You have one owner with about three hours a week, 30 days, the same ad account and the same budget to switch Meta agencies. Do not let the new manager spend week one rebuilding campaigns. That throws away delivery history you already paid for and starts learning again on the budget you were trying to protect.

I used to be the guy who rebuilt everything in week one. New campaign names, clean structure, my system. I told clients it was hygiene. I was wrong. Too often, what I called hygiene was making the account relearn who buys from you so I could work in a structure I preferred. The order I would insist on now is fixed: tracking first, creative second, structure last, if at all. Give the new manager that order in writing before they touch the account.

1. Before day 1: secure the account and its history

Cost: one hour of the owner’s time. Buys: the ability to make the switch without surrendering the account history. Start with access, not a kickoff presentation. Confirm the account ID the new manager will use and who controls access to it.

If your ad account was created inside the old agency’s Meta business portfolio, it cannot be moved out. It stays in that portfolio, so the handover needs to preserve access to that same account rather than open a fresh one (how access works when the account stays in the original portfolio). Confirm the access arrangement before you remove anyone. The same-account requirement is not a preference for familiar campaign names; it is the condition that keeps the history you have already paid to build.

Treat the pixel and dataset with the same care. Meta does not let you transfer a dataset between portfolios. It can be shared with you, or you can start a new one from zero (why dataset history matters for audiences and learning). Ask for the share, the appropriate Business Manager access and the raw creative files. If the old agency controls any of those pieces, settle the handover before the new manager needs them.

Then export your instant-form leads. Meta keeps them downloadable for 90 days from submission, so pull what is available before handover and again on switch day. Connect your own CRM before the old agency’s connection is cut (export leads and connect your CRM before handover). This is not glamorous work. Neither is discovering that the leads stopped arriving while two agencies discussed whose integration was responsible.

No access, no switch. I go deeper on that handover in how to hire a Meta ads manager without losing your account or data.

2. Week 1: leave delivery alone and fix measurement

Cost: 90 minutes of your three hours. Buys: numbers you can use to judge the next three weeks. I have watched new managers spend week one renaming campaigns while the pixel fired twice on every purchase. Do the opposite. Keep the producing campaign running, pause only obvious non-producers and make changes one at a time. A pause of seven days or longer counts as a significant edit that can send an ad set back into learning (how a seven-day pause affects learning). Delivery stays on. Measurement gets fixed.

The manager’s job this week is to show how Meta’s reporting relates to your store or CRM records. The tracking setup is Pixel as the browser layer and Conversions API as the server layer, using the same event_id for deduplication and checking the result in Events Manager with Test Events (Pixel, Conversions API and matching event IDs). If that setup is already working, do not rebuild it to look busy. Verify it.

Next, reconcile Meta-reported purchases or leads against Shopify or CRM records for the last 30 days. I ask for three checks on one page:

  • Meta-reported purchases or leads beside the store or CRM count.
  • Meta-reported cost per result beside cost per result calculated from your records.
  • Event match quality for the purchase or lead event.

Those figures do not need to be identical. They do need a difference you can explain before you call a new ad a winner. Record what changed in tracking and when; otherwise, a cleaner week-four report might just reflect a new way of counting. This is where the owner spends the 90 minutes: review the reconciliation, ask about unexplained gaps and sign off before testing begins. No tracking sign-off, no new ads.

3. Week 2: test creative inside the existing winners

Cost: two fresh image or video variants and one hour of owner approval. Buys: a creative test without a new campaign structure. Meta exits learning after about 50 optimization events per ad set in the seven days after its last significant edit. That count belongs to the ad set, not the campaign or the account. If an ad set is already producing purchases or qualified leads, use it rather than spreading the same budget across a set of new ones.

Week two has a narrow brief: no new campaigns and no new targeting. The manager adds the two variants to the one or two ad sets responsible for most results in the previous 30 days, then runs old and new ads side by side. Keep the old creative in place as a control. Otherwise, the manager can tell you the new ads got results but not whether replacing the old ones helped.

Adding an ad can itself count as a significant edit. So can changes to targeting, creative, the optimization event or bid strategy; a seven-day pause and a large budget jump also matter. The working budget rule here is to avoid moves over 20% in a day and scale around 20% every three to four days (significant edits and the 20% budget rule). The point is not that a creative test comes free. It is that you make one controlled change while leaving the account’s audience and budget structure in place, rather than changing five things and calling the result a test.

Have the manager note which ad set received each variant, when it went live and what the old ad was doing beside it. That small record makes week-four judgment possible. If they want a rebuild before they have tested creative, they are out of order.

