In partnership with
Podcast ad intelligence, for a fraction of the cost
Radar uses state-of-the-art transcription technology to transcribe and analyze where brands are advertising across 115,000+ podcasts.
See where your clients’ competitors are advertising on podcasts, read the exact copy for their ads, and verify your own clients’ ads have run.
Radar also allows you to search across millions of podcast transcripts to find relevant shows to buy ads on.
New podcast episodes and ad reads are available minutes after an episode airs.
Radar is built by former Twitter and Tesla engineers, using an AI-native transcription pipeline that delivers high accuracy and extensive data enrichment, at a fraction of competitors’ costs.
At my agency, we haven't added an interest to our campaigns in almost two years now.
Not because it stopped converting overnight.
Because Meta stopped building around it.
Most advertisers still treat their targeting box like a fence.
That's the mistake.
You're not building a wall around your audience anymore.
You're handing Meta a hint, one it can ignore the moment it finds someone better.
The Only Things Meta Still Enforces
Two boxes now.
Know the difference or you'll waste budget in the wrong one.
Audience Controls (hard limits).
Location, language, minimum age (you can push this up to 25), Special Ad Category restrictions, and exclusions.
Meta will not cross these. Set them and move on.
Meta Ads Audience Controls at Ad Set level
Audience Suggestions
Age above the minimum, gender, detailed targeting, custom audiences, lookalikes.
Meta treats every one of these as a suggestion (it’s in the name of the section), not a wall.
Since February 2026, that's official Meta policy, not just how the algorithm quietly behaves.
Meta Ads Audience Suggestions at Ad Set level
Get one thing wrong here and it costs you.
Your past purchasers or existing leads need to live in Controls, as an exclusion, not sit in Suggestions hoping Meta respects them.
I've seen accounts where the exclusion got left as a suggestion.
Meta happily re-served ads to people who already bought, because nothing was stopping it.
Meta's Numbers vs. the Independent Data
Meta reports median CPA improvements of 14.8% top-funnel, 9.7% mid-funnel, 7.2% bottom-funnel, at 99.9% confidence, from 150 A/B tests.
The catch: the top performers behind those numbers were shipping 12 to 19 new creative concepts a week.
An independent study of $834 million in ad spend found the opposite in a lot of accounts:
average ROAS dropped 7% during the Advantage+ rollout, hitting the previously high-performing advertisers hardest.
Where I Actually Draw the Line
Meta says you need 50 conversions a week before Advantage+ has enough to work with.
In my accounts, 25 a week is usually where results start holding steady.
Not exiting the learning phase officially, just enough signal that performance stops swinging wildly week to week.
We don't use lookalike audiences either, not on accounts under about 5K a month.
At that spend level, the seed audience is too small, and the lookalike Meta builds from it ends up too small to give the algorithm anything useful to work with.
What I Still Set (And What I Leave Blank)
Always set: location, minimum age, language, exclusions for past purchasers and existing leads.
Never set, agency-wide: interest targeting, detailed targeting, and lookalikes on accounts under 5K a month in spend.
Set only above 25 to 30 conversions a week: age range or gender, and only with a real reason like a narrow B2B buyer.
The One Truth
Your targeting box didn't disappear.
It just stopped being the thing that decides who sees your ad.
I called this back in DAS #92 - Creative Is the Targeting, Meta Finally Admits It.
This is Meta making it official on the input side too.
If you want the strategy-level segment thinking that still matters, that's DAS #23.
For the signal side of why Andromeda reads conversions the way it does, that's DAS #64.
And for the creative half of this shift, per-persona ads and what changed there, that's DAS #126.
Stop protecting a fence that isn't there anymore.
Feed the algorithm signal and creative, set the few things it still respects, and get out of its way.
TLDR:
Meta made detailed targeting official suggestions in February 2026: only location, minimum age, language, exclusions, and Special Ad Category stay as hard rules.
Meta's own data shows real gains (14.8% / 9.7% / 7.2% CPA improvement by funnel stage), but only with heavy creative output (12 to 19 concepts a week).
An $834M ad-spend study found ROAS dropped 7% during the Advantage+ rollout, hitting previously high-performing advertisers hardest.
Put exclusions like past purchasers in Audience Controls, not Suggestions, or you'll pay to re-acquire people who already bought.
My agency hasn't added an interest to a campaign in almost two years, and we skip lookalikes entirely under 5K a month in spend.
DAS is also sponsored by:
Hundreds of creators. Zero new hires.
partnerUP is the creator team your Shopify brand can't afford to hire. You guide the brief, the AI reaches hundreds of creators, ranks them by fit, handles contracts and payment, and tracks what each drives back to your store. Your first two creators are on us.
♥️ Tools we love to use at my agency:
My ultimate list of FREE Ad Libraries for spying and inspiration
Get my FREE Facebook Ads Hook Library and make it yours.
Email Marketing tools for e-commerce Klaviyo and for leads Mailerlite. We choose the one that fits our client’s needs between these two.
Best Pop-up Tool for e-commerce to grow your email list: Optimonk
Our go to No-Code Drag & Drop website and landing page builder with 170+ prebuilt templates: Sitejet
#1 Newsletter tool for founders, creators, writers is Beehiiv.
Get 30-days Free Trial + 20% OFF for 3 months with this link.
For Project Management we use Nifty.
* If you click the links above, we may receive a tiny commission that helps keep Digital Ad Snack running—thank you!