Google Ads Remarketing: The 5 Types and Which Ones Work

In December 2025, Google quietly dropped the minimum audience size for ad targeting from 1,000 users to 100. Across every network. Almost nobody noticed.

If you run a Shopify brand doing $2-5M a year, that one change matters more than any remarketing tactic published in the last three years. Here's why: the segments you actually wanted to build (people who viewed a product, added to cart, and left in the past seven days) never used to hit 1,000 users on Search. So the platform wouldn't serve them. You were forced into one blunt "all site visitors, 30 days" list and told that was remarketing.

That constraint is gone. And most Google Ads remarketing setups running right now were built around a limit that no longer exists.

This article covers the five types of remarketing in Google Ads, what each one actually does, and which are worth your budget in 2026. It also covers the two that have quietly stopped being campaigns you build and started being settings inside campaigns Google runs for you. Along the way, we'll get into the thing no other guide on this topic will tell you: why your remarketing ROAS is almost certainly the most over-credited number in your account.

What is Google Ads remarketing?

Google Ads remarketing is the practice of showing ads to people who have already interacted with your brand, visited your site, watched your videos, used your app, or bought from you before. Google builds these audiences from your first-party data, then serves ads to them across Search, Display, YouTube, Gmail, and Shopping.

One piece of housekeeping before we go further, because it trips people up.

Google renamed the furniture. What used to be called "remarketing lists" are now "your data segments" in the Google Ads interface, and what used to be "audience types" are now "audience segments." If you're hunting for a remarketing tab in Audience Manager and can't find it, that's why. The official Google Ads documentation now uses "your data" throughout.

Remarketing vs retargeting: same thing. "Remarketing" is Google's word. "Retargeting" is what the rest of the industry says. Meta calls it retargeting, Google calls it remarketing, and the mechanic is identical. Don't let anyone bill you extra for knowing both words.

The 5 types of remarketing in Google Ads

There are five types of remarketing in Google Ads:

1. Standard remarketing shows display ads to past site visitors as they browse other sites and apps across the Google Display Network.
2. Dynamic remarketing shows past visitors the specific products they viewed, pulled live from your Google Merchant Center feed.
3. Remarketing lists for search ads (RLSA) adjusts your bids and ad copy on Search when a past visitor searches again.
4. Video remarketing targets people who watched your YouTube videos, subscribed, or engaged with your channel.
5. Customer Match uploads your own customer email list to reach known buyers across Google's properties.

That's the list every article on this topic gives you. Now here's what each one is actually worth.

Standard remarketing, and why PMax mostly ate it

Standard remarketing is the original version: someone visits your site, gets added to a list, and then sees your banner ads across the two million-plus sites and apps in the Display Network.

It still exists. But for most Shopify brands, running a standalone Display remarketing campaign in 2026 is a strange choice, because Performance Max already serves that same inventory automatically. If you're running PMax (and if you sell products on Shopify, you almost certainly are), your remarketing audiences feed it as an audience signal rather than a hard targeting rule.

That distinction matters. A signal tells Google's models where to start looking. It does not fence the campaign in. Advertisers who assume they've "set up remarketing" by adding a signal to PMax are often surprised to find their budget reaching plenty of people who've never heard of them.

Standalone Display remarketing is still defensible when you want hard control over frequency, creative, and spend on a specific segment. It's just no longer the default.

Dynamic remarketing, and the feed it depends on

Dynamic remarketing is the version that shows someone the exact sunglasses they looked at on Tuesday, at the price they saw, with the image from your product page.

For an ecommerce brand, it's the strongest of the five. It's also the most fragile, because it doesn't run on your website. It runs on your Google Merchant Center feed.

The mechanic: your remarketing tag fires a product ID when someone views a page. Google looks up that ID in your GMC feed and builds the ad from the feed data. If the ID in the tag doesn't match the ID in the feed, there's no lookup and no product ad. You get a generic banner instead.

It fails silently. No error, no alert, no red flag in the interface. Your campaign keeps spending and keeps reporting impressions. It just stops being dynamic.

This is the same class of problem as identical variant titles capping your Shopping reach. The feed is the input, and bad inputs make everything downstream worthless.

RLSA, the bid layer most accounts skip

Remarketing lists for search ads doesn't show ads to past visitors. It changes how you bid when a past visitor searches.

Someone who browsed your collection page last week and now searches "waterproof running jacket" is worth more than a stranger typing the same query. RLSA lets you bid up for that person, write different ad copy for them, or open up broader keywords you'd never bid on cold.

