When to Scale a Campaign vs When to Kill It

Most campaigns die too slowly and scale too fast

Media buyers face this decision constantly: is this campaign a winner that needs more budget, or a loser that's draining money? The wrong call in either direction is expensive.

Killing a winner too early means abandoning a campaign that was about to hit its stride. Scaling a loser means pouring money into something that was never going to work. And the middle ground -- keeping a mediocre campaign running at low spend "just in case" -- is often the worst choice of all because it consumes budget and attention without producing meaningful results.

Here's the framework for making this decision with data instead of hope.

The decision matrix: four campaign states

Every campaign falls into one of four buckets based on two variables: statistical confidence and performance relative to target.

1. Clear winner: Scale

Signals:

• CPA or ROAS meets or exceeds target
• At least 50 conversions (enough for statistical reliability)
• Performance has been consistent for 7+ days
• Audience size is large enough to support higher spend

Action: Increase budget by 20-30% and monitor for 5-7 days. If performance holds, increase again. Never more than 30% at a time -- large budget jumps destabilize the algorithm.

2. Clear loser: Kill

Signals:

• CPA is 2x+ target OR ROAS is less than 50% of target
• At least 30 conversions (enough to rule out bad luck)
• Consistent underperformance for 7+ days
• No obvious external explanation (landing page outage, tracking break, seasonal dip)

Action: Pause immediately. Don't reduce budget gradually -- that extends the waste period without changing the outcome. Redirect budget to winning campaigns.

3. Promising but unproven: Test more

Signals:

• CPA is within 20% of target (above or below)
• Fewer than 50 conversions
• Running for less than 14 days
• Performance is trending in the right direction

Action: Keep running at current budget until you hit 50+ conversions or 14 days. Do not scale or kill based on insufficient data. This is the hardest state because it requires patience when the data is ambiguous.

4. Mediocre performer: Optimize or kill

Signals:

• CPA is 30-80% above target
• More than 50 conversions and 14+ days of data
• No clear trend toward improvement
• Performance isn't terrible enough to kill outright

Action: This is where most campaigns get stuck. The answer depends on whether there are specific optimization levers to pull.

The optimization checklist before killing a mediocre campaign

Before killing a campaign that's close but not hitting targets, run through these checks. If three or more optimizations are available, try them before cutting.

Creative exhaustion

Check: Compare the campaign's CPA in week 1 vs. the current week. If CPA has increased 30%+ from the initial period, creative fatigue is likely the issue, not the campaign structure.

Fix: Introduce 3-5 new creative variations. Keep the same targeting and structure. If performance recovers within 7 days, the campaign is viable with regular creative rotation.

Audience saturation

Check: Look at frequency metrics. If weekly frequency exceeds 4-5 for a prospecting campaign, you're running out of new people to reach.

Fix: Broaden the audience. Expand lookalike percentages from 1% to 3-5%. Add new interest layers. Or switch to broad targeting and let the algorithm find new pockets.

Landing page mismatch

Check: Compare CTR (click-through rate) to on-site conversion rate. High CTR with low on-site conversion suggests the ad is attracting the right people but the landing page isn't converting them.

Fix: Test a new landing page or improve the existing one. This is a landing page problem, not a campaign problem. Don't kill the campaign -- fix the page.

Bid strategy misalignment

Check: Is the campaign on the right bid strategy for its maturity? New campaigns on target CPA or target ROAS often struggle because the algorithm needs 30-50 conversions to calibrate. During the learning phase, CPAs can be 2-3x target.

Fix: If the campaign has fewer than 50 conversions, switch to maximize conversions (uncapped) temporarily to accelerate learning. Once you hit 50+ conversions, switch back to target CPA/ROAS.

Attribution window mismatch

Check: Compare 1-day click attribution to 7-day click attribution. If the campaign looks mediocre on 7-day click but good on 1-day click (or vice versa), the attribution window may not match the product's consideration cycle.

Fix: Adjust your evaluation to match the true consideration window. A $500 product with a 14-day consideration cycle will look bad on 1-day click attribution. Evaluate on 7-day click or use server-side attribution with a 14-day window.

