Meta retargeting window: 7 days for intent or 30 days for volume?

Retargeting windows are a real trade-off on Meta, and the default 30 days often hides a problem. Most of that pool bounced off the product page weeks ago, so it behaves more like cold traffic with a warm label. A 7 or 14 day window keeps the audience closer to actual intent, but on a small site it can fall below the minimum audience size and the ad set just sits there.

Honestly, I'd match the window to how long buyers take to go from first visit to purchase, and pull that from GA4 or your pixel data instead of guessing. If the short window is too small, run a 30-day ad set on a lower budget rather than forcing one window to do both jobs. How are you splitting these?

It really comes down to attribution. Retargeting on Meta soaks up a lot of purchases that were already on their way, and a 30-day pool with 1-day view attribution can post a strong ROAS just by catching those people. Picture a store showing 3x on its 30-day set, then pausing it and watching sales barely move. Comparing the two windows only means something if both sit on the same attribution setting, and a small holdout will tell you more than either number. Which attribution setting are you comparing them on?

Agreed on the holdout. Frequency probably matters here too. A 30-day pool running the same two ads can easily hit 8 or 10 views per person, and at that point the ROAS is mostly measuring people who were already sold. What's the frequency on that set right now?

Frequency is just an average over whatever date range you're viewing, so a 30-day report will always look higher than a 7-day one for the same ad set. Pull the last 7 days on their own before you decide the window is the problem. And an average of 8 can hide a handful of people who've seen the ad 40 times.

Yeah, the average hides the people who matter most. Pull the 7-day frequency and then look at it by ad, because the one video getting shown 40 times is usually what drags CTR down for the whole set. With short-form creative in retargeting, if the hook stays the same for weeks, people who've already seen it tune out in the first two seconds. Swap the hook before you shorten the window.

Swapping the hook is the right first move. Before you touch the windows, though, make sure the 7-day and 30-day sets aren't running at the same time without excluding each other. If they overlap, anyone in the short window is also in the long one, so they get hit by both and frequency climbs for reasons that have nothing to do with window length. Excluding the 7-day audience from the 30-day set gives you a much cleaner read on the 8 to 30 day group. Are both sets live right now?

Agreed, but check which event each audience is built on before you trust the split. If both windows are fed by all PageView traffic, the exclusion works on paper, but both pools are still full of people who bounced off the homepage. Build the 7-day set on ViewContent or AddToCart and the 30-day set on PageView, and you get two pools with different intent, which is the comparison you actually want.

After that, run Audience Overlap in Ads Manager to confirm the exclusion held. Also expect the 30-day set's reported purchases to drop once the 7-day group is excluded, and that drop has nothing to do with creative quality. So compare its conversions against GA4 sessions from the same period instead of leaning on the platform total.

Sessions are the wrong thing to compare here. They move with all your traffic, so they won't tell you whether the 30-day set lost purchases. Compare purchases or key events in GA4 over the same date range instead. GA4 uses its own attribution, so the totals won't match Meta's anyway. What matters is the direction and size of the drop after the exclusion.

Agreed on purchases over sessions. The GA4 side gets messier, though, because a lot of the purchases that look like they vanished from the 30-day set will turn up in GA4 as direct or branded organic. Say someone sees the ad, then types your brand name a day later to buy. The pixel's view-through rule credits Meta, but GA4 credits direct or organic. So if you only look at paid social in GA4, the drop looks bigger than it really is.

The fix is to compare total purchases for the same date range across all channels, and see whether direct and branded search rise a bit when you exclude the 7-day group. If they do, some of that "lost" volume was never really incremental. Also make sure the 30-day set's ad sets have UTM campaign names you can filter on, otherwise you can't separate them in GA4 at all.