Google Ads · focused work
Ecommerce remarketing campaigns
Remarketing sells to people who have already met the store: someone who looked at a product, someone who left a full cart, someone who bought once and has not returned. Most of the work is not creative. It is the lists. Who is in them, how long they stay, who is excluded, and what a person is shown at each distance from a purchase.
This is the right work if
- The only remarketing running is one display campaign showing the same banner it showed last year
- Buyers see ads for the item they already purchased because no exclusion audience exists
- Audience lists are defined by URL patterns that no longer match the current site
- Membership windows were left at the default and do not match how long shoppers take to decide
What it is
What remarketing work involves
The foundation is data. Tagging has to record product views, cart additions and purchases with the same product identifiers the shopping feed uses, because dynamic ads render the actual item from that match. Audiences are then defined on those events rather than on URL patterns that break the next time the theme changes, and customer lists uploaded from the store and the email platform are refreshed on a schedule instead of being loaded once and forgotten. Consent choices reduce how many people can be added, so list sizes need watching after any change to the consent banner.
The campaign layer decides who sees what. Membership windows are set from the store's own measured time between first visit and order, and between one order and the next, not from a default. Exclusions stop a buyer being shown the thing they just bought and stop an active email subscriber receiving a paid duplicate of a message they already have. Frequency is capped. Budgets and reporting are kept separate from prospecting, because remarketing is the easiest place in any account to be paid for a sale that was going to happen regardless.
How it is done
The work, in order
What changes
- Lists describe real behaviour and keep working when the site changes
- People stop seeing ads for products they have already bought
- Each audience receives a message written for where it sits in the buying cycle
- Remarketing spend and revenue are reported apart from prospecting
Fix the tag and the product identifiers
Confirm the remarketing tag fires on list view, product view, cart and purchase, and that the product identifier it sends matches the identifier in the shopping feed. Without that match, dynamic ads fall back to generic creative and the whole format loses its point.
Define audiences by behaviour and value
Build separate lists for product viewers, cart abandoners, checkout starters, one-time buyers, repeat buyers and lapsed customers. Where order value varies widely, split the buyer lists by value so the budget can follow the customers worth more than the average.
Set windows from the store's own data
Read the distribution of days between first session and order, and between consecutive orders, then set membership durations to cover the bulk of each. A store selling a considered purchase needs a long window; a consumable needs one tied to how long the product lasts.
Upload and refresh customer lists
Push customer data from the store and the email platform on a schedule, segmented so a lapsed customer, an active subscriber and a first-time buyer can be addressed differently. Stale lists quietly shrink as matches expire, which looks like falling performance.
Exclude, then cap
Remove recent purchasers from acquisition messaging, remove people already receiving the same offer by email, and set a frequency cap. The complaint that ads follow people around is nearly always an exclusion that was never built rather than a targeting decision.
Match message to distance
A product viewer sees the product with the reason to choose it. A cart abandoner sees delivery, returns and payment reassurance. A past customer sees the consumable, the refill or the next item in the range. One creative for all three wastes the segmentation.
Platform notes
Shopify
The Google channel can install the tag, but checkout and thank you pages have their own event handling, so purchase exclusions and the identifier sent on purchase should be verified rather than assumed to match the feed.
WooCommerce
Tagging usually arrives through a plugin or a tag manager container, and caching plugins can serve a page without firing the event, so list growth needs checking against real session counts after any caching change.
How long should someone stay on a remarketing list?
As long as the purchase decision realistically takes, which the store's own data can tell you. A short window on a considered purchase drops people before they decide. A long window on an impulse item keeps paying to reach someone who moved on weeks ago.
Should past customers be excluded from every campaign?
Exclude them from acquisition messaging and from anything offering a first-order incentive they already used. Keep them in campaigns built for them, such as replenishment or a new range, since an existing customer is usually the cheapest revenue in the account.
Does remarketing take credit for sales that would have happened anyway?
Some of it does, and platform reporting will not separate the two. The only reliable answer comes from holding the campaign back for a set of regions or a period and comparing total revenue, which is worth doing once the spend is large enough to matter.
What did consent requirements change about list building?
Fewer visitors are added, so lists grow more slowly and small audiences can fall below the size needed to serve. The practical response is fewer, broader lists rather than many narrow ones, plus customer lists uploaded from first-party data where consent for that use exists.
Part of
Related work and answers
Find the leak.
A free Growth Analysis ranks what your store should fix first, by revenue at stake.