📗 This content is the Smart Bidding Ninja Script user manual, the same one the script writes in the «Manual» tab of its spreadsheet. Here you can search it, listen to it, highlight it and save it to favourites.
The suggestion system audits your scenarios automatically and warns you when it detects that a campaign has spent a long time under a scenario that is probably not the right one. It changes nothing without your permission: it suggests, you decide.
What it analyses
For each managed campaign, the script reviews its last 30 days of runs under the current scenario and looks for two specific patterns:
- 🔻 SCENARIO TOO AGGRESSIVE: if 70% or more of the informative runs produced budget DECREASES, the scenario is probably pushing too hard and the campaign cannot sustain the pace. Suggestion: one level more conservative (e.g. .ninja2 → .ninja1).
- 🔺 SCENARIO TOO CONSERVATIVE: if 70% or more of the informative runs were INCREASES (or blocks caused by .max caps) and the metric is healthy, the scenario falls short. Suggestion: one level more aggressive (e.g. .neutro → .ninja1).
Confidence levels
Not all suggestions carry the same weight. We distinguish two levels depending on how much data backs the pattern:
| Level | Requirements | Where it appears |
|---|---|---|
| HIGH | 30+ days in the same scenario, pattern met in ≥70% of informative runs, minimum 7 samples | Suggestions tab + MONTHLY email |
| MEDIUM | 14+ days in the same scenario, pattern met but with fewer samples | Suggestions tab only (not sent by email) |
💡 The suggestions in the monthly email are always HIGH confidence. We only put statistically solid things in your inbox — when a suggestion reaches your email, we want it to be actionable beyond doubt.
How to apply a suggestion
- Open the "Suggestions" tab.
- In the "Apply" column there are checkboxes next to each pending suggestion. Tick the ones you want to execute.
- Wait for the next run of the script (at most 1 hour with an hourly trigger, or until the next nightly run).
- The script will remove the current scenario label and apply the suggested label in Google Ads. Automatically.
- Intra-day mode will detect the label change and, within the next hour, will recalculate the target according to the new scenario.
🔄 If you change your mind before it runs, untick the checkbox. The script only applies the ones that are ticked at the moment of the run.
Colour code of the tab
| Colour | Status | Meaning |
|---|---|---|
| Light green | PENDING - HIGH | Solid suggestion, waiting to be applied |
| Yellow | PENDING - MEDIUM | Early signal, waiting to be applied (not sent by email) |
| Light grey | APPLIED | Already applied, kept for 30 days as history |
Campaigns that do NOT get suggestions
- Campaigns with an active auto-rollback (DETECTADO_CAMBIO_MANUAL or REVERTIDO). The system is already intervening through another mechanism — we do not want contradictory signals.
- Campaigns with fewer than 7 runs in the last 14 days under the current scenario. Without enough data, we cannot suggest anything well founded.
- Campaigns in the .lanzamiento scenario: it is assumed to be a temporary decision by the user for the learning phase, so no automatic changes are suggested.
Analysis frequency
The Suggestions tab is regenerated once a day during the nightly cycle. Suggestions that stop being detected (because the pattern no longer holds) are kept for 3 days before being deleted, in case the user had already seen them and does not want to lose track.
Suggestions from monthly budget drift
Besides the behaviour-pattern suggestions, the script generates extra suggestions when it detects that the month's spend is clearly off course:
- OVERSPEND (projection > 100% of the monthly budget): the script proposes raising the scenario by one level on up to 3 campaigns, prioritising those that have spent the most this month and those on the least demanding scenario (room to tighten). The goal: appear in fewer auctions, slow the spending pace, reach the end of the month on budget.
- UNDERSPEND (projection < 90%): it proposes lowering the scenario by one level on up to 3 campaigns, prioritising those with the best performance over the last 14 days. The goal: release the brake where things do work and use the available budget to get more volume.
- These suggestions are always MEDIUM confidence (yellow in the tab). Reason: they are based on a linear projection of the month, not on 30-day statistical patterns.
- Unlike behaviour suggestions, budget ones are sent in the WEEKLY email on Mondays (they do not wait for the monthly one). Reason: every week counts within the month, and urgent feedback lets you react in time.
⚠️ This feature is ONLY active if you have the monthly budget set in the Config tab with "Yes" and a valid amount. If the monthly budget is empty or "No", the script generates no budget suggestions. The monthly email will include a subtle reminder to switch it on if it is not.
Other protections of the system: budget suggestions are NOT generated during the first 3 days of the month (the projection is unreliable that early), NOR on campaigns that already have a PENDING behaviour suggestion (that one prevails), NOR on campaigns with an active auto-rollback, NOR on campaigns in .lanzamiento.
Suggestions cross-checked with targets (v43.46+)
If you have a CPA/ROAS target defined (in Config or through .cpa/.roas labels on the campaign), the detector also runs a cross-check against the observed deviation. This is the most informative suggestion because it combines the adjustment pattern with the distance to the target:
- 🔺 RAISE SCENARIO (pattern + target): the campaign has spent ≥14 days below the target CPA (or above the target ROAS) AND the budget adjustments have been predominantly upwards, with no sign of the pattern slowing down. The system has been giving it room and the campaign keeps over-delivering. There is room to tighten.
- 🔻 LOWER SCENARIO (pattern + target): the campaign has spent ≥14 days above the target CPA (or below the target ROAS) AND the budget adjustments have been predominantly upwards but performance is not improving. The pressure is not bearing fruit, it is worth easing off.
📌 The "Reason" column of the Suggestions tab makes the difference clear: "pattern + target: RAISE scenario" for the cross-checked ones, "pure pattern:" for the classic ones. Anti-spam: once a cross-checked suggestion has been issued for a campaign, it is not issued again for 14 days even if the pattern persists (the control is automatic).
If a campaign has no target defined (neither in the account nor in a label), only the classic pure-pattern detector is evaluated. The cross-checked suggestion is NOT activated for that campaign — with no target there is no distance to measure. This preserves backwards compatibility: MANUAL accounts without targets keep working exactly as before.