> For the complete documentation index, see [llms.txt](https://docs.sellermagnet.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-activity-log.md).

# PPC Activity Log

A record of the PPC changes SellerMagnet made, whether by you, your bid rules, or the AI Optimizer, with a 24-hour undo for many of them.

{% hint style="info" %}
**Difficulty:** 🔴 Advanced · **Reading time:** \~14 min
{% endhint %}

{% hint style="success" icon="rocket" %}
**Open this page in your dashboard:** [**Go to PPC Activity Log \[BETA\] →**](https://dashboard.sellermagnet.com/ppc/activity-log)
{% endhint %}

{% hint style="warning" icon="triangle-exclamation" %}
**BETA:** The PPC Manager is currently in beta. Some features may behave unexpectedly or change based on user feedback. Please report any issues at <info@sellermagnet.com>.
{% endhint %}

## 📋 Overview

The **Activity Log** records every action taken on your Amazon PPC campaigns. Whether a change was made manually, by an automation rule, or by the AI Optimizer, it's logged here with full details including the source, old value, new value, and reason.

Think of it as your PPC black box recorder. If campaign performance suddenly improves or drops, the Activity Log tells you exactly what changed, when, and why.

{% hint style="info" %}
**Why this matters:** On teams with multiple users managing PPC, the Activity Log is your single source of truth. No more guessing "who changed the bid on that keyword?", every action is attributed and timestamped.
{% endhint %}

The page is available on both the PPC Manager and the PPC AI Optimizer plan. From top to bottom it has the filters, five summary cards, the six charts (collapsed behind a **Show charts** toggle by default), and the Action History table, which scrolls inside its own card instead of growing the whole page.

***

## Summary KPIs

Five summary cards give you an instant snapshot of the selected period:

| Card                | Description                                                                                                                                                                                                                                                                                                                                                                                                                  |
| ------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Total Actions**   | Number of recorded actions in the selected Time Range and Profile. It does not count entries from the System source or from other internal bookkeeping sources (such as opportunity scans), so it can be lower than the number of rows in the table. If you pick a Source, the card counts only that source                                                                                                                  |
| **Suggested**       | The number of suggestions waiting in the Approval Center right now. This is the live queue, not a count of log rows, and snoozed suggestions are not included. It does not change with Time Range                                                                                                                                                                                                                            |
| **Measured Change** | The observed change in ad profit (sales minus ad spend), comparing the 14 days after each change with the 14 days before it, summed per currency. It counts changes from the last 90 days that are at least 31 days old, measured on the changed keyword or campaign itself, with no other change to it in that window. Reverted changes are excluded. It is an observation, not proof that the change caused it (see below) |
| **Approval Rate**   | Applied actions divided by applied + rejected + expired actions. Suggestions that expired without a decision count against the rate. Next to the percentage the card shows the applied count out of the total, plus the number that expired, for example 12 / 20 with 5 expired                                                                                                                                              |
| **AI Actions**      | Actions from AI Autopilot and the AI Engine, plus the current Suggested count                                                                                                                                                                                                                                                                                                                                                |

**Reading the Measured Change card:**

* The large figure is in your project currency (or in the only currency measured). If you advertise in several currencies, hover over the card to see the figure per currency, with the number of changes behind each one.
* The detail next to the label, "{n} awaiting measurement", is the number of changes that are not yet old enough to be included.
* A positive figure means ad profit rose around the measured changes, a negative figure means it fell. Unlike a spend figure, higher is better.
* It follows the Profile selector but not the Time Range: it always looks at changes from the last 90 days that are at least 31 days old.
* It shows "-" when nothing could be measured yet.

