Saddl OS · Advertising
The Optimizer in Saddl OS reads your Amazon, Noon Core Ads and Noon Minutes campaigns with the same decision logic and machine-learning layer. It turns what it finds into budget, bid, negative and harvest actions, each with a reason and a confidence level.
Saddl’s team runs the Optimizer for you, as part of Marketplace Advertising or Marketplace Management.
StrategyBalanced
Campaign budgetsBidsNegativesHarvestSignals
“Brand defence” converts at 2.1× the account average. Confirm it as a DEFENSE campaign so efficiency cuts don’t hit it?
Amazon: pushed automatically. Approved actions go to Amazon through the Ads API, within your budget and approval limits.
Push approved actionsNoon: a report the team applies. Noon advertising actions can’t be pushed automatically. The Optimizer exports the full report from Downloads, and the Saddl team applies each action in Noon by hand, then checks the effect.
Downloads · Full report (CSV)Every recommendation passes the same four steps, on every platform.
Campaigns, targets and search terms, with stock cover, contribution and the campaign’s own objective alongside.
Each campaign is measured against its objective and your strategy, not the account average. Every action gets an impact and a confidence level.
Low-stock products are held, recent changes cool down, halo effects are respected and judgment calls escalate to a person.
Amazon actions push through the Ads API within your limits. Noon actions go out as a full report the Saddl team applies by hand.
Inside the Optimizer
Each tab answers one question, with the reason and the guardrail shown beside the numbers. Sample data.
How many decisions the engine made, how much waste it found, what growth it opened and how spend is reallocated.
Guardrail Unchanged spend is shown too, so you can see what the engine chose to leave alone.
How spend is reallocated across actions
Budgets are raised, held or cut campaign by campaign, against the target for that campaign, with the impact and the engine’s confidence beside each.
Guardrail Products low on stock are held until restocked, and recent changes cool down before the next one.
Inventory hold: 2 campaign actions paused. The advertised products have under 14 days of supply. They resume automatically once restocked.
9 more held back this run: 8 in cooldown, 1 with halo active.
Escalations that need a human decision, plus period-over-period diagnostics that say what moved the result and where to look next.
Guardrail Judgment calls go to a person. They are never turned into an automatic change.
“Brand defence” converts at 2.1× the account average. Confirm it as a DEFENSE campaign so efficiency cuts don’t hit it, or dismiss to keep it on PROFITABILITY.
Bid changes by targeting type: exact keywords, product targeting, broad and phrase, and auto and category, with the signal behind each move.
Guardrail Bids move in bounded steps, and terms with too little data are held.
Exact keywordsProduct targetingBroad / phraseAuto / category
Search terms that spend without earning are proposed as negatives, with the spend they would save.
Guardrail A weak term is checked against stock, conversion and the offer before it is negated.
Search terms that convert are promoted to their own targets, so winners get their own bids and budgets.
Guardrail Only terms that pass the harvest thresholds (clicks, orders and return) are promoted.
Every recommendation from the run, exported with its reason, impact and confidence. On Noon this is the work order: the Saddl team applies each action by hand and records the result.
Guardrail Noon advertising actions are never pushed automatically. Amazon actions can be, within your limits.
Two settings shape every action, so the same numbers can call for different moves.
Account
Set for the account, for example Balanced. It tells the engine how to trade growth against efficiency.
Campaign
Each campaign carries its own objective, such as profitability or defence, and is judged against it. A brand-defence campaign isn’t cut for failing a profitability target.
Every action
What changed, why, the impact expected and how sure the engine is, so approving an action means reading one line.
Same logic. Two routes.
Amazon: pushed by API
Noon and Noon Minutes: report, applied by handNoon advertising actions can’t be pushed automatically, so the team applies them by hand. The judgment behind them is the same as on Amazon.
It is the Optimizer: one decision engine that reads your campaigns, targets and search terms and proposes budget, bid, negative and harvest actions, each with a reason, an expected impact and a confidence level. Saddl’s team runs it for you.
On Amazon, approved actions can be pushed through the Ads API, within your budget and approval limits. On Noon Core Ads and Noon Minutes, actions cannot be pushed automatically, so the Optimizer exports a full report and the Saddl team applies each action by hand.
No. The same decision logic and machine-learning layer applies to Amazon, Noon Core Ads and Noon Minutes. Only the execution route differs.
Actions on products with low supply, for example under 14 days, are held and resume automatically once restocked. A stocked-out product defers a negative-keyword candidate instead of having its keywords cut for poor performance.
The strategy is set for the account, for example Balanced, and tells the engine how to trade growth against efficiency. Each campaign also has an objective, such as profitability or defence, and is judged against that.
Judgment calls, such as whether a campaign should be protected as a core performer, are escalated for a decision. Approvals and execution permissions stay within limits agreed with you.
A bid tool adjusts bids. The Optimizer also decides budgets, negatives and harvest, checks stock and objectives before acting, explains each action and escalates the calls that need judgment. Saddl’s team then reviews the results with you.
The free audit reads your campaigns, stock and product economics together, and shows where spend is leaking and what to fix first.