Freeze Meta spend for 30 days: one ecommerce store, one fixed media ceiling, and no extra budget to rescue unnecessary complexity. Before approving another dollar, make the management team defend the campaigns already spending it.
Eleven campaigns spending $18k a month can look like a sophisticated operation. If four chase the same buyer with the same offer, I want to know what the extra structure buys. “Testing” is not an answer. It is a label on a folder.
This plan gives the team one month to establish a baseline finance can inspect, consolidate genuine overlap in stages, protect distinct creative concepts, and judge the result on business outcomes. The constraint is fixed spend, not a predetermined campaign count. A campaign with a real commercial job stays. A campaign that exists because nobody wants to delete it has to make a better argument.
1. Day zero: fix the ceiling, name the owner, sign the rules
Name one operator and one approver. That is the entire org chart for this plan. The operator runs the account daily and logs every change. The approver, usually the founder or finance lead, owns the ceiling and the authority to stop the work. Nobody else changes structure, budgets, or creative during the freeze.
Cost against the constraint: fewer opinions get a turn at the controls. What it buys: when performance moves, there is a record of who changed what.
Separate the management fee from media spend before starting. A percentage-of-spend fee pays the manager more when you spend more, even if efficiency falls. I used to tell clients that aligned incentives. I was wrong.
Write the terms plainly:
- Media ceiling: say, $15k across the 30 days.
- Management cost: flat and recorded separately.
- Business target: contribution margin from Meta-attributed orders, with refunds accounted for.
- Supporting revenue target: for example, at least $42k net collected revenue after refunds, tracked through Shopify alongside a fixed 7-day click view.
- Decision owner: the named approver, not whichever person replies first in Slack.
Revenue and margin are not interchangeable. The revenue target makes the plan readable; contribution margin tells you whether the orders are worth buying. The test is whether simpler management produces better business outcomes from the same $15k, not whether the dashboard displays a prettier ROAS.
Refunds belong in the starting rules, not the explanation for a disappointing finish. Meta’s purchase-time reporting does not net out subsequent returns, so reported ROAS can overstate the revenue you keep. Returns-adjusted ROAS is attributed revenue minus refunds on those orders, divided by spend. Freeze the definition alongside the target.
Put the proposed merge, protected campaign boundaries, and rollback rule in a one-page memo now. My working loss limit is a drop of more than 15% in net collected revenue across seven consecutive days after the merge, against the agreed baseline, with the offer and ceiling unchanged. The approver signs before any structural change.
What gets cut: promises that consolidation alone will lift ROAS. Fewer campaigns do not create demand. They can reduce fragmentation and make creative and offer performance easier to read. That is the proposition being tested.
2. Days 1–5: build a baseline finance can inspect
Use the last 28 days to reconcile three records: Shopify revenue, Meta-attributed sales, and actual ad spend from the bill. Reconcile does not mean force them to match. It means explain what each number represents and where the gaps sit.
I want one sheet the finance lead can open without a narrated tour. Include:
- Date and actual Meta spend.
- Meta-reported revenue under the agreed attribution view.
- Shopify collected revenue.
- Refunds, with the orders they relate to identified.
- Net collected revenue after refunds.
- Product and shipping costs used in the contribution calculation.
If Meta claims $58k on $15k spend while Shopify shows $51k collected, do not paste the platform ratio into the approval memo and call the job finished. Explain the gap. The figures describe different views of the business; the sheet needs to make those differences visible.
Cost against the constraint: five days without structural changes. What it buys: a baseline that does not change its definition when the result becomes inconvenient.
Account for margin and lag before judging anything. Subtract product cost, shipping, and refunds from attributed order revenue to calculate contribution. Do not deduct refunds again if the starting revenue figure already excludes them. Keep media and management costs visible separately.
Then record conversion lag. If half the purchases happen three to seven days after the click, day-three ROAS is an unfinished number, not a verdict. The same patience applies to refunds that have not yet arrived.
Log live confounders in the same sheet: promotions, discount codes, stockouts, price changes, and email drops. These do not make the exercise useless. They make an unexplained before-and-after claim useless.
What gets cut: platform ROAS as the final decision metric. Keep it as a diagnostic. Finance signs the baseline before the operator touches structure; if it is not ready, the account stays frozen rather than rushing into a merge.
3. Days 6–10: make every campaign justify its existence
List every live campaign with its objective, audience, offer, spend share, and reason for being separate. “We have always had it” does not fit in the last column.
Overlap often becomes obvious on one page: three prospecting campaigns using broad targeting and the same bestseller creative, followed by a retargeting campaign repeating the same discount. Extra ad sets divide the available learning volume; overlapping delivery can leave some barely spending. That is the useful mechanism to examine, not the shorthand that you are paying several times to enter your own auction. The operating principle is to consolidate by default and split for a real reason.
Protect those real reasons in writing:
- Countries with different shipping costs.
- Product groups with different margin structures.
- Offers with genuinely different economics.
- Delivery, measurement, or budget requirements that need a separate boundary.
Prospecting with a first-order offer and retargeting with a bundle upsell can deserve separate treatment. Two campaigns carrying the same message do not earn that distinction just because one has “warm” in its name. The structure I favor is one main sales campaign, with separate retargeting only when messaging or economics differ, plus a testing lane where the budget supports it.