{% hint style="success" %}
**Quick health check:** Read **Approval Rate** and **Measured Change** together. A high approval rate with a positive Measured Change means the changes you let through have been followed by better ad profit. Remember that Measured Change only covers changes at least 31 days old, so it trails what you did recently, and that it is an observation, not proof. A low approval rate can mean you keep rejecting suggestions, or that suggestions expire before you review them: check the expired count on the card and the [Approval Center](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-approval-center.md). If you keep rejecting what the AI or your rules propose, tighten the rule scope or switch to Manual Approval mode.
{% endhint %}

***

## 🔍 Filters

| Filter          | Options                                                                                                                                    | Default          |
| --------------- | ------------------------------------------------------------------------------------------------------------------------------------------ | ---------------- |
| **Time Range**  | Last 24 Hours, Last 7, 14, 30, 60, or 90 days, or Custom… (pick your own From/To dates)                                                    | Last 30 days     |
| **Source**      | All Sources, AI Autopilot, Bid Rules, User Approved, Manual, System                                                                        | All              |
| **Action Type** | All Types, Bid Change, Budget Change, Campaign State, Negate Keyword, Harvest Keyword, Target ACoS unreachable, Suggestions expired unseen | All              |
| **Profile**     | All Marketplaces, or a specific advertising profile (marketplace)                                                                          | All Marketplaces |

The last two Action Type options are explanations written by the PPC worker rather than changes to your campaigns (see [Worker notices](#worker-notices-and-launch-scorecards) below). The **Export** menu sits in the same filter area, at the top right.

{% hint style="info" icon="clock" %}
**All timestamps are in UTC.** The date each action was recorded is stored and displayed in UTC, so your local time may differ. Keep this in mind when investigating "what happened yesterday".
{% endhint %}

**Faster ways to set the same filters:**

* **Time-range preset chips.** Above the time-range dropdown, quick chips for **Last 24h**, **7d**, **30d**, and **90d** let you jump to the common windows in one click. Choosing **Custom…** in the dropdown reveals **From** and **To** date pickers.
* **Source and Action Type chip strips.** Under each of those two dropdowns is a strip of one-click chips (**Autopilot**, **Bid rule**, **Approval**, **Manual**, **System** for Source; **Bid change**, **Budget change**, **Campaign state**, **Negate keyword**, **Harvest keyword**, **Target ACoS unreachable**, **Suggestions expired unseen** for Action Type). Click a chip to apply that value without opening the dropdown.

> **Filters stack in the table.** You can combine every filter simultaneously. For example, show only "AI Autopilot" + "Bid Change" actions from the "Last 7 days" on one profile to see exactly what the AI changed to your bids this week in that marketplace.

**The KPI cards and charts do not follow every filter.** The summary cards and the charts read only the Time Range and the Profile. Choosing a Source or an Action Type changes the table and the export, but not the charts, and the only effect on the cards is that a Source narrows **Total Actions** and **AI Actions**.

| Filter          | Table and export | Charts | KPI cards                                                                                                          |
| --------------- | ---------------- | ------ | ------------------------------------------------------------------------------------------------------------------ |
| **Time Range**  | Yes              | Yes    | Total Actions, Approval Rate and AI Actions. Not Suggested (live queue) or Measured Change (fixed 90-day lookback) |
| **Profile**     | Yes              | Yes    | All five                                                                                                           |
| **Source**      | Yes              | No     | Only narrows Total Actions and AI Actions                                                                          |
| **Action Type** | Yes              | No     | No                                                                                                                 |

{% hint style="info" %}
**Hidden entry types.** Two internal entry types, "entity archived on Amazon's side" and "opportunity scored", are left out of the table, the charts and the export. The Action Type filter has no option for them.
{% endhint %}

### Filter Combinations for Common Investigations

| Investigation Goal                                    | Time Range | Source       | Action Type     |
| ----------------------------------------------------- | ---------- | ------------ | --------------- |
| "What did the AI do this week?"                       | Last 7     | AI Autopilot | All Types       |
| "Why did my spend spike yesterday?"                   | Last 7     | All Sources  | Bid Change      |
| "Which campaigns did a bid rule change the state of?" | Last 30    | Bid Rules    | Campaign State  |
| "What did I manually change last month?"              | Last 30    | Manual       | All Types       |
| "Were any new negative keywords added?"               | Last 14    | All Sources  | Negate Keyword  |
| "What keywords were harvested from search terms?"     | Last 30    | All Sources  | Harvest Keyword |

