Why Adgent
Product All features It reads Account memory Daily verdict Ground truth Trust Gate Creative intelligence Change Ledger It builds Campaign build Media plan Live dashboards Agents & reports
Solutions By team For agencies For in-house By industry Ecommerce & DTC Lead generation Travel & hospitality Marketplaces Local & multi-location Mobile apps & games
Pricing About Blog Request demo

Change Ledger

The reason you can let it touch the account.

Every change is captured with the exact state it replaced, so putting it back is a restore rather than a reconstruction. That is what makes the difference between a tool you supervise and one you actually use: when a change is genuinely undoable, approving it stops being a small act of courage. The caution you have been carrying was never about the tool — it was about not being able to get back.

18 Aug 22:29 · you approved daily budget · Google Search campaign 1,500 → 1,650 TRY was: 1,500 TRY, kept same session · reverted put it back 1,650 → 1,500 TRY the exact figure it replaced — not an estimate verified back at 1,500 TRY in Google Ads itself which is why approving it was easy
Undo is only real if something kept the old value
Why reversible changes matter

The cost of a reversible mistake is an afternoon. The other kind is a quarter.

Ask anyone why they hesitate before letting software change a live account and the answer is never "I want an audit trail". It is that a bad change on a client account is expensive, visible, and hard to walk back — so the safe move is to do nothing, and doing nothing is its own cost.

Without it
  • Every change is a decision you have to be sure about
  • So the ones you are unsure about wait for a meeting
  • Reversing means someone remembering what the number used to be
  • The account drifts, slowly, in the direction of whatever nobody dared touch

Caution reads as discipline. It is mostly just a cost nobody has priced.

With it
  • You approve it and watch. The old value is held, so Thursday is a real option.
  • You test the thing you were unsure about — which is usually the one worth testing.
  • Reversing is a restore, not four people reconstructing Monday from memory.
  • You can be wrong on purpose, cheaply, which is the only way anyone learns an account.

The ledger does not make you safer. It makes you faster, because being wrong got cheap.

This is the part that sounds like a compliance feature and is not. A team that can undo anything runs more changes, learns the account faster, and stops routing every ordinary decision through a call. The log is the by-product. The behavior it unlocks is the point.

A team that can undo anything runs more changes.

Reversibility is not paperwork. It is what makes a wrong call cheap, and a cheap wrong call is how an account gets learned quickly.

  • 1approval between a proposal and a change
  • 0changes that arrive without the state they replaced
  • undos — reverting a revert is just another entry
Can you undo a change in Google Ads?

A log tells you what happened. It cannot put anything back.

Plenty of tools keep a history, and it reads convincingly right up to the moment you need it. The question is whether the replaced value was captured at the instant of the write — because if it was not, no amount of logging will reconstruct it afterwards.

01

The before-state, captured at the write

Not "budget changed" but the figure it was, stored at the moment it stopped being true. That is the only thing a revert can actually use, and it has to be taken before the change — never inferred later from a report.

02

Reversible or not, said before you approve

Some changes a platform will not take back — a bid strategy with no prior manual bid, a placement that cannot return to its automatic default. You are told which kind you are approving while deciding, not when you try to reverse it and find out.

03

The reversal goes through the same gates

An undo is a write like any other: it re-reads the live account, runs the same guardrails as the forward change, and is logged as its own entry. Reverting a revert works too. Nothing here is a special path that skips the checks.

How reverting a change works

It is not a button. It is a recommendation you can refuse.

Reversing something is a decision, so it is treated as one. Adgent proposes the inverse change, tells you what it thinks the right call is, and says plainly when the answer depends on something only you know.

Undoing something, six weeks on
This ad was paused a while back and I do not think that was deliberate. Can you put it back?
I can, and here is what I would weigh. It went ACTIVE → PAUSED on 10 August at 23:47, from chat, approved by you. Reactivating restores it exactly — the other four ads in the set are untouched.

If the pause was accidental, this undoes it cleanly. If it was intentional, hold off on approving. I do not know which, and I would rather ask than guess.
It was accidental. Put it back.
Done — PAUSED → ACTIVE, logged as its own entry. Watch the next few days: if it keeps spending with no clicks, reactivating was the wrong call and it should go back to PAUSED. I can reverse this one too — the revert has a before-state of its own.
The same ad, three entrieseach reversible
10 Aug 23:47 · you, from chat ACTIVE → PAUSED 11 Aug 22:20 · revert, approved PAUSED → ACTIVE same session · reverted again ACTIVE → PAUSED undoing an undo is just another entry

Nothing here is a one-way door.

Tracking changes you did not make

It also catches the changes nobody made.

Ad platforms move things on their own: a recommendation applies itself, a bid strategy flips, a budget reallocates under an automation somebody enabled last year. Those are read from the platforms’ own change feeds, so drift arrives with a date and a source rather than surfacing three weeks later as an unexplained CPA rise.

The honest limit: a platform-side change cannot be undone from here. No before-state was captured, because Adgent did not make the write. What it can do is date it, name it, and offer to set the value back as a change of its own.

Which settles the argument nobody can normally win

"Who changed this, and when?" costs teams more time than almost any other question in a shared ad account, and it is usually unanswerable — so it gets settled by whoever sounds most confident. Here it has a date and a source, including when the answer is the platform itself.

And the before-state is the one thing that cannot be backfilled. Start next quarter and the entries you most want — the bid strategy that flipped at 03:14, the budget that reallocated itself, the setting nobody remembers touching — are already gone, because nothing captured what they replaced at the moment they were replaced. A ledger opened in November cannot restore August. Every month of waiting is a month you can describe but not undo.

Read-only, approval, then undo

Read-only → approve → reversible.

Three gates, and a change has to pass all of them. None of these is a setting you have to find and switch on.

Next

Cheap to undo is only half of it. The other half is not doing it at all.

A reversible change is the floor. The layer above it is the one that declines to make the change in the first place when the account's own numbers say it will not work.

See what it would have caught

Ask it what changed in your account last month.

Point it at an account and it will read the platforms' own change feeds — the automated recommendations that applied themselves, the settings that moved, the dates nobody has. Read-only; nothing is written.

  • Every change dated, with its source
  • Including the ones nobody on your team made
  • Read-only until you switch execution on
What comes back

The drift in your account from the last thirty days, with dates and sources — usually including at least one change everyone assumed a person made.

We usually reply within a couple of working days.

Thanks — we’ve got it.

We’ll be in touch within a couple of working days.