Latest · Ad platforms

Performance Max Punishes New Lease-Ups and Rewards Stabilized Assets

September 2, 2026

PMax is the wrong tool for a lease-up. You're probably running it anyway because someone in a webinar called it "the future of Google Ads." Webinars love the future. It's cheaper than talking about the present.

Here's what PMax actually needs to work: enough conversion volume for the algorithm to learn what a good lead looks like, and enough history to know where to find more. A lease-up has neither. You're feeding a machine-learning model a handful of leads a week and asking it to spot a pattern that doesn't exist yet.

Stabilized properties are a different animal. Give PMax a year or two of clean, tagged conversion data and a CRM that actually closes the loop back to the ad platform, and it's got something to chew on. It'll find pockets of intent you'd never catch bidding manually on ten keyword themes.

The lease-up failure mode looks the same everywhere. Spend goes out across Display, YouTube, Discover, and Search inventory you can't see into. Google's own PMax reporting gives you almost no placement-level visibility, so when your marketing director asks why cost-per-lead spiked, your honest answer is "the algorithm did something."

For a stabilized property doing steady renewal-adjusted leasing, that's a fine trade for the extra reach. For a lease-up burning a fixed budget on a fixed clock, it's not.

Then there's signal quality. PMax optimizes toward whatever conversion event you hand it. Feed it "form submitted" and it will happily go find you a pile of cheap, low-intent form fills off Display inventory, because that's the easiest box to check. It's not lying to you. It's just answering the question you asked.

A stabilized property with real volume can layer in a second signal, like a scheduled tour or an applicant, and let the algorithm chase quality instead of raw count. A lease-up rarely has enough of either event to make that split mean anything.

Here's what I'd actually do. On a stabilized asset with at least six months of clean conversion history, turn PMax on with tour scheduled as the primary signal, not raw leads, and fund it properly instead of tossing it scraps. Then leave the bids alone for three to four weeks. That's the minimum runway for the algorithm to stop guessing and start learning.

Here's what I would not do. Put PMax on a new lease-up in its first ninety days. Full stop.

There's no conversion history for it to learn from, so you're not paying for leads, you're paying tuition. Run manual Search instead: tight geo, tight keywords, and daily eyes on what's working. Once you've got a real trickle of conversions and some renewal-stage data behind it, that's when you hand PMax the keys.

Budget size matters too. A small independent operator running one or two lease-ups a year never builds the account-level data volume that makes PMax worth the loss of control, no matter how patient they are. A bigger portfolio running several stabilized properties through PMax can pool that learning across the account in a way a single-asset operator never will.

The urge to put every campaign on autopilot makes sense. Budgets are tight, and PMax promises to do the thinking for you. But it only ever optimizes toward the data you feed it.

A lease-up doesn't have data worth optimizing yet. It has vibes and a leasing deadline.

Source: