Skip to main content

Use case: Campaign mistakes

Catch costly campaign changes before they go unnoticed.

A budget typed as 2,000 instead of 200. A copied campaign that kept its old countries. AdFence compares each authorized change with the account's recorded history and the expectations your team configured, and surfaces the ones worth a second look.

AdFence does not know which value was intended. It shows the recorded change against the account's history, and whether it was meant is your team's call.

Daily budget

Spring Launch, Prospecting

Intended

$200

Recorded

$2,000

Entered at 17:42 by an authorized user, the team confirms, with one extra zero

Usual daily budget, last 30 days
$180 to $240
Configured expectation
Up to $400 per day
Size of the change
10x the previous value

Sample. Names and figures are invented.

Changed evidence

One extra zero, caught against the account's own history.

A media buyer raises a daily budget at the end of the day and types 2,000 where 200 was meant. The change is legitimate, the person is authorized, and the account is still about to spend ten times its normal day.

Sample scenario. Names and figures are invented.

Daily budget

Spring Launch, Prospecting

Before
$200
changed to
After
$2,000
Changed by
maria@example.com
Changed at
17:42
Access
Authorized user, the team confirms

Against the account

  • Usual daily budget, last 30 days

    New$180 to $240

  • Configured expectation

    NewUp to $400 per day

  • Size of the change

    New10x the previous value

What happened next

  • Recorded spend since the change

    New$260

  • Budget corrected

    $2,000changed to $200

    Set back in the platform at 18:05 by the same buyer.

  • Incident record

    NewKept, with previous and current values

AdFence compares the change with the account's recorded history and with the expectations your team configured. It does not know why the change was made. Whether it was intended is your team's call.

The correction is made by your team in the platform. The record keeps the change, the correction, and the spend recorded between them together, so the review is about the setting rather than the person.

Impact

The mistakes authorized people make.

None of these needs an attacker. They need a duplicate button, a long day, and nobody comparing the new value with the old one.

  • Budgets

    The number is wrong, or on the wrong campaign.

    • Budget entered with an extra zero
    • Budget set on the wrong campaign
    • Lifetime and daily budgets confused
  • Targeting

    The audience came from somewhere else.

    • Country list carried over from a copied campaign
    • Audience or placement left on the duplicate's settings
    • A test campaign launched to everyone
  • Destinations and settings

    The ad points somewhere, or behaves, in a way nobody chose.

    • URL typo, or the wrong product page
    • Wrong domain in a duplicated ad
    • Schedule, bid, or attribution setting carried over

Tell them apart

Authorized but unintended is its own problem.

The same budget jump can mean three different things. The response is different for each, so the pages are too.

Review and response

What AdFence surfaces, and what stays with your team.

Available history and configured expectations turn a change into something a person can judge quickly. The judgement, and the correction, stay human.

What AdFence surfaces

  • The change, with previous and current values, who made it, and when
  • How far the new value sits from the account's recorded history
  • Whether it crosses an expectation your team configured
  • An alert to the channels your team chose

What stays with your team

  • Whether the change was intended
  • Correcting it in the platform, or leaving it if it was meant
  • Updating the configured expectations when the business really changed
  • Any automatic response, which runs only if your team explicitly configured and pre-approved it

How AdFence behaves

  • Supported checks start after connection and run every 15 minutes.
  • Read-only by default. Monitoring never edits an ad account on its own.
  • Coverage varies by platform, permissions, integrations, configuration, and plan.
  • Automatic actions run only when they are explicitly configured and pre-approved.
  • No guarantee of prevention, recovery, refund, or reinstatement.

Retain the record

Keep the change and the correction together.

AdFence records the change, the comparison that flagged it, the correction, and the spend recorded in between, so the review can be about the process rather than the person.

What the record can support

  • Client reporting
  • Team reviews
  • Process fixes
  • Onboarding

The record covers changes recorded after the account was connected. It documents what changed and when. It does not guarantee a refund of the spend in between from any platform.

Correction timeline

Sample scenario. Names and figures are invented.

  1. 17:42Daily budget changed$200 to $2,000
  2. 17:45Compared with history10x the usual range
  3. 17:45Alert sentSlack
  4. 18:05Budget corrected$2,000 to $200
  5. 18:05Recorded spend in between$260

Questions, answered

What teams ask before connecting.

How does AdFence know a budget was a mistake?

It does not. It knows the change sits far from the account's recorded history, or crosses an expectation your team configured, and it surfaces that for a person to check. Whether it was intended is your team's call.

Is this the same as detecting a hacked account?

No. A takeover is an unauthorized change; a mistake is an authorized one. The same budget jump can be either. The ad account security use case covers the warning signs of unauthorized access. This page covers what authorized people get wrong.

Can AdFence put the budget back on its own?

Not by default. Monitoring is read-only. Corrections are made by your team in the platform, and any automatic response runs only if it was explicitly configured and pre-approved in advance.

What history does AdFence compare against?

The account's own activity recorded after connection, plus any expectations your team configured. A brand-new connection has little history, so early comparisons lean on the configured expectations.

Does this catch a mistake before it spends anything?

Not always. Supported checks run on a cadence after connection, so some spend can occur between the change and the review. The aim is to surface the change for review well before a month-end spend report would.

Start protecting

Give the extra zero a second look.

Read-only by default. Changes outside the usual range reach your team for review, and corrections stay with your team.

  • 7-day free trial
  • Read-only by default
  • Cancel anytime