Microsoft Opened a PMax Import Door Right as Google Forced the AI Max Rebuild

Microsoft rolled out expanded Google PMax import support on April 16, 2026, now covering new customer acquisition goals, audience list mapping, and landing page reporting. The timing matters: Google's forced Dynamic Search Ads migration to AI Max hits in September, and advertisers already rebuilding campaigns can land the rebuild on a second platform with roughly 30 to 70 percent cheaper CPCs. The 30 percent new-customer conversion value floor is the setting most imports will leave blank, then blame Microsoft for.

The timing is the product

Microsoft has had a PMax import tool since October 2022. What changed this week is that the tool now handles the one campaign type every PMax advertiser is currently being asked to rebuild: campaigns with new customer acquisition goals. Search Engine Land confirmed the full NCA import path is live for all advertisers. Google announced in early April that Dynamic Search Ads will automatically migrate to AI Max by the end of September. Most serious PMax accounts also run NCA goals inside the purchase objective. Those two migrations are happening in parallel, and the overlap is where Microsoft is betting.

Before this month, the Microsoft import path would throw errors on anything NCA-related. You would clone the campaign, realize the customer list logic did not port, and abandon the import. That's fixed. Audience lists that used to throw unsupported-list errors now either map cleanly to Microsoft remarketing lists or trigger a fallback option. "All visitors" and "all converters" convert 1:1, according to Microsoft's January product release.

I don't think Microsoft accidentally shipped an NCA import feature in the same quarter Google forced a PMax rebuild window. The polite framing is "reducing friction." The honest framing: if you're rebuilding the campaign anyway, Microsoft wants to be the copy-paste destination.

What actually imports, and where the map breaks

Reading Search Engine Land's breakdown alongside Microsoft's own January update, here is what the current import handles:

• Google PMax campaigns with NCA goals, including both "bid higher for new customers" and "new customers only" modes
• Website visitor segments, which become Microsoft remarketing lists
• Final URL reporting with spend, clicks, impressions, conversion value, and ROAS by landing page
• Existing Microsoft NCA settings are preserved, not overwritten by the import
• Expanded location targeting coverage with Microsoft's own phrasing of "more reliable imports"

And here is what does not yet work cleanly:

• Customer Match lists do not port. Microsoft treats customer match differently, and the import gives you a fallback option, which usually means the campaign runs without that audience layer until you rebuild it natively.
• Google PMax seasonality signals do not transfer. Seasonality adjustments on Microsoft are portfolio-bid compatible now, which is useful, but the historical signal stays behind.
• Asset-level diagnostic data from Google's PMax reporting does not fully mirror. You get the structure, not the trail.

Plan on a cold start. The learning phase resets when the campaign lands on Microsoft, and most imported PMax campaigns need two to three weeks of runway before the ROAS numbers mean anything. That's the hidden cost nobody puts on the product page.

The 30 percent rule hidden in Microsoft's NCA docs

The most useful detail in Microsoft's own PMax NCA guidance is one line most advertisers miss: set the new customer conversion value to at least 30 percent of the average revenue from a usual sale. Treat that number as a floor, not a reward.

The logic, spelled out: if your average sale is $120, Microsoft wants you to tell the algorithm that acquiring a net-new customer is worth at least $36 on top of that sale. Without the floor, PMax will trend toward whichever conversion is cheapest, and that's almost always a returning customer through a branded search click. With the floor, you nudge the bid algorithm to value acquisition even when the immediate ROAS looks worse.

Google's equivalent setting lets you put any number in. Microsoft is giving you the benchmark. If you're importing a PMax from Google and you never set a new-customer value there, the import carries that absence across. You'll see flat acquisition performance and assume Microsoft is the reason. The problem is the missing floor, not the platform.

One more detail worth knowing: Microsoft recommends refreshing customer lists daily or weekly. Most agencies I've seen running Microsoft do monthly refreshes. That's not enough. The algorithm reads list freshness as a signal about the real world, and stale lists quietly shrink the acquisition pool.

Why this probably still isn't a clean swap

Microsoft's ad revenue is projected around $19.5 billion in 2026 against Google's roughly $225 billion in search ad revenue. That ratio hasn't moved much in three years. The pitch that Microsoft Ads is "untapped cheap inventory" has been told since 2019, and most advertisers who tested it ran into the same wall: the volume isn't there for high-intent B2B or niche ecommerce. CPCs being 30 to 70 percent cheaper doesn't matter when impression volume is 90 percent lower for your keyword set.

What has changed: the slice of search traffic flowing through Copilot, ChatGPT search, and Bing-integrated AI has tipped enough that the impression side of the math is less dismissive than it used to be. Microsoft's ad revenue grew 13.4 percent between 2024 and 2025, which is not nothing. The NCA import is Microsoft betting that PMax advertisers running the AI Max migration will finally test the platform as a secondary channel now that the switching cost just collapsed.

From what I've seen, this isn't a reason to shift budget off Google. Run a 90-day parallel test on Microsoft using a campaign clone you're already rebuilding anyway. The marginal effort is low. The downside is a few hours and some spend that would have gone somewhere else. Our earlier take on the AI Max transition math itself still holds on the Google side, but now you have a second place to run the same rebuilt campaign.

If you're already rebuilding, do these five things by Friday

A short list, because most teams don't have three weeks to sit with it:

1. Pull the NCA-enabled PMax campaign from Google that you're migrating to AI Max. Clone it with the Microsoft import tool today, not after the September deadline. The lift is a few clicks now; it will be a full rebuild if you wait.
2. Set the NCA conversion value at a minimum of 30 percent of your average sale. If you already set it higher in Google, keep the higher value on Microsoft.
3. Upload customer match lists natively on Microsoft. The import will not do it, and leaving it empty is worse than spending 30 minutes doing it manually.
4. Give it three weeks of learning with 25 percent of what you're currently spending on Google. Don't make any cut or scale decision inside the learning window.
5. Check the new Final URL report at week two to see if spend is landing on pages you want. That's where PMax on a new platform quietly wanders, and the Final URL report is how you catch it.

Microsoft didn't build this import because it wanted to be helpful during the AI Max transition. It built it because Google just opened the largest window in five years where PMax advertisers are looking at their account structure with fresh eyes. My guess: Microsoft converts maybe 8 to 12 percent of serious PMax accounts into a real second-rail test by Q4. Whether any of those tests survive the learning window is the only question that actually matters.

Notice Me Senpai Editorial