***

## Charts

Between the summary cards and the table, click **Show charts** to open six chart cards (click **Hide charts** to collapse them again; your choice is remembered for next time). Once open, they load automatically. They follow the **Time Range** and **Profile** filters only, so Source and Action Type have no effect on them. They use the same hidden-entry rule as the table. The charts read at most 5,000 entries from the selected period, so on very busy accounts they may not reflect every entry.

| Chart                             | Subtitle                         | What it shows                                                                                                                                                                                                            |
| --------------------------------- | -------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Actions Over Time**             | Daily action count by source     | A line chart with one line per source group (Manual, Autopilot, Bid Rules, Preview Apply, Other) and one point per day. Days without activity count as zero, and only groups that have activity in the period get a line |
| **Success Rate Trend**            | Daily % of applied actions       | The share of each day's decided entries that were applied. Applied, rejected, failed and error entries count as decided. Suggested, pending and expired entries are not counted                                          |
| **Action Type Breakdown**         | Top 8 types + Other              | A donut chart of the eight most frequent action types, with everything else grouped as Other                                                                                                                             |
| **Source Distribution**           | Who/what made the changes        | A donut chart of the share of entries per source group. Sources other than Manual, Autopilot, Bid Rules and Preview Apply are grouped as Other                                                                           |
| **Top Affected Entities**         | Top 10 by action count           | A bar chart of the ten entities (keywords, campaigns and so on) with the most actions, each labeled Keyword, Campaign, ASIN or Other                                                                                     |
| **Activity Heatmap (Day × Hour)** | When automation fires most - UTC | A grid of Monday to Sunday by hour of the day, in UTC. Darker cells mean more actions, with a Less / More legend. Hover a cell to see the count for that day and hour                                                    |

If there is nothing to draw, a chart shows "No activity yet - autopilot actions will appear here once enabled." If the charts cannot be loaded, each card shows an error message with a **Retry** button instead of the empty message, so a failure is never shown as "no activity".

***

## Activity Table

![PPC Activity Log - Actions timeline and filters](https://4139654583-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fih2knSvVpBcq5rIeXVjM%2Fuploads%2Fgit-blob-01fadc218aab81ba2bc7f58ea70a3faffd8ad0ae%2Fppc-activity-log.png?alt=media)

![PPC Activity Log filtered to AI Autopilot source only](https://4139654583-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fih2knSvVpBcq5rIeXVjM%2Fuploads%2Fgit-blob-e955eefd198c01c66a6d030257653601f796c072%2Fppc-activity-log-2.png?alt=media)

| Column       | Description                                                                                                                                                                                                                                                                                                                            |
| ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **When**     | How long ago the action happened (a relative time such as minutes or hours ago, or a date once it is a week old). Hover over it for the full date and time. The table is always sorted newest first and grouped under day separators that read "Today", "Yesterday", a weekday name or a date, each with a count of that day's entries |
| **Source**   | Color-coded badge (see [Understanding Sources](#understanding-sources)). For manual and approved actions, the specific user's email is shown beneath the badge                                                                                                                                                                         |
| **Activity** | What happened: the action type (Bid Change, Budget Change, Campaign State, etc.) together with the keyword, campaign, or ad group affected                                                                                                                                                                                             |
| **Campaign** | Campaign name, with the marketplace flag                                                                                                                                                                                                                                                                                               |
| **Change**   | Previous value to new value, with a direction arrow and the percentage/absolute delta                                                                                                                                                                                                                                                  |
| **Status**   | Outcome badge (see [Status badges](#status-badges))                                                                                                                                                                                                                                                                                    |
| **Reason**   | Explanation of why the AI or rule recommended the change. For worker notices and campaign launches it also holds a translated explanation or a scorecard (see [Worker notices](#worker-notices-and-launch-scorecards))                                                                                                                 |
| **Undo**     | One-click reverse of the action within a **24-hour** window (see [Undo](#undo))                                                                                                                                                                                                                                                        |
| **Result**   | Whether the change worked, measured once 31 days have passed (see [Result column](#result-column))                                                                                                                                                                                                                                     |

There is no sort control: the column headers are fixed, and the order is always newest first.

