Live signal in. Budget and creative decisions out.
Optimization does not wait for a weekly review call. As live performance comes in, budget shifts, creative refreshes and audience adjustments trigger automatically - and every one is logged with the reason it fired.
One optimizer, watching every campaign in real time.
Optimization runs continuously against a signal window that considers spend, click-through rate, add-to-cart rate, real conversion rate and net revenue - all at the creative level.
When a variant clears its threshold, budget flows toward it. When one drops below the floor, it is paused and a fresh variant is queued in from the generation stage.
- Signal window is per-creative, not per-campaign
- Budget shifts are bounded - no single variant runs away with all spend
- Fatigue detection triggers regeneration before performance collapses
- Every action gets a reason line - visible, auditable, and reversible
One optimization decision, shown with its reasoning.
Real decisions run continuously. This is one, unpacked - the type of thing that fires many times a day, per product, without a human touching the campaign.
Meta creative A - CTR fell from 1.8% to 0.9% over 72h at $250 spend.
Product-level floor was 1.2%. Fatigue detection triggered.
Creative A paused. Variant D queued in from generation stage.
Every step above stored on the campaign timeline, viewable any time.
A live view of every decision, and the ability to override.
Optimization is autonomous by default. It is not opaque. Everything the factory does is visible on a per-product timeline - with the signal, the threshold, and the action.
Any decision can be overridden. Pin a variant on, block a budget shift, force a creative refresh. The optimizer respects your overrides and keeps its reasoning intact around them.
- Per-product timeline of every optimization action
- Filter by decision type - budget, creative, audience, pause
- Manual override on any variant, campaign, or platform
- Weekly digest sent to your inbox if you want the summary
- 42 budget shifts across 18 products - net +$220/day flowing to top variants.
- 9 creatives retired due to CTR drop or add-to-cart floor breach.
- 5 new lookalikes activated from top-hook audiences.
- 3 fresh variants generated to replace fatigued ads.
Not just budget. Every lever that moves ROAS.
Budget rebalancing
spendSpend shifts between platforms, ad sets, and creatives based on live per-variant profitability.
Creative rotation
freshnessFatigued variants are retired and replaced from the generation stage - before performance drops off a cliff.
Audience expansion
reachLookalikes and interest expansions are activated when core audience efficiency starts to plateau.
Bid strategy tuning
biddingBid caps and target ROAS values are adjusted per campaign based on live cost efficiency.
Anomaly detection
safetySharp spend spikes, tracking gaps, or approval reversions get flagged and paused pending review.
Signal-triggered regeneration
regenerateWhen a hook stops working, generation runs a fresh set with a different angle - not a rerun of the same idea.
About how automatic optimization actually behaves.
How aggressively does it shift budget?
+
Can I stop it from touching a specific campaign?
+
What if the optimizer disagrees with me?
+
How long before I see meaningful optimization?
+
Hand over the daily calls. Keep the strategic ones.
You set the guardrails - budget floors, must-hit ROAS, priority products. The optimizer runs inside them.
