Autonomy modes and variables
Access from: Processes → Workflows → your workflow → Variables tab
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.
The autonomy mode decides how much initiative the AI takes over your campaigns. Each mode adjusts several parameters at once, so you don't need to understand them one by one to get started.
The four modes
| Mode | What it does |
|---|---|
| Maintenance | The AI only adjusts bids: it neither harvests new terms nor proposes keywords, and it takes longer to block terms with no sales. Maximum control, minimum risk |
| Growth | On top of adjusting bids, it turns search terms that already sell into keywords, with moderate thresholds. Recommended for most accounts |
| Expansion (default) | It harvests winning terms as keywords, proposes brand-new keywords and blocks spending-but-not-selling terms sooner; it also prunes exhausted parent keywords. Maximum autonomy and coverage, at the cost of more spend |
| Custom | You don't pick it: it is what appears as soon as you edit any parameter by hand. You control each one separately |
What each mode adjusts
These are the parameters that make the difference between modes:
| Parameter | Maintenance | Growth | Expansion |
|---|---|---|---|
| Discovery level | Bids only | Harvesting | Harvesting + new keywords |
| Discovery: add AI generation | No | No | Yes |
| Analysis window (days) | 30 | 14 | 14 |
| Exclude recent days | 2 | 1 | 0 |
| Placements: min clicks to adjust | 30 | 20 | 20 |
| Min clicks with no sales to block | 25 | 15 | 8 |
| Projected-ACOS threshold to block | 2 × target | 1.5 × target | 1.5 × target |
| Isolate relevant terms instead of blocking them | No | Yes | Yes |
| Auto-prune parent keyword | No | No | Yes |
The progression is the same across all three columns: the further right, the faster the workflow reacts and the more new ground it explores — at the cost of more spend.
Applying a mode overwrites your manual settings
Applying a mode rewrites every workflow variable with that mode's values: bid limits, analysis window, placements, discovery, negatives and auto-pruning. Not only the parameters in the table above.
Your term lists (own brand, competitors, exclusions, types to harvest and match types) and your campaign selection are always kept. Epinium asks you to confirm before applying the change.
This cuts both ways, usefully: if you have been trying out values and want to go back to a known point, press the mode again and everything returns to its profile.
Variable groups
Below the mode selector, the Variables tab groups every parameter. Each variable has its own explanation in a tooltip inside the interface:
| Group | What it governs |
|---|---|
| Campaign selection | Whether the workflow automatically takes campaigns with a target ACOS or only the ones you tick by hand |
| Analysis | How many days of history are looked at to decide, and how many recent days are excluded because of Amazon's attribution lag |
| Bid limits | Min and max bid at workflow level (0 = no limit) |
| Placements | Minimum clicks before adjusting a placement, and the relative weight of Product Pages |
| Harvesting | Discovery level, AI generation, brand and competitor lists, search-term types to harvest, match types and competitor-ASIN targeting |
| Negatives | How many clicks with no sales are needed to block a term, the projected-ACOS threshold, and whether relevant terms are isolated instead of blocked |
| Discovery | When new keywords are needed (keyword floor and cap, spend window) and how many are created per run |
| Pruning | When an exhausted parent keyword is archived and how many at most per run |
Bid limits stack with the campaign's
Min and max bid exist in two places: on the workflow (affecting all its campaigns) and in each campaign's AI Settings. When both are set, the more restrictive one applies: the effective max bid is the lower of the two, and the effective min bid the higher.
Changing mode safely
- Launch the workflow in Dry run (test mode) with the current mode and review the decisions.
- Change mode and simulate again: compare what it would do differently.
- When you are happy with the result, run it for real.
- If you overshoot on aggressiveness, step down a mode — the next cycle already applies the new profile.
Neither changing mode nor editing variables touches anything in Amazon on its own: the changes reach your campaigns on the next run.
Next steps
- Smart Campaigns Optimizer — what it decides on each cycle and which campaigns it acts on
- What the workflow has done — check the effect of the mode you picked
- Per-campaign AI Settings — objective and limits for a specific campaign
- Workflows — Variables tab, scheduling and history