Facebook Audience Network placement audit

The campaign looks healthy from the top. CPC is down, CTR is up, delivery is green across the board. Then you run Breakdown by Placement in Ads Manager and the picture falls apart. Say your Facebook feed placement sits at a 1.1% link CTR, which is unremarkable. Audience Network rewarded video sits at 38%. Thirty-eight out of every hundred people who saw the ad clicked it. And of the few thousand clicks it collected, exactly zero became a purchase.

(Those figures are an illustrative scenario, but if you have spent on Audience Network, you have probably seen your own version of them.)

I have pulled that breakdown on my own accounts and had the reaction every buyer has: who is even clicking this? The short answer is that many of them did not click on purpose, some clicked for an in-game reward, and intent was never part of the transaction. The longer answer is worth twenty minutes of your month, because placement junk hides inside healthy-looking account averages and quietly spends budget your feed placements would have converted.

Short answer: A Meta placement audit means pulling Breakdown by Placement over 30 days with five columns: spend, link CTR, CPC, conversions, and cost per result. The clearest junk-traffic signal is a placement CTR several times your feed norm paired with zero conversions, usually Audience Network rewarded video. Exclude it, or constrain placements on click-optimized campaigns.

The takeaways

• A placement CTR several times above your own feed norm, paired with zero conversions, is the strongest junk-traffic signal in a Meta account. Real interest converts at some rate. Accidental and incentivized clicks convert at none.
• The audit is one breakdown and five columns: Breakdown by Placement over a 30-day window, then spend, link CTR, CPC, conversions, and cost per result, sorted by spend.
• Advantage+ placements is safest when you optimize for purchases and riskiest when you optimize for clicks or landing page views, because cheap junk inventory looks like success to a click-optimized delivery system.

Why does Audience Network show a huge CTR with zero conversions?

Because a large share of that traffic was never interest to begin with. Audience Network places your ads inside third-party apps, and Meta's own format documentation describes rewarded video as ads people watch in exchange for an in-app reward: extra lives, coins, another round. A person in that flow taps whatever ends the interruption fastest, and sometimes that is your ad. Banner and interstitial placements have a related problem: in a game, the ad sits millimeters from the controls a thumb is mashing. A tap aimed at "jump" lands on your headline.

Meta does filter invalid traffic, and its terms prohibit publishers from encouraging accidental clicks. The structural issue survives the filtering anyway. The click was an accident or a toll, so the post-click behavior is exactly what you see in the breakdown: instant bounce, no scroll, no add-to-cart, nothing. A 38% CTR with zero purchases is not an audience mystery, it is mechanics.

There is a second layer, and it is the one that costs the money: the auction. Meta's delivery system pushes spend toward the cheapest available instances of your optimization event. If the campaign optimizes for link clicks or landing page views, Audience Network is the cheapest place in the system to buy that event. Spend shifts there, your reported CPC improves, your boss is briefly pleased, and your pipeline starves. The dashboard rewards the exact behavior that hurts you.

How do I run a placement audit in Ads Manager, step by step?

Open the campaign view, set a 30-day date range, click Breakdown, and choose "By delivery: Placement". Add five columns: Amount spent, CTR (link click-through rate), CPC, your conversion event, and cost per result. Sort by spend. That is the entire setup; the judgment is in reading it. Here is what I look for, in order:

1. CTR outliers against your own baseline. Your feed CTR is the norm; every other row gets compared to it. A placement at 3 to 10 times feed CTR deserves suspicion. A placement at 30 times feed CTR with no conversions has already answered the question.
2. Use link CTR, never CTR (all). CTR (all) counts reactions, comments, and profile taps. Link CTR isolates the click you are paying to get.
3. Cost concentration. Check what share of spend each placement family takes. A row eating 15% of budget while contributing 1% of conversions is the budget leak, even if its CPC looks like a bargain.
4. The usual suspects by name. Audience Network rewarded video, Audience Network native/banner/interstitial, and to a lesser degree Facebook in-stream video. These are where accidental and incentivized clicks live.
5. Check the conversion column honestly. Use your standard attribution setting and a window long enough for delayed conversions to land. A placement with three days of data has not proven anything in either direction.

One caution before you exclude anything: small numbers lie in both directions. A placement with 40 clicks and no conversions might just be early. The same noise that crowns fake winners in creative tests also produces fake losers in placement breakdowns. My personal rule is to act only when a placement has spent enough that, at the account's average conversion rate, I would have expected several conversions and got none. Then the silence is evidence.

When is Advantage+ placements fine, and when should you constrain it?

Keep Advantage+ placements when the campaign optimizes for purchases or another down-funnel event with real volume. The delivery system is then graded on the thing you care about, junk clicks do not help its score, and in my experience it learns on its own that rewarded inventory does not buy purchases. Constrain placements manually when you optimize for clicks, landing page views, traffic, or reach, because there the cheapest inventory wins by definition.

It is worth being fair to Meta here. Advantage+ placements exists for a defensible reason: Meta's guidance says broader placement eligibility gives the delivery system more auction options and generally lowers average cost per result. That is often true. Excluding placements is a trade, you give up reach and auction flexibility to buy traffic quality, and on a purchase-optimized campaign with strong conversion signal the trade frequently is not worth making.

So the honest decision rule is conditional, and it mirrors how I think about killing an underperforming ad: act on evidence at the level where the damage shows. If the placement breakdown shows Audience Network taking real spend and returning nothing over a fair window, exclude it (Manual placements, untick Audience Network) or, on Advantage+ campaigns where full exclusion is locked, use the available controls like inventory filters and block lists. If the breakdown shows it taking 2% of spend and occasionally converting, leave it alone. You have bigger leaks.

How often should you re-audit placements?

Monthly, plus after any structural change: a new campaign, a changed optimization event, a big budget increase, or a switch between manual and Advantage+ placements. A placement leak is also worth ruling out when performance drops overnight for no obvious reason. The breakdown takes five minutes once the columns are saved as a preset, so the honest constraint is remembering, and a recurring calendar entry solves that. Delivery behavior drifts as budgets scale and as Meta adjusts its own systems, so a placement mix that was clean in March can leak by June.

The deeper habit behind all of this is refusing to let one flattering metric speak for a whole row of the report. A placement-inflated CTR is the same failure mode as a lucky early ROAS: one number, viewed alone, telling a story the rest of the data contradicts. That is the reading discipline the whole ad-intelligence side of Adscalr is built around. It scores creatives on a blend of six metrics (hook rate, CTR, CPI, ROAS, share rate, revenue per install) precisely because a single jumpy number, whatever inflated it, cannot fake the blend.

Run the breakdown this week. The worst case is five minutes spent confirming your spend is clean. The more common case, in my experience, is finding a quiet line item that has been billing you for thumbs aimed at a jump button.