser account setup includes Floodlight conversion tracking and Campaign Manager 360 links where needed. If you have questions about your own tags or models, talk to our team.