**Above the table:**

* **Search.** The "Search the log..." box filters the rows that are already loaded. A note under the box says "Search covers loaded entries. Load older entries below to search more history." While you search, a count reading "{shown} of {total} loaded entries" appears next to a **Clear** button.
* **Columns.** The columns button (tooltip "Show or hide columns") lets you choose which columns are visible.
* **Undo pill.** When the view contains applied changes that can be undone, a pill reads "{n} of {m} changes in this view can be undone".

**Pagination.** The table loads 500 entries at a time. The footer reads "Showing {shown} of {total}", and a **Load older** button loads the next page. Only the first page includes the live "Suggested" rows, so they do not repeat on later pages.

### Status badges

| Badge                                                                    | Meaning                                                                                                                                                                                                 |
| ------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Applied**                                                              | The change was sent to Amazon                                                                                                                                                                           |
| **Suggested**                                                            | A suggestion queued for your approval                                                                                                                                                                   |
| **Rejected**                                                             | You rejected the suggestion                                                                                                                                                                             |
| **Expired**                                                              | The suggestion expired before it was decided                                                                                                                                                            |
| **Failed**                                                               | The change could not be applied. The Reason column shows a short headline and a hint, and a **Technical details** disclosure holds the raw error                                                        |
| **Sending**                                                              | The worker is sending this change to Amazon. Hover over the badge for the hint: if the worker restarted mid-send, a later run checks the campaign's synced state and marks the change Applied or Failed |
| **Held**                                                                 | An approval batch whose suggestions were held back instead of being queued. This is a wait, not a failure                                                                                               |
| **Notice**                                                               | An explanation from the worker rather than a change (see [Worker notices](#worker-notices-and-launch-scorecards))                                                                                       |
| **Detected**, **Error**, **Pending**, **Queued**, **Skipped**, **Stale** | Other status values the worker can record. The Reason column gives the context                                                                                                                          |

### Result column

The **Result** column shows whether an applied change worked. Hover over the header for the definition: the 14 days after the change compared with the 14 days before it, measured once 31 days have passed. Rows that were not applied show "-".

| Chip                     | Meaning                                                                                                                                                                                                          |
| ------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Too early**            | The change is not old enough yet. The chip shows "Result on {date}"                                                                                                                                              |
| **Improved**             | The result was better. The chip shows "Measured on {date}", and hovering shows the before and after ACoS and spend                                                                                               |
| **No clear change**      | Nothing clearly moved                                                                                                                                                                                            |
| **Got worse**            | The result was worse. The cell also shows the previous value and a link to open the campaign so you can restore it, because the 24-hour undo window is long over by then                                         |
| **Not control-adjusted** | The change was measured, but too few comparable untouched entities existed to grade it, so the raw before and after is shown but not graded as better or worse                                                   |
| **Not measurable**       | The change could not be measured. Hover for the reason, for example "Another change hit the same target within 14 days, so the effects cannot be separated." or "This change was undone, so it is not measured." |
| **Result unavailable**   | The result data could not be read. Reload to try again                                                                                                                                                           |

When several changes to the same target are graded together, the cell also shows "Chain of {count} changes" with the first and last date. Like the Measured Change card, a result is an observation, not proof that the change caused it.

### Undo

{% hint style="success" icon="rotate-left" %}
**One-click Undo.** Many applied changes can be reversed straight from the **Undo** column for up to **24 hours** after they were made. The button reads "Undo ({remaining} left)", for example "Undo (23h 59m left)", counting down. Clicking it opens the confirmation "Reverse this action?" with **Yes, undo** and **Cancel**, then sends the reversal to the Amazon Ads API and restores the previous value, no need to dig through Amazon's console.
{% endhint %}

**What can be undone:** bid changes, budget changes, campaign state changes (including pausing a campaign), negated keywords, keyword pause and enable, placement modifier changes, product ad pause and enable, and ad group default bid changes. Keyword and product ad state changes can be undone only for the paused and enabled states, not for archived. Harvested keywords cannot be undone, and neither can any other action type that is not listed here. A placement rebalance writes a summary row that cannot be undone, but its two child changes can each be undone separately.

