Custom Audience Retargeting: The Two Settings That Decide Whether It Works

What makes a custom audience "custom"

A custom audience is built from an advertiser's own first-party data — a pixel or ad tag tracking who visited a site, viewed a product, added to cart, or purchased — rather than a demographic guess or a modeled lookalike. That's the whole distinction worth remembering: a custom audience is people who already did something specific and traceable, not people who share traits with someone who did. Retargeting is simply the practice of showing ads back to that audience, and it's consistently the cheapest, highest-converting segment in most programmatic accounts, because it's recovering demand that already exists rather than trying to create new demand.

Custom audiences on Amazon specifically can be built from more than one kind of signal — site-visit and add-to-cart events tracked through the Amazon Ad Tag, or ASIN-level remarketing, which targets shoppers who viewed a specific product detail page on Amazon and didn't buy, without needing a pixel on an external site at all. The ASIN-based version is worth knowing about separately, because it works for sellers whose entire funnel lives on Amazon and never touches a brand-owned website with a pixel installed.

The lookback window: how long someone stays eligible

The lookback (or attribution) window sets how far back an action has to have happened for someone to remain in the retargeting audience — commonly 7, 14, 30, or 60 days. This isn't a technical detail to set once and forget; it should match the product's actual purchase cycle. A 7-day window suits impulse purchases, where buying momentum fades fast and a longer window just wastes spend showing ads to someone who's already moved on. A 60-day window suits considered purchases with a longer research cycle, where cutting the audience at 7 days would drop genuinely interested shoppers before they're ready to act. Setting every retargeting audience to the same default window regardless of product category is one of the more common, and more invisible, sources of wasted spend — invisible because the campaign still delivers and still reports some conversions, just fewer than a correctly-windowed audience would.

The frequency cap: how much is too much

A frequency cap limits how many times one person sees the same ad within a set period. Three to five impressions a day is a common general starting point for display, but retargeting specifically tends to need a stricter cap — two to three a day — because showing the same product to someone who already looked at it and didn't buy reads as pressure faster than a cold prospecting impression does. A useful pattern is pairing a daily cap, which prevents a burst of five impressions in one browsing session, with a weekly or flight cap, which prevents the slower grind of steady over-exposure across several days. Neither cap alone catches both failure modes.

The cost of getting this wrong isn't just wasted spend — a frequency cap set too loose can actively work against the brand. A shopper who sees the same retargeting ad a dozen times in a week without buying isn't sitting on the fence; something is stopping them, whether it's price, a competing option, or simple inattention, and no amount of repeated exposure fixes that. In that situation, tightening the cap and letting the creative rotate is a better use of the remaining budget than continuing to bid on the same person at the same frequency.

A worked example: what retargeting actually costs against prospecting

In a 31-day pull across 27 advertisers in our own Amazon DSP book, retargeting line items cost $4.47 per acquired customer, against $12.13 for prospecting — 2.7 times more expensive — on roughly equal spend across the two, $315,666 for prospecting against $306,785 for retargeting. Read purely as a return-on-spend comparison, retargeting won by a wide margin: it returned 7.27x against prospecting's 3.28x. But read as a new-customer question, the picture flips — prospecting delivered a 46.6% new-to-brand rate against retargeting's 29.2%. Retargeting is the cheaper, higher-ROAS channel; prospecting is the one actually growing the customer base. Which comparison is the right one to run a budget on depends entirely on whether the brand's repeat-purchase rate can absorb a $12 new customer profitably — a subscription or consumable category usually can, a one-time hard-goods purchase usually can't.

The mistake: reading the ROAS number and cutting prospecting

The predictable mistake, looking at those two numbers side by side, is shifting budget entirely toward retargeting because its ROAS is more than double prospecting's. That decision starves the pool retargeting depends on — without new visitors and cart-abandoners flowing in from prospecting, the retargeting audience shrinks and its own returns eventually decline too, the same dynamic that shows up whenever a funnel gets judged purely on its bottom-funnel numbers.

Where Dr. DSP fits

Dr. DSP is Amazon DSP run as a managed product by Full Circle, which has managed more than $500M in revenue across 100+ brands. Setting the lookback window to match the actual purchase cycle, and the frequency cap to avoid pressure without losing reach, are small decisions that quietly decide whether a retargeting audience converts or gets ignored. A reader who never buys anything from us should still leave this page knowing that a strong retargeting ROAS number and a healthy customer-acquisition funnel are two different things, and neither one alone tells the whole story.