Retargeting Windows: 1-Day vs 7-Day vs 30-Day Performance Compared

A retargeting window defines eligibility after a recorded interaction. It is not the attribution window and it does not guarantee that a person sees an ordered message sequence. Choose windows from the buying cycle, available eligible traffic and cash needed to collect evidence.

Design non-overlapping recency rules

Hypothetical source-side specification, to verify against the actual account’s supported inclusion/exclusion and retention controls:

A new eligible visit can move a person back into the most recent tier. A paid order should trigger the appropriate purchase suppression rule; a later eligible cart can create a new recovery task. Record whether the source’s rolling clock and exclusions actually implement this specification. Even with mutually exclusive tiers, delivery availability and auctions mean message order is not guaranteed.

Size the test from traffic and economics

Illustrative assumptions:200 unique permitted visitors per day, a $140 weekly retargeting cap and $20 paid-order CPA. The budget implies seven expected orders before uncertainty and delay. Do not multiply visitors by thirty as if every visit were a new eligible person: deduplication, purchase exclusions, expiry and permissions change the pool. These are planning inputs, not measured benchmarks.

Recent visitors may include accidental or support visits; older visitors may still be considering a high-value purchase. Combine compatible tiers when separate delivery produces too few mature outcomes, preserving eligibility and a useful message. Do not change to a traffic objective merely to make the pool deliver without checking purchase relevance.

Measure additional orders, not just attributed recovery

1. Define net paid orders and contribution after returns, ads and discounts. Distinguish them from swipes, views and attributed purchases.
2. Where a supported holdout is feasible, assign comparable eligible people or markets to retargeting versus no retargeting. Protect the holdout from other recovery campaigns and record limitations.
3. Hold other email/offers comparable; a person who would buy without the ad is not automatically an incremental order.
4. Approve cash-loss/spend caps and a conversion-maturity rule from the purchase cycle. An insufficient sample ends as inconclusive, not a winning recency tier.
5. If a holdout is infeasible, report descriptive cohort outcomes and do not call attributed ROAS causal lift.

Approve three messages before delivery

Prepare the factual reminder, objection answer and product introduction in NovaStorm as creative briefs and paused drafts. The user verifies stock, landing-page continuity and the actual saved audience rules. NovaStorm preparation does not implement the data-source exclusions or a guaranteed sequential delivery system. Continue with the buyer suppression guide.