When to Kill Losing Ads on Taboola & Outbrain (2026)

The single most common question in native advertising is also the most expensive one to get wrong: when do you switch off a losing ad or publisher? Taboola and Outbrain both have an answer, and that answer is built to drain your account. They tell you to wait until a placement has spent 10x your target CPA before you cut it. On a $20 target CPA, that's $200 of waste per ad before you're allowed to pull the trigger.

Now multiply that. The best-practice campaign structure is nine ads at launch, and $200 of permitted waste on each is $1,800 burned chasing data you already had on day two. This is Marcel Sattler, founder of native-advertising.net, who since 2015 has deployed over $100M across Taboola, Outbrain, Newsbreak, MGID, Yahoo Native, Mediago, and RevContent for DTC, lead-gen, and affiliate accounts. The platforms get richer when you follow their rule. Here is the rule Marcel runs instead.

Why the Taboola and Outbrain 10x CPA rule burns your budget

Taboola and Outbrain will tell you the same thing: find your target cost per acquisition, then turn off placements and publishers once they hit 10x that number. Set a $20 target CPA and you're told to let an ad spend $200 before cutting it.

The math falls apart the moment you structure a real campaign. Marcel's standard build is nine ads at launch: three headlines crossed with three images, which is the right way to test creative on native. At $200 of allowed spend per ad, that's $1,800 of mandatory burn before the 10x rule lets you clean anything up.

That money doesn't disappear into better data. It goes straight to the traffic platform. The 10x rule "basically just makes the traffic platform way more rich and you way more poor," and that is the entire reason it exists as the default recommendation. Whether you run an e-commerce brand on /solutions/ecommerce, a /solutions/lead-gen funnel, or an affiliate offer, the structure is the same and so is the leak.

Check CTR before CPA on a new account

On a brand-new account, CPA is not the first thing Marcel looks at. Before there's a single conversion, the conversion data is too thin to trust, so the first indicator is click-through rate, CTR on the ads and CTR on the publishers. The same check applies to both.

CTR has to sit in a healthy band, and that band moves with the conditions. Mobile CTR is different from desktop. A campaign in Brazil reads completely differently from one in the US, the UK, or Singapore. There is no single universal number, and anyone who hands you one is guessing. The move is to ask your Taboola or Outbrain rep for the benchmark CTR for your specific niche, device type, and country, then judge against that.

Two failure modes matter at the edges:

• CTR too high is an alarm signal. It usually means you're getting push traffic, the low-quality inventory you want to avoid.
• CTR too low means the creative is boring, not relevant enough to the audience, and won't earn the impressions to convert.

To keep push traffic out of the mix entirely, you need a managed account. A Taboola self-service account will always run with push traffic baked in; a managed account is what filters it. That distinction is also what lets advertisers in countries Taboola doesn't support directly get an approved managed account.

The two-day kill window for native campaigns

The rule of thumb is time-boxed and short. Marcel launches a campaign at $250 per day and checks CTR one day after it goes live. If the CTR picture is clearly bad after two days at the most, the campaign gets turned off or narrowed down hard.

Narrowing down is the part most advertisers skip. You don't just nuke the whole campaign; you turn off the specific ads and the specific publishers dragging it down. That targeted cutting is exactly where the "when do I switch this off?" question gets answered, and the answer is: early, and at the line-item level on both Taboola and Outbrain.

Two days is the ceiling, not the target. CTR moves fast enough that you usually know within 24 hours of launch whether the creative is landing, and a $250 daily budget gives you enough volume to read it without overspending into a bad signal.

Be aggressive early so the algorithm learns the right direction

Early optimization should be aggressive, not cautious. If you're timid about cutting in the first days, the algorithm learns from the wrong data and drives the campaign in the wrong direction. Aggressive pruning is what points the bidding engine at the inventory that actually performs.

The reason aggression is safe is that nothing you turn off is permanent. Kill an ad or a publisher by mistake and you simply turn it back on; it's not a problem. If you cut something and the whole campaign's performance tanks over the next two days, flip it back on and you're solved. The downside of cutting too hard is fully reversible. The downside of cutting too late is $1,800 you'll never see again.

This is why the timid 10x approach loses on both ends. It wastes money waiting for certainty, and it feeds the algorithm a muddy signal full of losing placements while it waits. Aggressive early cutting is faster, cheaper, and reversible.

The real cut threshold: 3x target CPA, not 10x

Here is the number that replaces the platform's 10x rule. Marcel does not wait for 10x CPA. On a $20 target CPA, once a single ad or placement has spent around $60 with nothing to show, it gets turned off.

The logic is simple arithmetic. The gap between the platform's $200 limit and a $60 cut is $140 of spend running in the wrong direction. Even if a sale lands somewhere in that $140, you're already too far over target to catch back up to a $20 CPA. Hitting $60 of dry spend is the signal that something in the funnel is broken and needs a rebuild.

Scale the same ratio up and it holds:

• $20 target CPA → cut around $60 (vs. the platform's $200)
• $80 target CPA → cut around $150 to $180 (vs. the platform's $800)

That $150-180 zone on an $80 CPA is where Marcel feels comfortable: far enough to give the placement a fair read, nowhere near the $800 cliff the platforms want you to ride off. It's a roughly 3x window, applied as a gut-tuned range rather than a rigid formula, and it's the difference between testing on a budget and donating one.

Find the bottleneck before you blame the ad

A dry $60 spend tells you something is wrong; it doesn't tell you what. The strength of native is that the funnel breaks into clear stages, ads, landing pages, and so on, and you can isolate exactly where the leak is. Marcel's favorite part of native is precisely this: you can almost always point to the bottleneck instead of guessing.

When the ads are the bottleneck, CTR diagnoses the failure direction:

• CTR too low → the creative isn't relevant enough. Get pushier and more clickbaity to pull engagement up.
• CTR too high → you've gone too clickbait and are inviting push traffic. Narrow it down.
• CPMs too high → the cost side is the problem, not the engagement side.

The fix follows the diagnosis. It might be a new ad, a different whitelist or block list, a fresh headline, or a different angle entirely. There's no single formula here, but there is a discipline: identify the bottleneck stage, read the metric, and rebuild that one thing rather than letting the whole campaign bleed at $200 an ad while you wonder.

Watch the full breakdown

Is your account a fit for the same play?

If you're running nine-ad campaigns and following the 10x CPA rule, you're handing the platforms $1,800 of avoidable burn per launch, and a managed account plus the right cut thresholds can stop that on the next campaign. The CTR-first, cut-at-3x approach works the same whether you're scaling a DTC brand, a lead-gen funnel, or an affiliate offer.

Book a strategy call and we'll pressure-test your current cut rules and account setup against your target CPA. If you want context first, see how this plays out across our case studies or browse the rest of the native ads resources.