Daily management should not mean daily disruption. In one advertiser’s account of a Meta agency takeover, the change log filled with more than 1,000 edits while new-customer acquisition got more expensive.

The advertiser published a self-reported takeover timeline on November 7, 2026, covering August 18 through September 5. I read it as one owner’s account, not an audited case study. Its value is the sequence: a working baseline, overlapping edits, a paused ad set, then worse business numbers.

That sequence cannot prove what caused the decline. It can show what a new manager should have to explain. The first test of daily management is whether it protects what already works.

August 18: A working account changes hands

The owner reported spending roughly $500 a day at a 1.7 to 1.9 ROAS when they handed over access on August 18. That was the starting point, not evidence that every campaign was healthy or that nothing needed fixing.

The reason for hiring was ordinary: a full-time job left no time to watch the account every day. The owner bought daily management to protect a working baseline. They were not asking someone to generate enough activity to fill a report.

I see the danger at that handover. A new manager has pressure to demonstrate value quickly. A rebuild looks decisive. Leaving a campaign alone looks suspiciously like doing nothing, especially when the invoice needs explaining.

The correct first move is less theatrical: record the baseline before changing it. Spend, CPA or ROAS, and customer volume need a shared starting point. The handover playbook linked here recommends preserving the prior three months and separating brand from non-brand so later comparisons have some meaning.

That is not a demand to freeze the account indefinitely. It is a demand to know what the new work must improve. Without it, the first month becomes an argument about which numbers count.

If your new manager cannot recite your baseline back to you, they are not ready to change it.

August 19–September 2: Activity starts replacing a test plan

The owner reported more than 1,000 changes across the takeover period. Two operators worked in the account, sometimes undoing each other’s edits. New campaigns and ad sets appeared, then stopped before they had much time to demonstrate anything.

This stretch of the account’s history matters because overlapping decisions make results harder to interpret. One person changes the test. Another reverses it. By the time someone asks what worked, the thing being measured is no longer the thing that launched.

Meta’s change history is useful here. Not because a long log automatically proves bad management, but because it lets an owner ask what each sequence of edits was supposed to accomplish. A thousand changes is a count. It is not an explanation.

There is also a delivery mechanism to respect. Meta ad sets typically need about 50 optimization events within seven days after the last significant edit to exit learning. Changes to targeting, creative, the optimization event, large budget shifts, or a pause of seven-plus days can restart learning.

Not every logged change does that. I would not infer a thousand learning resets from a thousand entries. The narrower concern is enough: repeated significant edits can interrupt delivery and keep moving the conditions of the test.

The same source recommends giving new ad sets at least seven days before editing and batching necessary changes rather than making several separate resets. That gives a test a decision window instead of a daily referendum.

Watch daily. Do not rewrite the experiment daily.

September 3: An established ad set gets switched off

On September 3, according to the owner’s log, the agency paused an established ad set with 52 conversions behind it. A second working ad stopped delivering because its parent campaign was paused. Both appear in the same reported takeover timeline.

This is the entry I would ask the agency to explain first. Not because 52 conversions make an ad set untouchable. They do not. A historical conversion count alone says nothing conclusive about its current acquisition cost, profitability, or reason to keep spending.

But an established ad set deserves a decision based on its performance, not a place in a general cleanup. The owner described it as working. If the manager disagreed, the explanation should have been available before the pause, not assembled after the results deteriorated.

There is a practical difference between stopping something that has crossed a written performance threshold and stopping it because a new structure looks tidier. The first is management. The second needs a much better argument than “we’re restructuring.”

A pause also has consequences beyond the immediate loss of delivery. If it lasts long enough, a pause of seven-plus days can restart learning. And when changes are poorly documented, their effects can later be mistaken for creative fatigue.

My rule is simple: write the pause criterion before switching off a working ad set. If nobody can explain why it must stop, keep the test separate and let the evidence develop.

September 5: Customers tell a different story from the change log

By September 5, the owner reported three business outcomes:

  • Spend was down about 14% versus baseline.
  • New customers were down about 49%.
  • New-customer acquisition cost had risen from $27 to $45.

Those figures come from the same self-reported account, not an independent audit. Read them together anyway. Spending less explains some reduction in customers. It does not, by itself, explain why customers fell so much faster than spend.

The reported acquisition cost makes the problem plain: the business was paying more for each new customer. Whatever the change log showed, that was not the outcome the owner hired the agency to produce.

I cannot attribute the entire decline to the takeover. A short window cannot rule out tired creative, an offer or landing-page change, broken tracking, or ordinary variation in who buys. The sequence raises questions; it does not settle them.

It does expose the reporting trap I discuss in Your Meta Agency Can Raise ROAS While New-Customer Growth Stalls: a headline metric can look acceptable while the business outcome you actually pay for gets worse. Here, the customer numbers demand attention regardless of how busy the account looked.

