Open the campaigns tab in most store accounts and the remarketing line is the one nobody worries about. It spends a modest amount, reports a return several times higher than anything else, and survives every budget review untouched. Now imagine switching it off in a few comparable regions for four weeks as a test. If orders in those regions barely move, the campaign was not lying about the sales it touched. It was taking credit for sales that were already on their way. Finding out which of those is happening in your account is the most useful thing you can do with remarketing.
Short version: Remarketing shows ads to people who already visited your store, and for ecommerce its strongest form is dynamic remarketing, which pulls the exact products someone viewed from your Merchant Center feed. It reliably produces the best-looking ROAS in the account, and that number is partly an illusion: you are targeting people who were already likely to buy, so the campaign takes credit for revenue that would have arrived anyway. Google also folds remarketing into Performance Max and Demand Gen, and with Display campaigns now moving into Demand Gen, that is where display and dynamic remarketing get built. This guide covers what Google remarketing for ecommerce looks like today, how to structure audience windows and exclusions, how to judge the results honestly, and why remarketing will never scale a store on its own.
Remarketing deserves a clear-eyed guide precisely because it is so easy to sell. Every screenshot of a remarketing campaign looks brilliant. The useful questions are what it actually adds, where it now lives inside Google's automated campaigns, and how much budget it genuinely deserves.
Four mechanisms carry the label, and they behave differently.
Dynamic remarketing
Dynamic remarketing is the ecommerce workhorse: ads assembled from your product feed showing the specific items someone viewed or added to cart. It needs a linked Merchant Center account, tagging that reports product IDs, and a feed in good shape, which is one more reason the product feed repays attention.
Display remarketing, now built in Demand Gen
Standard display remarketing shows static or responsive image ads to past visitors. It suits brand-level messages and stores whose catalogue is too small for dynamic ads to matter. The change this year is where you build it. Google began rolling out a tool in June 2026 to migrate Display campaigns into Demand Gen, and its documentation says new Display campaigns can only be created within Demand Gen, with remaining campaigns migrated automatically and no way back once moved. Remarketing lists and product feeds are both supported there, so display and dynamic remarketing now run as Demand Gen campaigns with the Display Network as one channel, alongside YouTube, Discover and Gmail if you choose them.
RLSA: remarketing lists on Search
RLSA, remarketing lists for search ads, does not create ads at all. It layers your visitor lists onto Search campaigns so you can adjust bids or messaging when a past visitor searches again, which is often the least glamorous and most quietly useful version.
Remarketing inside Performance Max and Demand Gen
This is where honesty is required. In Performance Max your lists are audience signals, starting points the system may expand beyond, not hard targeting, and PMax retargets your visitors as a matter of course. In Demand Gen your lists do target, but optimized targeting, if you switch it on, can reach beyond them, and lookalike segments built from your lists are moving to a suggestion mode by default. The practical consequence: if you run PMax, a large share of classic remarketing is already happening inside it, and a standalone remarketing campaign is often bidding against your own automation for the same people. The standalone campaign still earns its place for controlled messaging, strict frequency control and audiences you want to treat precisely, but its role has narrowed, and anyone pitching remarketing as a big untapped channel in a PMax account is describing an older version of Google Ads.
The moving parts are few but all mandatory. In order:
- Link Merchant Center to Google Ads. Dynamic ads draw titles, images and prices from the feed, so the link comes first.
- Tag product IDs on every key page. Use the Google tag's dynamic remarketing events or the equivalent Google Analytics 4 ecommerce events, so product views, add to carts and purchases report the item ID. On Shopify this rides on the same tagging work covered in our conversion tracking guide.
- Build lists from those events. Product viewers, cart abandoners and purchasers, each with its own window.
- Create the campaign in Demand Gen with the feed attached. Target the lists, choose channels, and use a product scope if some products should not be advertised.
- Supply brand assets. Logo, images and headlines, so the generated formats look like your brand rather than a template.
- Check the IDs match. Compare a sample of tagged item IDs with the IDs in your feed before you judge any results.
Most broken dynamic remarketing traces back to one of two faults: product IDs in the tag that do not match IDs in the feed, or lists too small to serve because the store does not yet have the traffic.
Audience windows and exclusions for ecommerce remarketing
Treating all visitors as one list wastes the entire point, because visitors differ enormously in intent. The diagram maps the usual tiers to the campaigns they belong in.
These are typical starting windows, not rules. Stretch them for considered purchases with long research cycles, and shorten them for impulse categories.
1 to 7 daysCart abandoners, the shortest window and the highest bids
14 to 30 daysProduct viewers, shown the items they looked at
30 to 90 daysAll visitors, low bids or a signal only
ExcludedRecent purchasers, for as long as your reorder cycle
Cart abandoners are the highest-intent group and justify the most aggressive bids, short windows and the most direct message. Reach them within days, not weeks.
Product viewers who never added to cart sit in the middle. Dynamic ads showing what they viewed, over a 14 to 30 day window, is the standard play.
