What the workflow has done
Access from: Advertising → Smart Campaigns Performance
Beta feature
The Smart Campaigns optimizer is in beta: it is enabled account by account, and both the interface and the default values may change. If you don't see the template under Available templates, nor the AI Settings button in your campaigns table, it is not enabled for your account yet.
This screen answers the question that shows up the day after you turn the workflow on: is it working, and what has it touched? At the top, three cards with the overall state; below, a table with one row per managed campaign; and for each campaign, the full history of changes with their reason.
Only managed campaigns appear here
The table lists the campaigns that some workflow has optimized at least once. If you have just created the workflow and it hasn't run yet, it will be empty — launch a run or wait for the early-morning cycle.
The summary band
The Window (days per period) selector at the top sets the period the three cards and the table look at, and it is always compared against the previous period of the same length.
| Card | What it tells you |
|---|---|
| Workflow execution health | How many runs completed, failed, were cancelled or are still running. Repeated failures here mean the problem is the workflow, not your bids |
| AI changes | How many changes the AI made in the period, broken down by action type. Each action is counted once, even if internally it takes several steps |
| Target vs actual ACOS | The target ACOS against the actual one, with its trend and the number of campaigns included |
The managed campaigns table
One row per campaign, with the verdict up front:
| Column | What it holds |
|---|---|
| Campaign | The campaign's name in Amazon |
| Verdict | Good / Watch / Bad / No data — see the rule below |
| Recent ACOS | The ACOS for the chosen period |
| Target | The objective set in that campaign's AI Settings |
| Δ vs previous | How much the ACOS went up or down versus the previous period |
| Reasons | The verdict explained in words: which threshold was crossed |
| Optimizing since | Since when the workflow has managed that campaign |
| Performance since start | Spend and sales accumulated since the workflow started optimizing it |
Each row's History button opens the detail of every change made to that campaign.
How the verdict is decided
The verdict compares the period's ACOS with your target and with the previous period:
| Verdict | When it appears |
|---|---|
| Good | The ACOS meets the target and holds or drops versus the previous period |
| Watch | It is above target but contained, or it meets the target while rising |
| Bad | The ACOS exceeds 1.5 times the target, or it rose by more than 40 % versus the previous period |
| No data | Not enough sales in the period, or the campaign has no objective configured |
"Bad" does not always mean the workflow is doing badly
A spiking ACOS can come from outside: you ran out of stock, you lost the Buy Box, or an aggressive competitor moved in. Before changing the autonomy mode, read the Reasons and the history: if the AI is already lowering bids and blocking terms, it is reacting and the problem is in the offer.
A campaign's history
The Workflow history tab lists everything the AI has decided about that campaign, most recent first. You can reach it from two places: the History button in the table above, or from Advertising → AI Campaigns by opening a campaign, where it sits as one more tab next to Ad groups and Performance.
| Column | What it holds |
|---|---|
| Date | When the decision was taken |
| Action | What it did (or decided not to do) — see the table below |
| Detail | The affected entity and the numbers: old and new bid, clicks, ACOS, orders… |
| Reason | Why it did it, in plain words |
Three filters narrow the list down:
- All actions — keep a single action type (only the negatives, for instance).
- All executions — keep the changes from one specific cycle, to see what the workflow did that day.
- Show simulations (dry-run) — include the test-mode runs, tagged as Simulation. They are excluded by default so they don't get mixed up with real changes.
The actions you will see:
| Action | What happened |
|---|---|
| Raise bid | The target sells with ACOS below the objective: raise it |
| Lower bid | The ACOS is well above the objective, or it spends without selling |
| Hold | The ACOS is in the acceptable band: leave it alone |
| Harvest | A search term proved sales and is promoted to a keyword |
| Keyword created | The keyword now exists in Amazon — the outcome of harvesting or discovery |
| Negate | A term was spending without selling and gets blocked |
| Placements adjusted | The placement adjustments were changed (or set to 0 %) |
| Quarantined / Reactivated / Blocked | The target went under observation, recovered sales, or was blocked at the minimum after 21 days |
| Wait | It was evaluated and the decision was not to act yet: the reason says why (insufficient signal, brand protection, collision with an active keyword…) |
Decisions are kept for 90 days
The history shows the last 90 days. The Wait entries are the most useful for understanding the workflow: they explain why it did not do something you expected.
The campaign's phase
In the Advertising → AI Campaigns table, the Phase column sums up at a glance where each campaign stands:
| Phase | What it means |
|---|---|
| Exploring and Structuring | It does not meet the target yet and is less than 30 days old. Visibility is prioritized to gather the data needed to decide |
| Active Performance | It meets the target. The AI makes continuous micro-adjustments to sustain profitability |
| External Friction Detected | More than 30 days without reaching the target. It usually means the brake is not in advertising but in the offer: price, competition or stock |
Next steps
- Smart Campaigns Optimizer — what it decides on each cycle and which campaigns it acts on
- Autonomy modes and variables — raise or lower the AI's initiative based on what you see here
- Workflows — run history and inspector, to debug a failed run