**What the Undo cell can show:**

| You see                     | Meaning                                                                                          |
| --------------------------- | ------------------------------------------------------------------------------------------------ |
| **Undo ({remaining} left)** | The change can be reversed now                                                                   |
| **Undone {when}**           | The change was reversed                                                                          |
| **Undo failed**             | The reversal did not go through                                                                  |
| **Expired**                 | The 24-hour undo window for this action has passed                                               |
| **Replaced**                | A newer change to the same item replaced this one, so undoing it would overwrite the newer value |
| **-**                       | Undo is not available for this action type, or the row was not applied                           |

**If the value has changed since.** Before it reverses a change, the page checks the live value. If it differs from the value SellerMagnet left, a second confirmation appears: "This item changed after SellerMagnet applied the action: it is now {live}, not {expected}. Undo would set it to {restore} and replace the current value." The button then reads "Set to {restore} anyway". Only confirm if you want to overwrite the current value.

### Worker notices and launch scorecards

Some entries are explanations written by the PPC worker rather than changes. They appear in the **Reason** column as a sentence in your language with a blue **Notice** badge, and most end in a next step. Examples:

* **Target ACoS unreachable:** your target needs a maximum bid that your conversion rate and order value cannot support. The next step is **Check target in settings**.
* **Suggestions expired unseen:** "{count} suggestions were held for {block\_days} days after expiring unseen {threshold}× in {window\_days} days." The next step is **Open Approval Center**.
* **Low engagement:** a line saying how many suggestions expired or were applied out of those decided in the period, and whether suggestion generation is being throttled until you review the queue.
* **Listing check:** clicks arrive but the conversion rate is well below the reference, with a prompt to check the listing, price, Buy Box and stock before cutting the bid.
* **Cycle degraded:** "This cycle ran with {count} safety check(s) falling back to a safe default. Our team has been notified."

**Launch scorecard.** For campaign launches, the Reason cell can hold a "Launch scorecard". It shows spend, orders and ACoS for "Days 1-14" and "Days 15-30", the number of converting search terms found for discovery launches, and a verdict: "hit its goal", "close to its goal", "missed its goal" or "too little traffic to judge". Until the launch is old enough, it shows the date it will be scored.

**Already in place.** A harvest that was applied but changed nothing gets an **Already in place** chip: "This keyword already existed in the destination ad group, so nothing was sent to Amazon. No action needed."

**Already excluded.** A negate that was applied but changed nothing - Amazon already had this exact negative - gets an **Already excluded** chip, the same no-op treatment as a harvest's **Already in place**: nothing was sent, no action needed.

**Archived campaign or ad group.** A negate skipped because its campaign or ad group is archived on Amazon reads **Skipped - campaign or ad group archived, no negative needed** instead of **Failed**. Nothing was sent, and no action is needed.

***

## Understanding Sources

The dropdown and the badges in the table use slightly different names. The filter chips use a third, shorter set of names.

| Badge in the table       | Source filter option | Color  | Meaning                                                                                                                         |
| ------------------------ | -------------------- | ------ | ------------------------------------------------------------------------------------------------------------------------------- |
| **AI Autopilot**         | AI Autopilot         | Purple | Change made autonomously by the AI Optimizer                                                                                    |
| **Bid Rule**             | Bid Rules            | Blue   | Change executed by a scheduled bid automation rule                                                                              |
| **Approved**             | User Approved        | Green  | AI suggestion that you approved in the Approval Center                                                                          |
| **Manual**               | Manual               | Orange | Change you made directly                                                                                                        |
| **System**               | System               | Gray   | Entry logged by SellerMagnet's own system rather than by you, a rule or the AI. System entries are not counted in Total Actions |
| **AI Engine**            | none                 | Indigo | Entry written by the AI engine itself. It counts toward the AI Actions card                                                     |
| **Opportunity Scan**     | none                 | Teal   | Entry from the pass that scores harvest opportunities                                                                           |
| **Outside SellerMagnet** | none                 | Orange | A change made outside SellerMagnet, in the Amazon console or another tool (see the FAQ for what this covers)                    |
| **Unknown**              | none                 | Gray   | The entry has no source recorded                                                                                                |

