Meta Ads Retargeting: Window and Suppression Design
How long should my retargeting window be?
Set early, consideration and long-lag boundaries from the observed share of purchases by cohort age in your own data. A window copied from another brand either stops before your buyers decide or keeps paying to reach people who already bought.
For the surrounding account decisions, compare Meta Ads Agency Handover: Access and Control and use Meta Ads Automation: A Safe Workflow Design as the next diagnostic.
In short
This guide owns one decision artifact: the filled, auditable structure below. Reader-supplied thresholds stay explicit; missing evidence stays missing.
Event-recency worksheet
| Observed purchase-lag band | Cohort calculation | Window band | Message sequence | Suppression |
|---|---|---|---|---|
| Same-session / earliest observed band | Purchases whose first eligible event and recognized order fall in band ÷ eligible recognized purchases | Reader sets boundary from its own cumulative lag distribution | Reminder of viewed product/offer continuity; no invented urgency | Recognized purchasers and invalid-consent records |
| Consideration band | Purchases whose observed lag falls after band 1 and inside band 2 ÷ eligible recognized purchases | Boundary chosen where the reader sees a meaningful change in cumulative purchase share | Evidence, comparison, or objection handling supported on destination | Purchasers, expired offers, unavailable items |
| Long-lag band | Purchases beyond band 2 but inside justified maximum ÷ eligible recognized purchases | Maximum follows observed data, purpose, and retention rule | Re-entry message appropriate to stale intent, with current terms | Purchasers, deleted/withdrawn identifiers, stale catalog items |
Cumulative purchase-lag share through day d = (eligible recognized purchases occurring within d days of the declared first event) ÷ (eligible recognized purchases old enough to reach day d). Source: consented event/order join by cohort age. Limitation: unobserved paths and selection remain. A holdout supports a causal read; a matched-period comparison is observational and must state concurrent changes.
Interpretation boundary
Use the purchase-lag recency bands only for its stated decision. Choose early, consideration, and long-lag boundaries from cohort-age purchase shares, then attach current messages and purchaser suppression. A matched-period result remains observational; use a suitable holdout for lift and exclude cohorts too young to reach each lag. Reader-supplied thresholds remain inputs, not universal standards.
Documented thresholds
Meta publishes the figures below; they are reproduced here with the article they come from. Retrieved 2026-09-22 — Meta revises this documentation without notice, so verify anything load-bearing against the current version before you act on it.
- Custom audiences do not backfill. Meta documents that a website custom audience starts populating from the moment you create it, and allows up to 5 rules per audience. A brand adding 300 purchasers a month needs roughly 3.3 months to reach 1,000 — the audience costs $0 to create and cannot be built retroactively.
- Lookalike source size. Meta requires at least 100 people and generally recommends 1,000–5,000. The hard floor is 10% of the bottom of that range and 2% of the top, so an audience can be technically valid and practically useless.
- How small is too small. Meta lists audience size first among the six documented causes of a learning limited ad set, and the operative test is whether the audience can generate about 50 optimisation events a week — at 30 a week you are at 60% of the threshold.
Source: Meta Business Help Centre — https://www.facebook.com/business/help/465262276878947, https://www.facebook.com/business/help/1474662202748341, https://www.facebook.com/business/help/269269737396981.
Why does retargeting cost more than prospecting?
A retargeting pool is small by design, so a higher cost per result there is a question to check, not proof of a fault. It becomes a problem in four cases:
| What you see | Likely cause | What to do |
|---|---|---|
| Frequency climbing past what your margin tolerates | Window too long for the size of the pool | Shorten the window or widen the trigger event |
| Existing customers still in the audience | Purchaser exclusion missing | Add the purchaser suppression from the worksheet above |
| Cost rising while conversions stay flat | Pool exhausted | Cut the budget rather than raising it |
| Retargeting outspending prospecting | Structural imbalance | Rebalance, because retargeting harvests demand that prospecting creates |
The last row does the quiet damage: an account that spends most of its budget on retargeting is paying to reach a pool that something else has to refill.
Can software help?
Bach.ai audits your connected Meta account, estimates the revenue impact of what it finds, and proposes specific fixes. It applies a change only after you approve it. Think of it as an automated audit layer that surfaces issues and proposed fixes for your review — not a replacement for your team’s judgment, and creative production is not its core job, though the Pro and Agency plans can generate a limited number of variants.
FAQ
How do purchase-lag cohorts determine retargeting windows?
Choose early, consideration, and long-lag boundaries from cohort-age purchase shares, then attach current messages and purchaser suppression.
Why can’t matched-period retargeting results establish lift?
A matched-period result remains observational; use a suitable holdout for lift and exclude cohorts too young to reach each lag.
What can recency bands support when attribution is not incrementality?
It supports the bounded operating choice encoded by the purchase-lag recency bands. It cannot replace missing source records or turn platform credit and observed association into incremental impact.
Method and sources
“Set early, consideration and long-lag boundaries from the observed share of purchases by cohort age in your own data.”
Source: Where this guide describes platform behaviour, it follows Meta’s published advertising and Marketing API documentation, which changes without notice — verify anything load-bearing against the current version before you act on it. Every threshold the guide asks you to supply is first-party, drawn from your own account exports and commerce ledger, because no external benchmark can stand in for your own margin structure.