The scaling playbook: how to grow winners without destroying them

Rule 1: Scale in 20-30% increments

A campaign producing $10K/day in revenue at 3x ROAS will not produce $30K/day at 3x ROAS if you triple the budget overnight. Meta, Google, and TikTok algorithms optimize within their current delivery pattern. Large budget jumps reset the learning phase and often degrade performance for 3-7 days.

Scale 20-30% at a time. Wait 5-7 days for performance to stabilize. Then scale again if metrics hold.

Rule 2: Monitor marginal performance, not average performance

After scaling from $50K to $65K, don't just look at your overall CPA. Compare:

• CPA before scaling: $45
• CPA after scaling: $52
• Marginal CPA on the incremental $15K: $68

The marginal CPA tells you the efficiency of the additional spend specifically. If the marginal CPA is 50%+ above the pre-scaling CPA, you're approaching the diminishing returns threshold for this campaign.

Rule 3: Scale horizontally before vertically

Instead of pouring more budget into one winning campaign, duplicate the winning elements across new campaigns targeting adjacent audiences.

Winning campaign: Lookalike 1% based on purchasers, video creative Horizontal scale: Create new campaigns targeting Lookalike 1-3%, Lookalike based on high-value purchasers, Interest-based audiences with the same creative

Horizontal scaling adds volume without pushing any single campaign into diminishing returns.

Rule 4: Set a kill line for scaled campaigns

Before scaling, define the performance threshold that triggers a pullback. Example: "If CPA exceeds $60 for 5 consecutive days after scaling, reduce budget to pre-scale level and reassess."

Without a pre-set kill line, you'll rationalize poor performance ("it's still in the learning phase," "it needs more time") and waste the extra budget.

Timing matters: when campaign state is misleading

Don't make decisions on Mondays

Conversion patterns vary by day of week. For most ecommerce brands, Saturday and Sunday have lower conversion rates, which makes Monday reporting look worse than the actual trend. Always evaluate on full 7-day cycles.

Account for algorithm learning periods

New campaigns and significant changes (budget increases above 20%, new creative, new audiences) trigger a learning period of 3-7 days where performance is unstable. Don't judge a campaign during the learning period. Wait for it to exit learning, then evaluate the post-learning performance over a full 7-day cycle.

Watch for seasonal distortion

A campaign launched during a peak period (Black Friday, back-to-school) will show inflated performance that doesn't hold afterward. Conversely, a campaign launched during a slow period may look weak but improve when demand picks up. Evaluate campaigns against seasonal benchmarks, not just absolute targets.

Frequently Asked Questions

How many conversions do I need before making a scale/kill decision?

The minimum is 30 conversions for a kill decision and 50 conversions for a scale decision. At 30 conversions, you can detect whether CPA is significantly above target (2x+) with reasonable confidence. At 50 conversions, you have enough data to trust that current performance represents the campaign's steady state. For expensive products with few conversions (under 5 per week), use micro-conversions (add-to-carts, initiated checkouts) as directional signals, but validate with actual purchase data over a longer period before making final decisions.

What if a campaign is killing it on clicks but not converting?

High CTR with low conversion rate almost always points to a landing page or offer problem, not a campaign problem. The campaign is finding the right people and the ad is compelling enough to click, but something breaks between click and purchase. Check your landing page load time (every additional second of load time reduces conversions by 7%), ensure the landing page matches the ad's messaging, and verify that your checkout flow isn't leaking customers. Don't kill the campaign -- fix the conversion path.

Should I kill a campaign that was performing well but suddenly dropped?

Not immediately. Sudden performance drops usually have identifiable causes: creative fatigue (check frequency), audience saturation (check reach), tracking issues (check pixel firing), or competitive shifts (check auction insights). Diagnose before acting. If the drop aligns with a creative that's been running for 3+ weeks, refresh the creative. If it aligns with a tracking change, fix the tracking. Only kill the campaign if the performance drop persists for 10+ days after addressing all identifiable issues.

---

Go Funnel uses server-side tracking and multi-touch attribution to show you which ads actually drive revenue. Book a call to see your real numbers.