Meta Now Auto-Excludes 2 Years of Buyers From New Customer Ad Sets
Meta now automatically excludes anyone who fired a purchase event in your ad account when an ad set uses the "Acquire new customers" option in Customer Lifecycle Strategy. The default lookback is 730 days, and you cannot switch the exclusion off. So that ad set now skips two years of buyers, including people who bought a different product. Check your purchase events and pick a shorter window.
Jon Loomer spotted the change and published the details today. In Ads Manager, the exclusion section now shows a line that reads "Automatically created: Customers from your conversion data (730 days)." When you expand it, you can choose 730, 180, 90, 60 or 30 days. That is the whole menu. Four of the options sit inside six months and the fifth jumps to two years, which is an odd spread. In practice, most accounts will probably leave the default alone.
On paper this is the feature many media buyers have wanted for years: a "new customers only" switch that works without hand-built exclusion lists. And in some accounts it will do exactly that. The catch is the definition of "customer." Meta now takes it straight from your purchase event, and that event is usually the least audited thing in the account.
How the setting worked before this update
When Customer Lifecycle Strategy showed up this spring, Loomer described it as a 2026 version of the old existing-customer budget cap. The exclusions came from the Existing Customers audience segment you defined in Advertising Settings. His warning back then was that "the accuracy of those exclusions will be dependent on how you define your audience segments." Many advertisers never defined that segment at all, so for them "Acquire new customers" mostly changed the label.
There was also some confusion about what "new" meant. EasyInsights reported in April that Meta treated a new customer as someone who hadn't bought in the last 180 days. Today's default is 730 days. If your nCAC dashboards or agency reports assumed a 180-day definition, the baseline just moved underneath them, and nobody sent a notice.
This also follows a year of Meta reshuffling exclusion controls. According to Common Thread Collective, Meta deprecated detailed targeting exclusions earlier in 2026 and launched exclusion-only custom audiences on August 10. Before that, the existing-customer percentage cap in Advantage+ Shopping went away too, and my read is that it cost brands visibility into how much "prospecting" spend was really going to people who had already bought. The pattern seems consistent: fewer manual levers, more decisions made by Meta from your own data.
Where the no-opt-out rule hurts
The no-opt-out part is the bit I would worry about.
Take a brand selling a $40 skincare starter kit alongside a $25/month membership. Under the new rule, an "Acquire new customers" ad set promoting the membership automatically skips everyone who bought the starter kit in the last two years. For many brands, those buyers are the warmest membership prospects they have. Loomer flagged the same problem with a membership example. His workaround is to switch to "Get conversions from all audiences" and add your exclusions manually under Legal and Operational Exclusions Custom Audiences. He doesn't think that is how Meta meant the feature to be used, and he would like the automatic exclusion to be optional. I agree with him.
The second problem is quieter. A two-year exclusion shrinks the pool for categories where people rebuy. Supplements, coffee, pet food, and anything else with a 30 to 90-day reorder cycle all fall in that group. A customer who bought once 20 months ago and lapsed is, for practical purposes, a new customer again. Meta now treats them as off-limits by default. Excluding more people tends to raise the cost of reaching the people left, so reported new-customer CPA is likely to go up even if nothing else about the account changes.
That matters because new-customer costs were already rising. Wicked Reports analyzed 55,661 Meta campaigns and found that the average new-customer acquisition cost for Advantage+ rose from $257 to $528 between May 2024 and May 2025. If your nCAC climbs another 15% to 20% this month, check the exclusion window before you blame the creative.
Your purchase event now decides who counts as a customer
This is the part that probably deserves an hour of your week. Meta builds the exclusion from people who fired the purchase event. That means every flaw in your purchase tracking now affects targeting as well as reporting.
Loomer's earlier post on exclusion gaps lists the usual problems. Purchase events fire at the wrong stage, such as on a lead form or an add-to-cart. Offline and phone sales never reach Meta. iOS opt-outs shrink the matched audience. In the same piece he puts customer list match rates anywhere from 20% to 70%, which tells you how much of a "customer list" Meta can actually find.
Mis-fires cut both ways, and that is a little uncomfortable. If your purchase event fires on something that isn't a purchase, Meta excludes people who never bought. You get a smaller audience and a higher CPA, and the people the ad set should reach never see it. If the event misses offline sales, real customers stay in the "new" pool and inflate your new-customer numbers. Both mistakes are invisible in the ad set itself.
Anyway, to be fair to Meta: Loomer's own conclusion is that if the exclusions look wrong, the cause is more likely your data than Meta's system. From what I've seen in how accounts get set up, I'd bet he's right more often than not.
A 30-minute audit before your next acquisition ad set
Here's roughly the order I'd go in:
- Check the purchase event in Events Manager. Confirm it fires once per real order, on the order confirmation page or via your Conversions API purchase call, and nowhere else. If the volume in Events Manager runs more than about 10% above your actual order count for the same week, something is double-firing.
- Change the window to match your repurchase cycle. For replenishment products, 90 or 180 days is probably closer to the truth than 730. For high-ticket, rarely repeated purchases like furniture or software annual plans, the two-year default is fine.
- Use "all audiences" plus manual exclusions for cross-sell campaigns. If the campaign sells product B to product A buyers, "Acquire new customers" works against you. Switch the setting and exclude only past buyers of product B.
- Upload a customer list if offline sales matter. Loomer suggests sending conversion data from more than one source when you sell online and offline, and adding a customer list custom audience when not every purchase is sent as an event.
- Log today's nCAC as a baseline. Note today's new-customer CPA for each "Acquire new customers" ad set. If it jumps more than 20% within two weeks with no creative or budget changes, the exclusion window is your prime suspect.
One more thing, and it's a bit off-topic: Loomer thinks most advertisers don't need to exclude customers at all. In his view, many advertisers saw algorithmic remarketing inflate their prospecting results and responded with blanket exclusions. He recommends defining audience segments and using breakdowns to see what each group actually contributes. I find that argument more convincing every time Meta takes away another manual control. If you can measure the mix, you don't have to fight it.
Meta's definition of "new" is now partly your job
My prediction: within six months, Meta adds a custom lookback (Loomer has already asked for any value from 1 to 730 days) or makes the automatic exclusion optional, because cross-sell brands will complain loudly enough. Until then, the 730-day default will quietly shape the reported new-customer CPA of every account that turns on "Acquire new customers" without opening the dropdown.
Honestly, I don't think the feature is bad. It's probably the most useful version of Customer Lifecycle Strategy so far. It just assumes your purchase event is clean and your customers buy once every two years, and for a lot of brands neither is true. Open the dropdown before you launch.
If you're also rethinking how Meta's AI handles your product data, our breakdown of how Meta's Muse picks products from just 2 websites covers a related problem with the catalog side.
By Notice Me Senpai Editorial
