Most shoppers who look at a product on Amazon leave without buying. Some compare a few options, some get distracted, some plan to come back later. Retargeting reaches those shoppers again with an ad, reminding them of the product they already showed interest in. Amazon DSP does this with more control than any other Amazon ad type, on Amazon and across other sites and apps.
This playbook covers how to build retargeting audiences, set lookback windows and frequency, choose creative, and measure results without fooling yourself.
Why retargeting on DSP
Sponsored Display can also retarget product viewers, and for many sellers it is the right starting point. DSP adds finer audience building, frequency caps, more placements off Amazon, and more detailed reporting. It suits brands with enough traffic to build meaningful audiences and the budget or partner to manage it. Sponsored Display vs DSP covers how to choose.
Step 1: build the audiences
Start with the audiences closest to purchase and work outward.
Product viewers who did not buy. Shoppers who viewed your product detail pages in a recent window. This is the core retargeting audience.
Cart abandoners, where available. Shoppers who added to cart and did not complete the order. Usually small and high intent.
Viewers of similar products. Shoppers who viewed competing or related products. Strictly this is not retargeting, but it often sits in the same plan, one step further out.
Past purchasers, for reorders or related products. Useful for consumables and for brands with a range of complementary items. Exclude them from campaigns aimed at new buyers. DSP audiences covers the full list of audience types.
Step 2: set lookback windows
The lookback window is how far back a shopper's view counts. Match it to how long shoppers take to decide.
Fast, low-price categories: short windows, measured in days to a couple of weeks. Shoppers decide quickly; after that, interest fades.
Considered purchases: longer windows. Furniture, electronics and other high-price items can take weeks to decide.
Split windows into separate line items when you can, for example recent viewers and older viewers. Recent viewers are usually worth higher bids. Splitting lets you bid on each separately and see which pays.
Step 3: control frequency
Retargeting without frequency caps can show the same ad to the same shopper many times a day. That wastes impressions and annoys people. Set a cap per shopper per day or per week, start conservative, and raise it only if results support it.
Also exclude recent purchasers of the advertised product. A shopper who bought yesterday does not need to see the ad again today.
Step 4: choose creative
Retargeting creative should remind, not introduce. The shopper already knows the product. Show it clearly, with price, rating and a reason to come back now if you have one, such as a deal or a bundle. Amazon's responsive ecommerce creative builds ads from your listing and is a reasonable default. Creative best practices covers principles that apply across ad types.
Step 5: measure fairly
Retargeting ROAS usually looks excellent. Part of that is real, and part is credit for shoppers who would have come back and bought anyway. Treat attributed ROAS as an upper bound.
Better signals: compare performance across lookback windows (if long windows perform almost as well as short ones, some credit may be borrowed), watch total sales and TACoS alongside DSP results, and run holdout tests where a share of the audience sees no ads. Amazon Marketing Cloud can compare exposed and unexposed shoppers in more detail. The AMC guide explains how.
Illustrative example: Parkway Home runs DSP retargeting on product viewers with a short and a long window. Attributed ROAS is high on both. When Parkway holds out a share of shoppers for a few weeks, the measured lift is meaningful but smaller than the attributed number. Parkway keeps the short window, trims the long one, and sets its budget from the lift rather than the attributed ROAS.
Step 6: budget and timing
Retargeting audiences are limited by your traffic. Spend more than the audience can absorb and frequency rises while results fall. Grow retargeting budget in step with traffic, especially around events, when product views rise and retargeting pools grow for a few weeks afterward. What Amazon DSP costs covers fees and budget planning.
Common mistakes
Retargeting everyone who ever viewed a product, with no window or frequency cap. Including recent buyers in the audience. Judging the program on attributed ROAS alone. Running the same creative for months until shoppers stop noticing it. Raising budget faster than the audience grows, so the same people see more ads instead of more people seeing one.
Each of these is easy to fix once seen. A monthly review of audience size, frequency and lift catches most of them before they cost much. Put that review on the calendar, and treat a falling audience size or a rising frequency as a signal to adjust before results drop. Retargeting rewards steady upkeep more than clever setup.
Frequently asked questions
What is the difference between DSP retargeting and Sponsored Display retargeting?
Both can reach shoppers who viewed your products. Sponsored Display is self-serve from the ad console, priced per click or impression, with simpler audience options. Amazon DSP offers more audience control, frequency settings, inventory and reporting, and is suited to larger budgets and teams that can manage it.
What lookback window should I use for Amazon DSP retargeting?
Match it to how long shoppers take to decide in your category. Fast, low-price purchases suit short windows of days to a couple of weeks. Considered purchases suit longer windows. Many advertisers run separate line items for recent and older viewers so they can bid differently on each.
How do I know if DSP retargeting is working?
Look beyond attributed ROAS, since retargeted shoppers include some who would have bought anyway. Compare results across lookback windows, watch new-to-brand share, and where possible run holdout tests or use Amazon Marketing Cloud to compare exposed and unexposed shoppers.
Off Hours keeps the Sponsored Products side of your account efficient while DSP works on returning shoppers, with schedules, budgets and alerts running on a 15-minute cadence. Start a free 14-day trial.