When to Restructure vs Kill a Campaign

Admaxxer is a DTC analytics platform with built-in Meta + Google ad ops. The hardest weekly decision in paid media isn't "what should I launch?" — it's "what do I stop and what do I fix?" This post gives you a three-bucket framework (**Kill / Restructure / Hold**) with explicit thresholds, plus the most common mistakes we see when brands get it wrong. ## TL;DR - **Kill** if spend has exceeded **3× CPA target** AND delivered **< 3 conversions** after 7 days. - **Restructure** if CPA is reasonable but you've run **6+ weeks** of the same creative/audience and fatigue is evident. - **Hold** if performance is within **20% of target** and you have **< 7 days** of data. - Never kill during a **platform reset window** (algorithm shifts, iOS updates, major holidays). - Never kill your **best-performing audience by mistake** because a single ad inside it underperformed. ## The framework The framework is three buckets, decided in order. Run through it for every campaign in your account weekly. ### 1. Kill A campaign earns the kill decision when two conditions are both true: - Spend has exceeded **3× your target CPA** (e.g., if target CPA = $40, spend > $120). - **Fewer than 3 conversions** in the campaign's entire learning window (typically 7 days). The logic: at 3× target CPA with no statistically meaningful conversions, there's nothing to optimize into. The algorithm has no signal, and continued spend is buying you a more expensive null result. Kill, archive, move the budget. Do not confuse "kill" with "pause and forget." Document why the campaign died (bad creative, wrong audience, mistargeted geography) so the next version doesn't inherit the same problem. ### 2. Restructure A campaign earns restructure when it's delivering reasonable CPAs but shows fatigue signals. The typical profile: - CPA is within **30–60% of target** (not stellar, not terrible). - You've run the **same creative and audience for 6+ weeks**. - Frequency is climbing (typically **> 3.5** on Meta prospecting). - CTR is declining week over week. - CPMs are climbing without a market-level explanation. Restructure means: new creatives, refreshed audience (not just renamed — genuinely different), possibly a new objective or placement mix. The goal is to reset the learning without losing the account-level signal. See our [creative rotation guide](/guides/creative-rotation). ### 3. Hold A campaign earns hold when: - Performance is within **20% of target** on the primary metric (CPA or ROAS). - You have **< 7 days** of data since the last meaningful change. - No obvious fatigue signals. Hold means: don't touch it. The worst thing you can do to a campaign in the hold zone is "optimize" it — every budget change, duplicated ad set, or audience tweak re-triggers learning and costs you signal. ## Common mistakes ### Killing during a platform reset window Meta does algorithm rollouts. Google does attribution model updates. iOS does privacy changes. Every 2–3 weeks, one of the major platforms ships something that reshuffles account performance for 3–7 days. If you kill a campaign during that window, you're killing based on platform noise, not campaign signal. The heuristic: if the whole account moved by 20%+ this week, don't kill anything. Investigate the systemic cause first. ### Killing the best audience by mistake A common pattern: a campaign has one bad ad and three good ones. The campaign's aggregate CPA looks mediocre, so the brand kills the whole campaign. The good ads get archived along with the audience that was actually performing. The fix: always look at **ad-level performance** before killing a campaign. If the average is pulled down by one ad, kill that ad, not the campaign. ### Restructuring too often The opposite failure: restructuring every 2 weeks because a brand confuses "variance" with "fatigue." If you haven't run the same creative for at least 4 weeks at decent spend, you probably don't have enough data to call fatigue. ### Killing based on one attribution model If Meta says 4× ROAS and your blended MER says 2.1, which is right? Usually blended MER — but the point is: don't kill a campaign based on a single platform's reporting. Check new-customer MER, check Shopify's own post-purchase survey data, and cross-reference with [ad-level LTV](/features/ad-level-ltv) before pulling the plug. ## What to do about it 1. **Weekly cadence.** Once a week, run every active campaign through the three-bucket framework. 2. **Document kills.** Keep a running log of killed campaigns with the reason. You'll learn patterns about what doesn't work in your account. 3. **Quarantine fatigue signals.** If multiple campaigns are fatiguing at once, the problem is creative strategy, not individual campaigns. 4. **Don't optimize Hold campaigns.** The best action on a Hold campaign is no action. ## Caveats This framework assumes you have meaningful daily spend (roughly **$100+/day per campaign**). At lower spend, variance dominates and none of the thresholds are reliable. It also assumes you're tracking blended MER — per-platform ROAS alone is not enough data to make these decisions well. Also: "kill" and "restructure" are not always binary. Sometimes the right move is to reduce budget by 50%, hold for a week, and reassess. Don't force a two-state decision when a continuous one fits better. ## FAQs **Q: What if I don't have 7 days of data?** A: Hold. The single most common mistake is killing campaigns at day 3 based on noise. **Q: How do I know if it's fatigue vs. a platform shift?** A: Check if multiple campaigns are degrading at the same time, and if the account-level metrics (total spend, total conversions) match what the platform itself reports for the vertical. **Q: Is 3x CPA the right kill threshold?** A: It's a default, not a rule. For high-AOV products where CPA variance is larger, 2.5x is often better. For low-AOV consumables, 4x is sometimes reasonable. --- **TRIAL_LINE:** Start your 7-day free trial — no credit card required. [See Admaxxer pricing](/pricing).