Select one consolidation step, not a full rebuild. For a smaller account, two or three live campaigns may be enough under the frozen ceiling. That is a working shape, not a quota. Do not merge a legitimate commercial boundary to make the account diagram look tidy.
Cost against the constraint: every separate campaign now needs a written defense. What it buys: a specific change the operator can evaluate rather than a pile of simultaneous edits.
What gets cut: duplicate prospecting structures and redundant lookalike stacks. What stays: boundaries that finance, fulfillment, or the offer actually requires.
4. Days 11–20: merge the overlap, not the creative ideas
Consolidation cuts containers, not reasons to buy. For a smaller account, start with four to six genuinely distinct ads in the surviving prospecting pool and explain the job each performs.
Price, giftability, the problem solved, proof, and an objection handled are different arguments. Thirty slightly different openings on the same argument are not thirty concepts. As this creative-diversity argument puts it, micro-variants can still be a portfolio of one.
That distinction matters because an account can become structurally simpler and creatively narrower at the same time. I want the first result, not the second. Protect distinct creative concepts while removing redundant campaign structures.
Keep the offer and spend ceiling stable while the merged structure runs. Same bestseller, same discount, same $500-a-day cap within the $15k ceiling. Fewer buckets splitting it.
Log every change with the date, what moved, and why. An entry should be concrete enough to reconstruct:
- Merged prospecting B into A on day 12.
- Paused the redundant lookalike stack.
- Moved $120 a day into the surviving campaign without raising total spend.
Then do the difficult part: leave the structure alone long enough to read it. No new audience because Tuesday looked soft. No budget spike because Wednesday looked brilliant. No mid-flight rebuild disguised as “being responsive.”
Allow a full week of collected orders before judging, and respect the longer conversion lag if the baseline shows one. The calendar does not make incomplete data complete.
Cost against the constraint: patience, plus the temptation to take credit for constant activity. What it buys: an interpretable read on the staged change.
What gets cut: micro-variants used to justify more campaign containers, and daily structural tinkering. Keep the distinct buying arguments in play. That is where the useful variety belongs.
5. Days 21–30: keep, reverse, or keep observing
Compare net collected revenue after refunds and contribution margin with the agreed baseline under the same media ceiling. Use the definitions signed on day zero. Do not switch attribution views halfway through because another one looks kinder.
Only then use Meta diagnostics to explain the movement. My checks are straightforward:
- Compare 1-day click with 7-day click reporting.
- Compare 7-day click with 7-day click plus 1-day view.
- Separate new and existing customers.
- Flag placements where view credit carries much of the reported result.
These are diagnostic checks before a lift test, not proof of incremental sales. They help show what the platform is claiming.
Cost against the constraint: accepting that a cleaner account with flat platform ROAS and less refund drag can still be a better business result. What it buys: a decision based on money retained, not merely revenue reported.
A before-and-after comparison does not prove consolidation caused the change. A sale ran. Email did some work. Stock changed. Meta took credit. The log exists so these explanations stay visible.
A proper incrementality test uses a holdout comparison; a better month alone is not causal proof. This playbook does not bolt that separate experiment onto a small account’s 30-day freeze. It produces an operating decision with an explicit limit on what the evidence can establish.
Apply the signed loss rule rather than inventing one after the result arrives. If net collected revenue falls more than 15% across seven consecutive days after the merge against the agreed baseline, with the offer and ceiling unchanged, roll back to the prior structure. Keep the spend ceiling fixed during rollback.
If volume is thin or conversion lag runs beyond seven days, extend observation by two weeks under the same frozen ceiling. An inconclusive result is not permission to buy a clearer-looking chart.
What gets cut: rollback theater on day two and a victory lap on day four. Neither is account management. They are reactions with a login.
6. Close the month with a decision memo, not a victory deck
Return to the one-page memo signed before the work began. Now add what actually happened:
- The change: which campaigns merged, when, and where the budget moved.
- The protections: which countries, margin groups, offers, and creative concepts stayed distinct.
- The outcome: net collected revenue, contribution margin, and the diagnostics that help explain them.
- The interruptions: promotions, stockouts, lag, refunds, or other logged confounders.
- The decision: keep, reverse, or extend observation under the freeze.
Cost against the constraint: no room to hide behind a polished reporting deck. What it buys: an auditable answer to whether the simpler account deserves to keep running.
A better month earns the simpler structure the right to continue. It does not automatically earn more spend. Those are separate approvals, and the second needs its own target.
At the end of the scheduled freeze, make one of three calls:
- Keep: retain the simpler structure and move the hours previously spent maintaining redundant campaigns into creative testing.
- Reverse: restore the prior structure if the loss limit triggered and rollback produced the better result.
- Extend: keep the ceiling frozen for two more weeks if lag or thin volume leaves the result inconclusive. That can happen with low-volume stores or a plan built around sellable stock during a promotion gap.
If nobody can own the log, reconciliation, and daily restraint, hire that work properly. Another campaign is not a substitute for an accountable operator.
When the constraint lifts, do not add budget and complexity in the same week. If you approve more spend, put it behind the winner rather than rebuilding the clutter you just removed. Keep the distinct creative arguments. Make every new campaign defend a job the existing structure cannot do.
The question was never whether Meta prefers fewer campaigns. It is whether your team can defend the campaigns it runs with net numbers. After 30 days frozen, that answer should fit on one page.

