A one-page ad decision log you can start today: what to record, a template with examples, and how to review it monthly.
Ad decision tracking means writing down, for every meaningful change to an ad account, what you changed, why, what you expected and what happened. A one-page decision log is enough: one row per decision, with a baseline, a hypothesis and an outcome you fill in later.
Ad platforms record that a change happened. They don't record why you made it. Without the reason, you can't tell a good call from a lucky one, and you repeat the same mistakes. This guide gives you the log, a template you can copy and the monthly review that makes it worth keeping.
AI Marketer recommendations are reviewed and approved by you.
Why the reason behind a decision gets lost
Three months after you doubled a budget, you remember doing it. You rarely remember what you saw that day, what you expected, or what you ruled out. Teams make this worse: the person who made the change moves on, and the reasoning leaves with them.
Meta does keep some history. Business history in Meta Business Suite is an activity log that people with full control of a business portfolio can download as a CSV file. According to Meta's page on downloading people permissions and business history, it lists actions such as new users being added, roles being updated and partners being granted access to assets. For each action it gives the time (GMT), a description, the event type, and the business user and object involved.
That log is useful for access and permissions. It is not a record of why you chose a budget or paused an ad, and it can't be. The "why" has to come from you.
Which ad decisions deserve a log entry
If you log everything, you will stop logging within a week. Log decisions, not tweaks. A good test: would you want to know, in three months, why you did this?
| Log it | Skip it |
|---|---|
| Pausing or relaunching an ad or ad set with meaningful spend | Pausing an ad that never delivered |
| Raising or cutting a budget by a large step | A small budget nudge inside your normal range |
| Changing the audience, placement or optimization goal | Fixing a typo in the ad name |
| Launching a new creative as a test | Duplicating an ad with no change |
| Changing a tracking or attribution setting | Reordering columns in a report |
The last row matters more than it looks. When you change tracking or attribution, every number after that is measured differently, and a log entry is the only thing that tells you where the line is. See our guide to iOS 14 and Meta ads tracking changes for what changed in tracking and what to do about it, and Madgicx Cloud Tracking for how to improve the conversion data these decisions rest on.
What each entry holds
Each entry has seven fields. Keep them short, one line each.
- Date. The day you made the change.
- What changed. The campaign, ad set or ad, and the exact change.
- Baseline. The number you are trying to move, as it stands now, over a named date range.
- Hypothesis. "Because X, I expect Y." If you can't fill this in, you are guessing, which is fine, but write that down.
- Expected result. The number you expect, and by when.
- Review date. When you will check.
- Outcome. Left blank until the review date.
The baseline is the field people skip, and it is the one that makes the log useful. "Cost per purchase was high" can't be checked. "Cost per purchase was $42 over the last 7 days" can.
A copy-ready log template with example entries
Make a sheet with these columns:
| Date | Change | Baseline | Hypothesis | Expected | Review on | Outcome | Score |
|---|---|---|---|---|---|---|---|
| Mar 3 | Paused Ad B in the retargeting ad set | Cost per purchase $58 vs $34 for Ad A, last 7 days | Ad B pulls budget from the better ad | Ad set cost per purchase falls toward $34 | Mar 10 | (fill in later) | (fill in later) |
| Mar 5 | Raised daily budget from $50 to $60 | Cost per purchase $31 for 7 days, stable | Results will hold with a small step | Cost per purchase stays under $36 | Mar 12 | (fill in later) | (fill in later) |
| Mar 8 | Swapped the lead video for a new hook | Frequency 3.2 and falling click-through rate over 7 days | The audience is tired of the old hook | Click-through rate recovers within a week | Mar 15 | (fill in later) | (fill in later) |
These numbers are made up to show the format. Use your own. Notice the budget entry: a small step, with a stated limit. A log also shows you how often you are touching the same ad set.
Record the outcome and score the call
On the review date, fill in what happened, using the same metric and the same date range as the baseline. Then give the decision one of three scores:
- Right call: the result moved as expected, and the reason holds.
- Wrong call: the result did not move, or moved the wrong way.
- Unclear: something else changed at the same time, or there wasn't enough data.
Be strict with "unclear". If you changed three things the same day, you can't credit any one of them. A decision can also be a good call with a bad result, and a bad call with a good one, so score the reasoning as well as the number.
Monthly review: find patterns
Once a month, read the whole log in one sitting. You are looking for patterns, not single entries:
- Which type of decision has the best hit rate: pauses, budget changes or creative swaps?
- Which hypotheses are repeatedly wrong? That is a belief to drop.
- Which "unclear" entries share a cause, such as changing several things at once?
- How often did you edit the same ad set within a week?
Write two or three rules from what you find ("no budget change above 20% in one step"; "one change per ad set per week") and add them to the top of the sheet. After three months, those rules are your own playbook, built from your own account. For more on structured thinking about these calls, see our post on making better advertising decisions.
Common reasons logs get abandoned
- Too many fields. Keep it to seven, one line each.
- Logging tiny changes. Use the "log it, skip it" table above.
- Nobody owns it. Name one person who reviews it each month.
- It lives somewhere you don't look. Keep it next to the dashboard you already open.
- No review date. An entry with no outcome is a diary, not a log.
Log decisions as you approve them in Madgicx
AI Marketer recommends actions on your Facebook and Instagram ad accounts, and you review and approve them. That review step is the natural moment to write the reason into your log, because the recommendation is on the screen in front of you. You can copy what changed and why into the sheet in a few seconds, then add your own hypothesis.
There is also a decision queue in beta. It shows recommendations with a confidence level, stacked into a plan that you push. It's in beta; contact our support team to get beta access. If you use it, the confidence level is a good thing to write in the hypothesis column.
Whether or not you use Madgicx, the habit is the same: decide, write down why, and check later.
FAQ
What is ad decision tracking?
It is keeping a log of the meaningful changes you make to ads, with the reason, the baseline and the expected result, and checking the outcome later. It turns changes into lessons.
What should I put in an ad decision log?
The date, what changed, the baseline number, your hypothesis, the expected result, a review date and the outcome. One line for each field is enough.
Does Meta keep a history of changes to my ads?
Meta Business Suite keeps a business history activity log of actions such as users added, roles updated and partners granted access, which you can download as a CSV file. It does not record why you made a change.
How often should I review my decision log?
Check each entry on its own review date, usually a week or two after the change, and read the whole log once a month to look for patterns.
Which ad changes are worth logging?
Log decisions you would want to explain in three months: pauses, large budget changes, audience or goal changes, creative tests and tracking changes. Skip small tweaks and typo fixes.