It's the least glamorous of the five and the most consistently underused. It's also where the 100-user change bites hardest, which we'll get to next.

Video remarketing on YouTube

Video remarketing targets people who watched your videos, subscribed to your channel, or engaged with your YouTube content. You link your YouTube channel to Google Ads and the audiences build themselves.

For a $2-5M Shopify brand, honesty applies here: if you don't have a real YouTube presence, this isn't your priority. It's genuinely powerful for brands already producing video, and close to irrelevant for brands that aren't. Don't build it because a checklist told you there were five types.

Customer Match, the only durable one

Customer Match is different from the other four in a way that's become important. It doesn't depend on browser tracking at all. You upload hashed customer emails and phone numbers, Google matches them to signed-in accounts, and you reach those people across Search, Shopping, YouTube, and Gmail.

Because it runs on your own first-party data rather than cookies, it's the one remarketing type that doesn't degrade when browsers change their minds about tracking. For a brand with a few thousand past purchasers, it's the most defensible audience you own.

Two rules to know. Google now enforces a 540-day maximum membership duration on Customer Match lists, effective 7 April 2025. Anyone added or refreshed more than 540 days ago drops off automatically, and Google's developer documentation confirms lists with longer durations were updated retroactively.

So a list you uploaded once and forgot about is shrinking every day. Refresh it on a schedule or it quietly dies.

What changed in 2026 (and what most guides still get wrong)

Most articles ranking for this topic were written before three things happened. Each one changes the answer.

The audience minimum is now 100 users, not 1,000

This is the big one. In December 2025, Google unified the minimum audience size to 100 active users in the last 30 days, across every network and every segment type: Search, Display, and YouTube, for both remarketing lists and Customer Match. Search Engine Land reported the change on 24 December 2025. Customer Lists in Search had already dropped from 1,000 to 100 earlier that year, in May 2025.

Previously the thresholds were 1,000 for Search and YouTube, and 100 for Display.

Here's why that reshapes remarketing for a mid-size Shopify brand. At 1,000 users, granular segmentation was theoretical. "Viewed a product, added to cart, didn't purchase, last 7 days" might be 200 people for a brand doing $3M a year.

The segment existed on paper and never served an impression. So you defaulted to one broad list and accepted that your cart abandoners saw the same ad as someone who bounced off your homepage in four seconds.

At 100, that segment works. You can separate cart abandoners from browsers, recent visitors from stale ones, high-value product viewers from clearance shoppers, and bid differently on each. The segmentation that used to require eight-figure traffic volume now works at $2-5M.

Most accounts haven't been rebuilt for this. If yours was structured before December 2025, its audience architecture was designed around a constraint that no longer applies.

Similar audiences are gone, and lookalikes moved

Google sunset similar segments entirely in August 2023 and removed them from all campaigns and ad groups. Several guides still discuss them as a live option. They aren't.

Lookalike segments now exist only inside Demand Gen campaigns, which replaced Discovery campaigns, all remaining Discovery campaigns were auto-upgraded by the end of March 2024. If you want audience expansion off the back of your first-party lists, Demand Gen is where that lives now.

Third-party cookies didn't die, but your reach is still uneven

The cookieless future got called off. Google abandoned third-party cookie deprecation in Chrome in April 2025, then retired ten Privacy Sandbox APIs, including Topics, Protected Audience, and Attribution Reporting, on 17 October 2025. Only CHIPS, FedCM, and Private State Tokens survived.

So cookies stay in Chrome. But Safari, Firefox, and Brave still block them, and always did. The practical consequence for your account is that cookie-based remarketing reach skews heavily toward Chrome users, and always has.

That's not a crisis. It's just a reason to treat Customer Match as the durable asset and everything else as reach that depends on someone else's browser policy.

Remarketing for Shopify brands: what actually breaks

Every guide tells you how to set remarketing up. Almost none tell you how it fails. In accounts we take over, it fails the same two ways.

Broken conversion tracking means broken audiences

Your remarketing audiences and your conversion tracking are built from the same tag. When the tag is wrong, both break; but only one of them is obvious.

An Australian jewellery brand came to us with an account that looked like it was working. Page views, cart additions, and checkout starts were all being counted as conversions, with duplicate purchase trackers firing on top. Smart Bidding was optimizing toward a number that wasn't revenue, and the audience segments built off that same tag were equally scrambled.