### Source Detailed Analysis

{% hint style="info" %}
**AI Autopilot vs Approved:** Both originate from AI suggestions, but the difference is critical. **AI Autopilot** changes were applied automatically (your project is in Fully Automatic mode). **Approved** changes went through the Approval Center and you explicitly approved them. If you see AI Autopilot actions you disagree with, switch to Manual Approval mode in PPC Settings.
{% endhint %}

> **System entries** are informational. If you see one you do not recognize, read its Activity, Campaign and Reason columns for context.

***

## Use Case Scenarios

### Scenario 1: Investigating a Sudden Spend Spike

> **Situation:** Your daily PPC spend jumped from EUR 80 to EUR 140 overnight and you want to know why.
>
> **How to investigate:**

{% stepper %}
{% step %}
Set **Time Range** to "Last 7 days."
{% endstep %}

{% step %}
Set **Action Type** to "Bid Change."
{% endstep %}

{% step %}
Set **Source** to "All Sources."
{% endstep %}

{% step %}
Find the day spend spiked: the table is always newest first and grouped under day separators ("Today," "Yesterday," then a weekday or date).
{% endstep %}

{% step %}
Look for large bid increases (green arrows in the Change column) on that day.
{% endstep %}

{% step %}
Check the **Source** column, was it a bid rule, AI action, or manual change?
{% endstep %}

{% step %}
Read the **Reason** column to understand the logic behind the increase.
{% endstep %}
{% endstepper %}

> **Common causes:** A bid rule with overly aggressive increase settings (e.g., +25% on any keyword with CVR > 5%) or the AI Autopilot responding to a sudden conversion spike by aggressively bidding up.

### Scenario 2: Auditing AI Performance Over 30 Days

> **Situation:** You have been running AI Autopilot for a month and want to evaluate whether it is helping.
>
> **How to investigate:**

{% stepper %}
{% step %}
Set **Time Range** to "Last 30 days."
{% endstep %}

{% step %}
Set **Source** to "AI Autopilot."
{% endstep %}

{% step %}
Note the total number of actions in the **Total Actions** KPI card. With a Source selected, the card counts only that source.
{% endstep %}

{% step %}
Filter **Action Type** to "Bid Change" and scan the **Change** column: in most cases the AI should be making more decreases (red, down arrows) than increases (green, up arrows). The KPI cards and charts stay as they were, because they do not follow Action Type.
{% endstep %}

{% step %}
Check the **Result** column and the **Measured Change** card: are the changes that are old enough to be measured marked Improved, and is the measured profit change positive? Cross-reference with your campaign performance: Is ACoS (Advertising Cost of Sale) trending down? Is wasted spend decreasing?
{% endstep %}

{% step %}
If the AI made changes you disagree with, consider switching to Manual Approval mode.
{% endstep %}
{% endstepper %}

### Scenario 3: Preparing a Team Performance Review

> **Situation:** You manage PPC for a team and need to report on what was done this month.
>
> **How to use the Activity Log:**

{% stepper %}
{% step %}
Set **Time Range** to "Last 30 days."
{% endstep %}

{% step %}
Filter by **Source** = "Manual" to see all human-made changes.
{% endstep %}

{% step %}
Use the **Export** menu to download the filtered view as CSV or Excel, or screenshot the results for your report.
{% endstep %}

{% step %}
Then filter by "AI Autopilot" and "Bid Rules" to show automation contributions.
{% endstep %}

{% step %}
Compare the ratio of manual vs automated actions to demonstrate the efficiency gains from automation. The Source Distribution chart shows the overall split for the period.
{% endstep %}
{% endstepper %}