Judge the takeover against the baseline, not against the volume of work performed. Activity filled the log. Customers did not follow.

September 5: The handover conditions this record should prompt

The reported timeline ends here. What follows is my reading of what an owner should demand before the next handover, not another event in this account.

If I were choosing a company to run Meta ads, I would put three conditions in writing before granting access:

  1. A shared baseline. Record spend, CPA or ROAS, and new-customer volume. Both sides agree what the first month will be judged against. Numbers first, vibes later. Preferably never.
  2. One accountable owner. Multiple people can contribute, but someone owns the decisions. Two operators must not quietly run competing experiments inside the same budget.
  3. Written pause criteria and test windows. Specify what would justify stopping a working ad set, how long a new test gets, and who can approve an exception.

For a working winner, a proposed rule might be to keep it running while a named challenger gets at least seven days in a separate ad set. The point is not to turn seven days into permission to stop thinking. It is to prevent an impatient edit from masquerading as a conclusion.

Account ownership belongs in the same conversation. Ask who runs the account day to day and who owns the account and creative on exit. Check where the assets live: an ad account created inside an agency portfolio stays there permanently. A handover should not create a new exit problem.

The early work should prioritize understanding the account and repairing tracking where needed, rather than assuming a full rebuild is the deliverable. Tests need room, but they should earn that room instead of automatically displacing established spend.

Ask what the provider will leave untouched, not just what it will change. Anyone who cannot name what stays on has told you something useful about how they manage.

This is where I put groas ahead of the activity-first operating model: accountability should be visible in the work, not confined to the pitch. The groas philosophy says every action comes with its reasoning in plain language, so the client can see what changed, why, and what comes next. That is a more useful standard than a large change count.

Applied to this record, that standard would require an explicit reason for pausing the 52-conversion ad set and a clear owner for the decision. “We made changes” would not answer either question.

A control-room logbook with overlapping edits crowding out one winning ad set

I would inspect that first month through the log, not the pitch deck. Stop Reading Meta Agency Case Studies. Ask for the Change Log. makes the same demand: show the actual operating decisions.

The checks are concrete. Are tests judged on CPA and new-customer revenue rather than reach or likes? Are Pixel and Conversions API reconciled to orders before anyone claims a win? Does the campaign structure let automation work, or does it keep getting rearranged?

A useful log connects each meaningful change to a reason and a decision window. Without that connection, a long log proves activity, not control.

November 2026: The published account changes the hiring question

The November 7 post makes the August–September sequence available for other owners to examine. It does not turn it into a verdict on every agency. It does offer a better interview question than “How often will you optimize?”

I would ask: who gets to pause an established ad set, after how many days, and on what evidence?

That question is harder to answer with a brochure. The 2026 agency list linked here matches providers to needs rather than ranking independently audited campaign results. Fit matters, but a service category does not tell an owner how decisions will be made inside the account.

The brochures answer who does what. They rarely answer what remains untouched while the new team learns. I would trade three case-study screenshots for one clear paragraph on that.

Hire for decision discipline, not a promise of constant intervention.

A green folder labeled winner left untouched beside separate test folders

A Meta management fee should buy unglamorous plumbing before it buys a new campaign naming convention. The fee expectations outlined here include Pixel and Conversions API measurement reconciled to orders, weekly creative testing, and reporting on CPA, new-customer revenue, and margin instead of reach and likes.

That is also how I read 5 Meta Ads Management Options, Ranked for Owners With No Marketing Team. Ownership, testing cadence, and measurement matter more than the logo on the invoice. A provider still has to show how those promises govern an ordinary Tuesday in the account.

The commitment I want is plain: working campaigns keep spending while tests earn their share separately. If something must stop, the owner gets a reason, not a surprise.

Today: Change control, not change count

I read this record for what it is: one owner, one account, and numbers typed into a forum post. There is no independent audit or backend access here. I cannot separate the agency’s edits from everything else that moved during the same weeks, and I cannot say those edits caused the full increase from $27 to $45.

I do not need that stronger claim to find the process concerning. As described, overlapping operators, short-lived tests, and unexplained pauses make it harder to protect a baseline or learn from a change. The reported business result moved the wrong way while the log filled up.

That does not convict every agency. It sets the burden of proof for the next hire. The provider worth paying for sells change control, not change count.

My reading of where the line points next is straightforward: more execution does not remove the need for ownership. It makes clear guardrails and readable reasons more important. Otherwise, faster work just creates a faster argument about what happened.

Ask for the baseline in writing, the person accountable for pausing a winner, and the decision window before a new test gets judged. Ask what stays untouched in week one. If the answer is a day-one rebuild with two people learning inside your budget, keep looking.

That is the standard I hold my own work to at groas: protect what works, isolate tests, fix tracking before claiming a result, and log each move with a reason an owner can read. Daily management means watching every day and touching only when the evidence says touch. Anything busier than that is motion billed as management.