4. Week 3: consolidate a starving ad set, if one exists

Cost: one decision and 30 minutes to approve it. Buys: more volume for an ad set that has a chance to learn. Too many ad sets, too many ads and too little budget per ad set can leave an account stuck in Learning Limited. Consolidation is the available fix: merge overlapping audiences, use broad targeting or CBO where appropriate, limit ads per ad set and cut low-spend ad sets (how fragmentation contributes to Learning Limited). It is not permission to redesign healthy campaigns.

Say you spend $3,000 a month and divide it among six ad sets. Asking each one to collect roughly 50 conversions in seven days is a hard ask. The manager should identify the one ad set that is short of volume and show its numbers before proposing a merge. For example: it spent $400 for two leads in 14 days; it is not getting enough results to learn; merge it into the week-two winner. That is a decision an owner can inspect in 30 minutes.

Week three allows one structural move, with receipts. If no ad set is plainly starving, make no structural move. The absence of a merge is not a failure to follow the playbook; merging a healthy ad set to tick a box would be. Keep budget moves small for the same reason you kept week two narrow. Consolidation is surgery on the starving ad set, not a rebuild of everything around it.

5. Week 4: judge results against your own break-even

Cost: one hour to pull store or CRM results against Meta spend. Buys: a day-30 decision grounded in your economics, not a report design. Break-even ROAS is 1 divided by gross margin. At a 40% gross margin, that gives you 2.5x as the gross-margin break-even point (the break-even ROAS calculation). Use your own break-even measure for the result you actually buy, whether that is a purchase or a booked job. Last month’s agency report is not the threshold.

I have seen cost per lead rise 15% while the close rate doubled because tracking finally counted real jobs. I have also seen ROAS fall after a rebuild that looked cleaner in screenshots. Those are reasons to put Meta cost per purchase or qualified lead next to the result in Shopify or your CRM, using the same 30-day window you established in week one. Ask what tracking changed before treating a reporting change as a performance change.

Say your break-even is $85 per booked job on $3,000 spend. If week four comes in at $71 with the same account and budget, cost per booked job clears your threshold. If it comes in at $112, you still have a tracking check, a creative test and a change log that show where to look. Neither number becomes more useful because the manager puts it on a slide labelled algorithmic warming.

This is also when two familiar pitches tend to arrive. Refuse both:

  • “We need to start fresh.” A fresh build discards the working ad-set setup and budget pacing you kept through the handover. Against this plan, it costs relearning time on the same budget and more of your three hours approving a new structure. It buys no evidence that the old structure was the problem.
  • “Give it 90 days before judging.” Thirty days is enough to judge whether tracking was reconciled, creative was tested inside the winners and any consolidation was supported by data. It is not a promise that every ad set will finish learning or every test will settle in 30 days. It is enough time to see whether the manager did the work in the agreed order.

Put that order in writing on day one. If the manager wants to change it, ask which result justifies the change. A pitch is not a receipt.

6. Day 30: keep, extend or leave

Cost: the final review within your three-hour week. Buys: a decision you can explain without another presentation. Use the break-even result and the change log together. A good number with no account history is hard to trust; an immaculate log with bad economics is not a win.

  1. Keep the manager if cost per result is at or under your break-even and the log shows the work in order: tracking reconciled in week one, two creative variants tested in the winning ad sets in week two, and a week-three merge only if an ad set needed one.
  2. Extend for 30 more days if tracking is clean and creative is directionally better, but the account has not yet cleared roughly 50 events per ad set since the last significant edit. Do not demand an instant verdict from a test you have just made.
  3. Leave if the manager rebuilt in week one, judges week four against a polished report instead of your break-even, or cannot tell you what changed and why. No fourth option exists.

The proof should take ten minutes to read. Meta Ads Manager activity history records who made a change, what changed and when (what activity history records). Ask for the same list in plain language beside it: date, what moved, why and what it cost against your budget. On a $3,000-a-month account, I want to understand the month in a few lines: matching Pixel and Conversions API event IDs checked, two ads added to a top ad set, one $400 test ad set merged if its results justified it, and two budget moves under 20%. Six readable lines beat sixty unexplained edits.

When the 30-day constraint lifts, add budget to the ad set that proved it, not to the account in general. Keep testing creative inside it and move budget around 20% every three to four days. If the evidence calls for a broader structural change, you can now make that case from a reconciled baseline rather than from a desire for tidy campaign names.

This is how groas takes over Meta accounts when I recommend us to an owner with no time: we keep your ad account, work inside it within 24 hours of the onboarding call and charge one flat fee of $1,499 a month for Meta with no cut of spend (see the existing-account onboarding and pricing). No rebuild for the sake of it, no 90-day waiting period, no dashboard you have to interpret. If your next manager will not agree to tracking first, creative second and structure last, keep looking until one will.