***

## Before/After: How the Activity Log Reveals Hidden Problems

| What You See in Campaigns        | What the Activity Log Reveals                                             |
| -------------------------------- | ------------------------------------------------------------------------- |
| ACoS dropped from 32% to 24%     | Bid Rules decreased bids on 45 high-ACoS keywords over 3 weeks            |
| Top keyword lost impressions     | A bid rule decreased the bid 4 times in 2 weeks (rule too aggressive)     |
| New keyword appeared in campaign | AI Autopilot harvested it from search term report 5 days ago              |
| Campaign paused unexpectedly     | A Campaign State entry whose Source and Reason show who or what paused it |
| Spend shifted between campaigns  | Manual budget reallocation by team member on Tuesday                      |

***

<details>

<summary><strong>⚠️ Common Mistakes to Avoid</strong></summary>

{% hint style="danger" %}
**Mistake 1: Ignoring the Activity Log after enabling automation.** Setting up bid rules or AI Autopilot and never reviewing the log is like hiring an employee and never checking their work. Review the log weekly at minimum. **Fix:** Set a calendar reminder to review the Activity Log every Monday morning.
{% endhint %}

{% hint style="danger" %}
**Mistake 2: Not filtering before investigating.** Scrolling through hundreds of unfiltered entries is inefficient and leads to missed insights. **Fix:** Always start by narrowing to a specific time range and source before analyzing.
{% endhint %}

{% hint style="warning" %}
**Mistake 3: Confusing correlation with causation.** A bid increase on Monday does not necessarily explain a sales spike on Wednesday - Amazon's attribution window means the cause could be older. The Measured Change card and the Result column are observations in the same way, not proof. **Fix:** Look at changes from 3-7 days before a performance shift, not the same day.
{% endhint %}

{% hint style="warning" %}
**Mistake 4: Overlooking entries you do not recognize.** An entry from the System source, or one marked Outside SellerMagnet, can change how a campaign behaves without you having made it. **Fix:** Read its Source, Activity and Reason columns, and if you cannot explain it, contact <info@sellermagnet.com>.
{% endhint %}

</details>

\## 📤 Export

Use the **Export** menu in the filter area at the top right of the page to download the **currently filtered view**. The menu offers two formats:

* **CSV** , for spreadsheets, BI tools, and plain archiving
* **Excel** , a ready-to-open `.xlsx` workbook

Whatever time range, source, action-type and profile filters you have applied are reflected in the file, so you can export, say, only "AI Autopilot" + "Bid Change" actions from the last 7 days. The file is not limited to the rows already loaded in the table.

{% hint style="info" icon="lightbulb" %}
**Tip:** Because the Activity Log shows the last 90 days, export on a regular schedule (monthly works well) if you need a longer-term audit archive for compliance or year-over-year review.
{% endhint %}

***

## Reading the Activity Log Like a Pro

### Pattern Recognition Checklist

Use this checklist during your weekly Activity Log review:

* [ ] **Bid oscillation:** Is the same keyword being increased and decreased repeatedly? This signals conflicting rules or an AI strategy that does not match your goals.
* [ ] **Bid floor hits:** Are many keywords hitting your minimum bid? Your min bid floor may be too high, or your decrease rules may be too aggressive.
* [ ] **Cluster changes:** Did many keywords change on the same day? This usually means a bid rule executed on its schedule. Verify the rule is producing the intended results. The Activity Heatmap shows when automation fires most.
* [ ] **Source distribution:** Is one source dominating? If 90% of changes come from a single bid rule, that rule might be too broadly scoped. The Source Distribution chart shows the split.
* [ ] **Unusual reasons:** Read the Reason column for anything unexpected, such as "Insufficient data" or "Confidence below threshold."
* [ ] **Results:** Look for **Got worse** chips in the Result column and decide whether to restore the previous value.