All visitors is the low-intent pool, including bounces. It belongs in long-window, low-cost brand exposure, or as a signal feeding automated campaigns rather than a target for real budget.
Past purchasers are a different job entirely. Exclude them from acquisition-style remarketing and message them for replenishment, cross-sell or a second purchase on a timeline that matches your product. Selling someone the mattress they bought last week is the fastest way to teach customers to ignore you.
Exclusions are half the architecture
Exclusions are the neglected half. Exclude recent purchasers from every campaign chasing the first sale, exclude cart abandoners from broader viewer lists so each person sees only the most relevant tier, and revisit windows so you are not paying to follow someone who last visited months ago. Every dollar spent showing ads to someone who already converted is measurable waste that the campaign's ROAS will happily hide.
List sizes after the 100-user change
In December 2025 Google lowered the minimum audience size to 100 active users for remarketing and customer lists, across Search, Display and YouTube, down from 1,000 on some networks. That makes small stores eligible for tiers they could not use before. Eligible is not the same as useful, though. A list of a few hundred people served daily reaches the same faces over and over, and a handful of conversions on it tells you almost nothing.
Check list sizes before you build tiers
Open Audience manager and note the active size of each list for each network. If your cart abandoner list is in the low hundreds, merge it with product viewers into one dynamic remarketing ad group and split it out later, once traffic grows. Fewer, fuller tiers beat many thin ones.
RLSA, PMax and Demand Gen: where remarketing overlaps
RLSA earns a short section because it is cheap leverage. Layering visitor and purchaser lists onto Search campaigns, in observation mode first, lets you see how past visitors convert differently, then adjust: bid more confidently on generic terms for warm visitors, tailor copy for returning customers, or exclude purchasers from acquisition keywords. None of it requires new creative or new budget.
Standalone remarketing campaign
- Lists are hard targeting
- You set frequency, message and windows
- Clean reporting per tier
- Risk: bidding against PMax for the same people
Remarketing inside PMax
- Lists are signals the system expands beyond
- Visitors are retargeted automatically
- Customer lists can now be excluded
- Risk: remarketing credit inflates PMax results
For PMax and Demand Gen the working rules are simple. Feed them your real lists, especially Customer Match, since the quality of the seed shapes the expansion. Google added first-party list exclusions to Performance Max this year, so an acquisition-focused PMax campaign can now leave a customer list out, which pairs well with the new customer acquisition goal. Do not fund a large standalone remarketing budget alongside a healthy PMax campaign without checking the overlap first. And when you compare channels, remember that part of PMax performance is remarketing wearing a different name, a point our PMax vs Shopping vs Demand Gen comparison covers in more depth.
Frequency and creative rotation
Cap frequency. Remarketing creative tires fast because the same small pool sees it repeatedly, and beyond a handful of daily exposures you are buying annoyance. Rotate creative on a schedule rather than waiting for performance to sag, because by the time the numbers show fatigue the audience has been tired of the ad for weeks. Dynamic ads rotate products on their own, but the frame around them, your logo, headlines and brand images, still needs refreshing from your team every few months.
This is the section most remarketing content leaves out. A remarketing campaign targets people already on their way to buying, then claims credit when they do. Cart abandoners especially: a meaningful share complete the purchase with no ad at all, through a brand search, an email or a direct visit, and the report assigns that revenue to whichever remarketing ad they happened to see. The campaign's reported ROAS is therefore the most flattered number in the account, and judging remarketing by it is grading the campaign on a test it wrote itself.
The calculator below makes that concrete. Enter your monthly remarketing spend, the return the campaign reports and your contribution margin, then move the slider for the share of buyers you think would have ordered anyway. It shows what the campaign actually adds, and the share at which it stops paying for itself.
If the verdict flips from profitable to loss somewhere in the middle of the slider, you do not need a better guess, you need a test. Only a holdout tells you which side of the line your campaign is on.
How to run a simple holdout
Pick a set of comparable regions, switch remarketing off in some of them for four weeks, and compare orders per visitor against the regions where it kept running. Keep everything else in the account unchanged during the test. Where your account is eligible, Google's conversion lift studies do a more rigorous version of the same thing. Then watch blended account performance, total revenue against total spend, when remarketing budget moves up or down; if doubling the budget does not move blended numbers, you are paying for conversions that were coming anyway. Our guide to what a good ROAS actually is covers why campaign-level ROAS misleads in general; remarketing is the sharpest case of it.
Keep view-through out of the headline
Remarketing audiences see a lot of impressions, so view-through conversions pile up quickly. Report click-based conversions as the main number and show view-through separately, or the campaign will look strongest exactly where it adds the least.
Privacy, consent and first-party data
The cookie pools remarketing draws on are shrinking. Consent mode means visitors who decline tracking never join your lists, browser tracking prevention trims them further, and list sizes for the same traffic are noticeably smaller than they were a few years ago. The durable response is first-party data: Customer Match lists built from the emails and phone numbers customers gave you directly. They do not depend on cookies, they persist across devices, and they feed every campaign type from RLSA to PMax signals and Demand Gen lookalikes.
