When to Scale or Kill an Ad Campaign: A Clear Framework

Every advertiser eventually stares at a mediocre campaign wondering whether it deserves more budget or a mercy killing, and knowing when to scale or kill an ad campaign comes down to a short set of numbers, not a feeling, a deadline, or how much you personally like the creative.

When to Scale or Kill an Ad Campaign: Let the Numbers Decide

The two decisions feel emotionally similar — both are big calls — but they should be driven by completely different evidence: scaling requires proof of stable, repeatable performance, while killing requires proof of a sustained, structural problem that adjustments have already failed to fix. Neither decision should ever be made from a single day's snapshot, no matter how tempting that is at nine in the morning after a bad night of numbers.

Signals to Scale

• ROAS stable or improving over at least 5-7 consecutive days, not just one good day.
• CPA consistently under your target, with enough volume to trust the average.
• Frequency still low, meaning the audience has real room left before fatigue sets in.
• The campaign has already exited the learning phase with a healthy number of optimization events behind it.

Signals to Kill

• CPA above breakeven for a sustained period after tracking has been verified as accurate.
• Rising frequency paired with falling CTR, the classic fingerprint of audience and creative fatigue together.
• Meaningful spend with zero or near-zero conversions past a threshold you set in advance, not one you invent in the moment.
• An audience that is fully exhausted, with reach no longer growing despite an unchanged or increased budget.

How Long Counts as Sustained

A single bad day, or even a rough three-day stretch around a public holiday, is not sustained underperformance. Look for the pattern to hold across at least two full attribution windows — roughly 14 days on a standard 7-day click setting — before treating either a scale or a kill signal as real rather than ordinary noise.

The Middle Ground: Optimize Before You Kill

Not every underperforming campaign deserves the axe immediately. If tracking, offer-page match, and audience size all check out, try one clean adjustment — new creative, a tighter audience, a bid strategy change — and give it a real data window before reaching for pause. Killing too early throws away learning phase progress that a small fix might have saved. As a simple rule: if this is the first time the campaign has crossed your threshold, adjust it; if it is the third time in a row after a genuine attempted fix, it is probably time to kill it for good.

How to Scale Without Breaking Performance

Raise budgets in increments of roughly 20% rather than doubling overnight, since large jumps reset the delivery system's learning and can spike costs for days. Where possible, scale horizontally first — duplicating a winning ad set into a new audience — before scaling vertically by simply pouring more budget into the same one.

Setting a Rule So You Are Not Deciding Emotionally Every Day

Write your scale and kill thresholds down before you need them: the exact CPA that triggers a pause, the exact ROAS that triggers a budget increase, and the exact number of consecutive bad days required before either decision. A rule set in advance, during a calm moment, beats a decision made mid-panic while staring at one bad morning's numbers.

This is precisely the discipline that is hardest to maintain manually, day after day, across every campaign and platform you run — and it is exactly what AGUDOT was built to automate. It reads your real campaign results every single day and automatically pauses or resumes spend against your daily budget based on rules like these, so the scale-or-kill decision gets made consistently and on time, whether or not you had a chance to log in that morning.