The fix was unglamorous: cut to a single primary purchase action, move to data-driven attribution, then restructure by product and region and rebuild the feed. Over the following five months the account did A$108,876 in spend against A$331,092 in tracked revenue, a 3.0x blended ROAS against a 1.3x break-even, above break-even in all five months including one month with a stock shortage.

None of that was a remarketing tactic. It was fixing the input that remarketing depends on.

Membership duration and frequency: stop paying to annoy people

Two settings that cost real money when left on defaults.

Membership duration controls how long someone stays in an audience. The default is often 30 days, which is fine for cart abandoners and wrong for post-purchase segments. If your average repeat purchase cycle is 90 days, a 30-day window means you stop talking to customers right before they're ready to buy again.

Frequency capping controls how often the same person sees your ad. Without it, a small remarketing audience with a healthy budget means the same 400 people see your banner 30 times a week. That's not persistence. That's spending money to make people dislike your brand.

The remarketing ROAS trap

Here's the part no agency likes to write down.

Your remarketing campaigns almost certainly report the best ROAS in your account. That number is also the least trustworthy one you have.

The mechanic is simple. Remarketing only targets people who already came to your site. Many of them were going to buy anyway. Someone who added to cart on Monday, thought about it, and came back Wednesday to purchase would likely have purchased regardless, but if a remarketing ad touched them in between, that ad takes the credit.

So an 8x remarketing ROAS sitting next to a 3x non-brand Shopping ROAS is not telling you that remarketing is 2.6 times better. It's telling you that remarketing is claiming credit for demand that other campaigns created.

Act on that number naively and the logic runs: shift budget from the 3x campaign into the 8x campaign. Which fails immediately, because remarketing is capped by the traffic the rest of your account generates. Double the budget on an audience of 4,000 people and you don't reach more people. You reach the same people more often, frequency climbs, and returns fall off a cliff.

Remarketing is a margin improver, not a growth engine. It makes the traffic you already paid for convert slightly better. It cannot manufacture new demand.

The way out is to stop reading campaign-level ROAS as a performance verdict and start reading blended ROAS against your actual break-even. That's the number that reflects your margin structure, and it's the only one that answers whether the account as a whole is making money. If you haven't worked out what yours is, run your own numbers through our breakeven ROAS calculator. It takes AOV, COGS, shipping, transaction fees, and return rate and gives you the ROAS you actually need to clear.

A pet accessories brand we launched from zero makes the point from the other direction. Nine weeks in, Google was producing 16% of total revenue on 8% of total ad spend. That gap is the signal worth watching: not the ROAS on any single campaign, but what the channel contributes relative to what it consumes.

Which remarketing type should you actually build?

For most Shopify D2C brands in the $2-5M range, the honest answer is two of the five, plus a bid layer. Not all five.

The sequence that works: fix conversion tracking, confirm the feed matches, build dynamic remarketing, upload and schedule Customer Match refreshes, then layer RLSA onto existing Search campaigns. Everything else is optional.

And if your account was structured before December 2025, revisit the segments themselves before you build anything new. There's a good chance the granular audiences you wanted originally are now viable and nobody has gone back to build them.

That kind of rebuild is a large part of what we do. Most brands come to us with a Google Ads account that already exists and isn't working, rather than a blank one. You can see how we run Google Ads for Shopify brands if you want the detail.

The takeaways

Google Ads remarketing in 2026 looks different from the version most guides describe. Five things to leave with:

• There are five types, but for a Shopify brand the real shortlist is dynamic remarketing and Customer Match, with RLSA as a bid layer on top.
• The audience minimum dropped to 100 users in December 2025. Segmentation that was impossible at $2-5M scale now works, and most accounts haven't been rebuilt to use it.
• Dynamic remarketing runs on your Merchant Center feed, not your website. When product IDs don't match, it fails silently and keeps spending.
• Customer Match lists expire after 540 days. Refresh on a schedule or watch your best audience shrink to nothing.
• Remarketing ROAS is over-credited. It's a margin improver, not a growth engine, and it can't scale past the traffic your prospecting generates.

If you're running Google Ads remarketing on a Shopify store and you're not sure whether your audiences are built correctly, whether your feed is matching, or whether that 8x campaign ROAS is real, that's exactly what an audit answers.

Book a Discovery Call and we'll go through your account, tell you what's actually working, and show you what a rebuilt audience structure would look like. No obligation, and you'll leave the call knowing more about your account than when you joined it.