Google is pushing in the same direction. Its October 2026 Demand Gen update added a Data Strength Uplift metric and expanded Data Manager, the tool for connecting customer and conversion data. If an app or developer syncs your customer list automatically, check that the connection was moved to Google's Data Manager API, which Google made the route for programmatic Customer Match uploads this year. Google has also said it will classify conversion-based customer lists by customer type automatically, so review how your lists are labelled in Audience manager rather than assuming they are still set the way you left them. A store that grows its email list is building the remarketing asset that survives whatever happens to cookies next, which is one more argument for treating email capture as paid media infrastructure rather than a nice-to-have.
Budget, and a realistic example
Remarketing is a small slice, typically somewhere around 10 to 20 percent of paid budget, and it cannot be more than that for a structural reason: it only reaches people who already visited. The audience is capped by your traffic. Push more budget at it and you buy frequency, not reach, and returns fall off quickly. Prospecting, through Search, Shopping, PMax and video, fills the pool that remarketing then works; starving prospecting to fund remarketing shrinks next month's remarketing audience.
A worked hypothetical makes the shape clear. A store spending 5,000 dollars a month on ads with 20,000 monthly visitors might hold remarketing to around 750 dollars: perhaps 400 on dynamic ads for cart abandoners and viewers, 200 on Customer Match and past-purchaser campaigns, and 150 on a long-window brand layer. The dynamic campaign will report a spectacular ROAS. The honest question is whether the account's blended return, which you can sanity-check against your break-even with our ROAS calculator, improves when that 750 is spent versus when it is not, and that is the number to manage.
No purchaser exclusions. Paying to chase people who already bought, while the flattered ROAS hides the waste.
One undifferentiated audience. Cart abandoners, viewers and bounces get the same message and the same bid, so the budget over-serves the least valuable group.
Judging it on its own ROAS. The campaign grades its own homework. Blended numbers and holdouts tell the truth.
Duplicating what PMax already does. A standalone remarketing campaign competing with PMax for the same visitors buys the same person twice.
Splitting lists too thin. Lists of a few hundred people are now eligible, but five thin tiers produce noise and heavy frequency. Merge until each tier has enough people to learn from.
Scaling it like a growth channel. The audience is capped by traffic; extra budget becomes extra frequency, not extra customers.
Frequently asked questions
What is dynamic remarketing in Google Ads?
Dynamic remarketing shows past visitors ads built automatically from your product feed, featuring the specific items they viewed or added to cart. It requires a linked Merchant Center account, tagging that reports product IDs, and audience lists built from those events. With Display moving into Demand Gen, it now runs as a Demand Gen campaign with your feed attached. For most ecommerce stores it is the highest-performing remarketing format.
Is remarketing worth it for ecommerce?
Yes, as a small supporting layer with clean exclusions, and no, as a growth strategy. It recovers some abandoned carts and keeps the brand present during consideration, but its audience is limited to your existing traffic and part of its reported revenue would have arrived anyway. Judge it on blended account performance and holdout tests, not its own ROAS.
Do I still need a remarketing campaign if I run Performance Max?
Often not a large one. PMax already retargets your site visitors as part of its normal operation, so a standalone campaign frequently overlaps with it. Standalone remarketing still makes sense for precise control over message, frequency and specific segments such as past purchasers, but check the overlap before funding both generously.
Where do I build display remarketing now that Display is moving into Demand Gen?
In Demand Gen. Google's documentation says new Display campaigns can only be created within Demand Gen, and existing ones will be migrated automatically. Remarketing lists and product feeds are supported there, so you target your lists, attach the feed for dynamic ads, and choose the Display Network as a channel, with YouTube, Discover and Gmail available if you want them.
How much budget should remarketing get?
Typically 10 to 20 percent of paid spend for an established store. The audience is capped by your traffic, so remarketing cannot absorb more budget productively; beyond a point extra spend buys frequency rather than reach. Prospecting fills the pool remarketing works, which is why the larger share always belongs to demand capture and demand creation.
Bringing it together
Remarketing done properly is unspectacular: dynamic ads on a healthy feed, audiences split by intent with purchasers excluded, frequency capped, first-party lists growing underneath, and a budget sized to the audience rather than to the flattering ROAS on the screen. Its job is to finish sales the rest of the account starts, and it should be judged on whether blended numbers improve, not on the campaign report. If you want an outside read on how your remarketing, exclusions and PMax overlap actually look, request a free audit, and our ecommerce Google Ads services page covers how the full channel mix fits together. If you are weighing visual prospecting next, our Demand Gen campaigns guide covers the channel where display remarketing now lives.
About the author
This guide is written by Vasant Chaudhary, founder of Improtics, a Google Ads agency focused on ecommerce. He has more than five years of experience managing over 50 ecommerce and lead generation accounts across the US, UK and India. He focuses on audience strategy, conversion tracking and measurement, the levers that decide whether remarketing adds revenue or just reclaims credit for it. Get in touch or start with a free audit call.
