Performance Max final URL expansion: leave it on or lock your landing pages?

Picture a PMax campaign for a store with about 40 products. Final URL expansion is on by default, so Google can send traffic to any page on the site it thinks will convert. A week in, the landing page report shows clicks going to the shipping policy page and to a category nobody sells from anymore. Those pages might pick up a few orders, but the cost per purchase can end up well above what the product pages deliver.

With a catalog that small, I'd turn expansion off, or add URL exclusions for the pages you don't want, and then look at the landing page report in week one. The algorithm will spend on whatever converts even a little, whether or not that's where you want the budget going.

Has anyone with a much bigger catalog kept it on and found the extra reach pays off?

Iris's point holds for a 40-product store, but I'd split it by margin rather than catalog size. A page that converts at a higher cost per purchase only becomes a problem if the orders it brings in are low margin. So the real question is whether you'd still take a shipping-page order at your target CPA, or whether it's just noise.

Margin is the right lens, but I'd check the conversion setup before judging the page. A shipping policy page rarely sells anything on its own, so if it's showing purchases, the tag is probably firing on something loose, like a thank-you page reached from a footer link or a generic form submit. PMax will happily chase whatever's cheapest to count. Before you judge any page by its CPA, confirm the conversion action is purchase-only and passes order value. If it isn't, fix that first, and the margin question mostly answers itself.

I wouldn't read a shipping-page purchase as a broken tag. The landing page report shows where a session started, not where the order happened. Someone clicks a PMax ad, lands on the shipping page, reads the returns rules, then moves to a product page and buys. That purchase still belongs to the session, so the shipping page gets credit it didn't earn on its own. Sure, confirm the conversion action is purchase-only and passes value, but look at the full path before deciding the page is just noise.

I'm not sure "that purchase still belongs to the session" holds up here. The landing page report in Google Ads credits a conversion to the page the click landed on, whether the buyer read three pages first or went straight to checkout. So the shipping page isn't getting credit for helping a sale. It's getting credit because PMax sent the click there.

That changes the fix. If the policy page is pulling clicks from people who then buy elsewhere, the problem is which traffic PMax is choosing, and exclusions or a different final URL setup would address that. You can check it in GA4 by looking at the paths for those sessions. If most of them read the policy page and leave, the orders are coincidence and that page's CPA is the real number.

Fair, the landing page report credits the page the click landed on, so Marco's path argument doesn't change what that report shows. For PMax I'd keep the rule simple: exclude any non-product, non-category page by default. Those pages rarely sell on their own, and PMax will keep sending clicks there as long as a few sessions buy. Only let a page back in if its clicks hit your target CPA for two or three weeks in GA4, with the path included. It's easier to earn a page back than to chase a leak after the budget has already gone there.

Agreed on the rule, but I'd tie the window to your purchase lag instead of a flat two or three weeks. If shoppers usually browse for ten days before buying, a page can look dead in week two while its orders are still on the way. Check the lag in GA4 first, then run the test for at least that long plus a week before you judge the page.

The lag matters for another reason too: Google Ads only counts conversions inside its conversion window. Orders that land after that cutoff drop out, so a page can look weaker in Google Ads than in GA4 over the same dates. Pick one source for the test and stick with it, or the two numbers will disagree and you'll spend a week arguing about which is right.

I'd also freeze everything else in the campaign for that window. If you raise the budget, add an asset group or move the tCPA in week two, the late orders come from a different setup, so you can't tell whether the page earned them or the change did. Change only the exclusion, then judge.

Are you pulling the lag from the Google Ads conversion lag report or from GA4?

Atlas, it depends on what you're deciding. For a tCPA or conversion window call, use the conversion lag report in Google Ads, because that's the data the bidding actually sees. If you want to know whether shoppers really come back later, GA4's path data gives you the more honest read.

Agreed, but the lag report has a catch. The numbers for recent clicks are still filling in. If you pull it on day 10 and see 60% of conversions in the first three days, the rest haven't happened yet, so that's no evidence the lag is short. Only trust the curve for clicks old enough to have closed the window you care about, which is often 30 days when the consideration cycle is long.

The same problem hits the page test. If you judge the exclusion on the last seven days, you're comparing immature clicks with mature ones. Compare the same click dates across both periods, and wait for the window to close before you call a page dead.