SSubida Skills

Esc to close · results from all notes, skills and people

Marketing›Meta ads

Fixing Meta underspend — CTC's four phases, and why they never turn off ads

Common Thread Collective's 4-phase answer when Meta won't spend at target efficiency — (1) creative expansion with big concept swings, (2) bid-surface expansion by duplicating evergreen campaigns into other bid types, (3) an "underspend" ASC for ads that got <3× target CPA in spend, (4) relaxing the ROAS/cost constraint only when the business justifies it — plus the "never turn off an ad" principle and how others disagree.

Updated 2026-10-24·3 sources·meta-adsunderspenddeliverybiddingcreative-diversityaccount-management

Fixing underspend, and never turning off ads

TL;DR

CTC’s four phases for “Meta won’t spend as much as I want at the efficiency I need” (Tony Chopp, Jul 2026):

  1. Creative expansion. Add new creative to the evergreen acquisition and retention campaigns and active marketing moments. Make big concept swings, not minor variations. This is “almost certainly” the path.1
  2. Bid-surface expansion. Duplicate evergreen campaigns into lowest cost, bid cap or cost cap with the same creative pool. A different bidding profile reaches a different audience, with no extra segmentation.2
  3. Underspend ASC. One value-optimised campaign with a single ad set and the same tROAS. Load it with evergreen ads that spent less than 3× the CPA target. No new creative and no ads that already scaled.3
  4. Constraint relaxation. Loosen the ROAS target or cost cap beyond what the math says only when business performance justifies it. “Be very careful.”4

Plus the never turn off an ad rule. Under CBO with cost controls, Meta has already cut a poor ad’s allocation. Turning it off removes optionality without improving outcomes. Tony calls turning ads off the most common mistake brands make.5

Details

The reasoning. Meta’s documentation says historical performance doesn’t deterministically predict future delivery (the breakdown effect, see The breakdown effect — judge Meta results at the ad-set level, not per ad). Ad-level kills based on yesterday’s ROAS rely on a backward-looking sample the model already outperforms.6

The operator’s three jobs in CTC’s model. One expert operator (“profit engineer”) runs the system day to day:

  1. ensure budget allocation has no limits at campaign level
  2. set bids that match the business economics
  3. keep launching new creative (“feed the machine”)7

Context on phase 3. “The vast majority” of creative put into Meta won’t achieve the results you want. The underspend ASC gives neglected ads one more chance without disturbing the core.3

Disagreements / caveats

  • Turning ads off.
    • Jon Loomer, as a last resort after several batches: turn off one ad that dominates delivery but performs miserably and watch the aggregate. He calls it peace of mind more than a likely fix.8
    • Dara Denney turns off ads that miss CPA with meaningful spend (see Meta creative testing setups at $5K, $50K and $1M+ per month).
    • Ben Heath warns against the opposite reflex of killing proven ads to force spend into new ones.9
  • Giving neglected ads a chance. Ben and Loomer use push delivery on the specific ad, inside the same ad set, instead of a separate campaign (Getting spend onto new Meta ads — push delivery vs the creative testing tool). CTC’s extra campaigns are a structure-heavy alternative.
  • The phases are CTC’s methodology for ecommerce accounts with cost controls. Without cost controls, an unwatched poor ad can keep spending.

See also

Footnotes

  1. How Meta Actually Works in 2026 @ 12:57 — Tony Chopp ↩

  2. How Meta Actually Works in 2026 @ 13:43 — Tony Chopp ↩

  3. How Meta Actually Works in 2026 @ 14:32 — Tony Chopp ↩ ↩2

  4. How Meta Actually Works in 2026 @ 15:22 — Tony Chopp ↩

  5. How Meta Actually Works in 2026 @ 16:13 — Tony Chopp ↩

  6. How Meta Actually Works in 2026 @ 04:56 — Tony Chopp ↩

  7. How Meta Actually Works in 2026 @ 17:03 — Tony Chopp ↩

  8. Your Ad Creation Process Might Be The Problem @ 06:25 — Jon Loomer ↩

  9. NEW Way To Test Facebook Ads! @ 06:44 — Ben Heath ↩