***

## 🔧 Troubleshooting

### Activity Log shows no entries

| Check                                | Solution                                                                           |
| ------------------------------------ | ---------------------------------------------------------------------------------- |
| Are filters too restrictive?         | Reset all filters to "All" and check if entries appear                             |
| Is the time range too narrow?        | Expand to "Last 90 days" to verify data exists                                     |
| Are any rules or AI features active? | If no automation is running and no manual changes were made, the log will be empty |
| Is the API connection active?        | Verify your Amazon Advertising API credentials in Settings                         |

### Entries appear but are missing details

| Check                                    | Solution                                                                                      |
| ---------------------------------------- | --------------------------------------------------------------------------------------------- |
| Is the Reason column empty?              | Some manual changes may not have an attached reason. This is expected                         |
| Is the Change value blank?               | For "Campaign State" actions (pause/enable), there is no numerical value, only a state change |
| Is the Result column "-" or "Too early"? | Only applied changes get a result, and a result is measured once 31 days have passed          |

### Cannot find a specific change

| Check                                   | Solution                                                                                                                                                                        |
| --------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Was the change made in Amazon directly? | The page can show an **Outside SellerMagnet** badge, but do not rely on the log to capture edits made in the Amazon console. Check the Amazon Ads console as well (see the FAQ) |
| Is it an older entry?                   | The table loads 500 entries at a time and search covers only loaded entries. Click **Load older** at the bottom of the table, then search again                                 |
| Was the change in a different account?  | Confirm you are viewing the correct marketplace/account                                                                                                                         |

***

## ❓ FAQ

<details>

<summary><strong>How long is Activity Log data shown?</strong></summary>

The Activity Log shows the last 90 days, which is also the longest Time Range you can choose. If you need a longer history for compliance purposes, export your data regularly.

</details>

<details>

<summary><strong>Can I export the Activity Log?</strong></summary>

Yes. Use the **Export** menu at the top right of the page to download the currently filtered view as either **CSV** or **Excel** (.xlsx). This is useful for reporting, auditing, and sharing with team members.

</details>

<details>

<summary><strong>Are changes made directly in Amazon Seller Central tracked here?</strong></summary>

Not reliably. The Activity Log is built around changes made through SellerMagnet (manual changes, bid rules, AI Autopilot, and system actions). The page also has an **Outside SellerMagnet** source badge and a **Changed outside SellerMagnet** action type, meant for edits that SellerMagnet's sync detects in the Amazon console or another tool. Such an edit is also meant to start the same cooldown that applies after a change made in SellerMagnet. The Source filter has no option for these entries, so use the search box to find them. Do not treat the log as a complete record of edits made directly in Amazon, and check the Amazon Ads console when you need to be sure.

</details>

<details>

<summary><strong>I see a "System" action I don't understand. What should I do?</strong></summary>

Open the row's Activity, Campaign and Reason columns for context, and compare the date with what you changed around that time. System entries are not counted in the Total Actions card. If it is still unclear, contact <info@sellermagnet.com>.

</details>

<details>

<summary><strong>Can I reverse a change shown in the Activity Log?</strong></summary>

Often, yes. Applied bid and budget changes, campaign state changes, negated keywords, keyword and product ad pause and enable, and a few other types have a **one-click Undo** in the **Undo** column for up to **24 hours** after the change was made. Clicking it confirms ("Reverse this action?"), then sends the reversal to the Amazon Ads API and restores the previous value. If the value has changed since, you get a second confirmation before it is overwritten. After the 24-hour window, or for non-reversible actions (such as harvested keywords), there is no Undo button: the cell reads **Expired** or "-". In that case, read the previous value from the **Change** column and set it back manually. For ongoing protection, use Manual Approval mode in the [Approval Center](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-approval-center.md) so you approve changes before they are applied.

</details>

<details>

<summary><strong>Why does the Approval Rate card disagree with my own count of approvals and rejections?</strong></summary>

Approval Rate is applied divided by applied + rejected + expired, so suggestions that expired without a decision count against it. The detail next to the percentage shows the applied count out of the total, plus the number that expired.

</details>

***

## ➡️ What's Next?

{% content-ref url="/pages/gVXllc12Kbkb0xuLJfxw" %}
[PPC Settings](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-settings.md)
{% endcontent-ref %}

{% content-ref url="/pages/pQTsyjXdDCQAMg0Eueka" %}
[PPC Analytics](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-analytics.md)
{% endcontent-ref %}
