> For the complete documentation index, see [llms.txt](https://docs.sellermagnet.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sellermagnet.com/ppc-management-beta/ppc-manager-overview/ppc-ai-optimizer.md). # PPC AI Optimizer {% hint style="info" %} **Difficulty:** ๐Ÿ”ด Advanced ยท **Reading time:** \~15 min {% endhint %} {% hint style="success" icon="rocket" %} **Open this page in your dashboard:** [**Go to PPC AI Optimizer \[BETA\] โ†’**](https://dashboard.sellermagnet.com/ppc/ai-optimizer) {% endhint %} {% hint style="warning" icon="triangle-exclamation" %} **BETA:** The PPC Manager is currently in beta. Some features may behave unexpectedly or change based on user feedback. Please report any issues at . {% endhint %} ## ๐Ÿ“‹ Overview The **AI Optimizer** (also called **AI Autopilot**) is SellerMagnet's most advanced PPC feature. It continuously analyzes your campaign data and generates smart recommendations : from keyword-level bid adjustments to campaign budget reallocation and anomaly detection. Depending on your automation mode, these suggestions are applied automatically or sent to your **Approval Center** for review. The AI Optimizer is not a black box. Every recommendation comes with a **confidence score**, a **human-readable explanation**, and a **projected impact estimate** so you always understand *why* a change is suggested and *what* to expect if you approve it. {% hint style="info" icon="circle-info" %} **What "Autopilot" means here.** "AI Autopilot" refers to recommendations the AI **applies automatically**, the behavior you get in **Auto-Apply** mode. In **Manual Review** mode nothing is applied on its own: suggestions wait in the [Approval Center](/ppc-management-beta/ppc-manager-overview/ppc-approval-center.md) until you approve or reject them. The mode is set by your **Approval Mode** in [PPC Settings](/ppc-management-beta/ppc-manager-overview/ppc-settings.md) (Manual Review / Auto-Apply / Paused). When a change is applied automatically, it is recorded in the [Activity Log](/ppc-management-beta/ppc-manager-overview/ppc-activity-log.md) with **AI Autopilot** as its source, so you can always tell auto-applied changes apart from ones you approved manually. Whichever mode you choose, every change still passes through your [safety guardrails](/ppc-management-beta/ppc-manager-overview/ppc-settings.md#safety-guardrails) (Min/Max Bid and daily change limits) on its way to Amazon. {% endhint %} > **Prerequisites:** Before using the AI Optimizer, make sure you have completed the setup on the [**PPC Settings**](/ppc-management-beta/ppc-manager-overview/ppc-settings.md) page, specifically your Target ACoS (Advertising Cost of Sale), bid boundaries, and approval mode. The AI uses those values as guardrails for every recommendation it generates. *** ## ๐ŸŽฅ Video Walkthrough {% hint style="warning" %} **Video tutorial coming soon.** A full guided walkthrough of this feature will be available here shortly. {% endhint %} *** ## Strategy Selection Choose how aggressively the AI should optimize. Your strategy selection tells the AI **what to prioritize** when making trade-offs between impressions, clicks, sales volume, ACoS, and profit. | Strategy | Description | Prioritizes | Best For | | ------------------------- | ------------------------------------------------------- | ------------------------------------------------ | ---------------------------------------------------------------------------------------- | | **Balanced** (default) | Equal weight on growth and profitability | Even balance of sales volume and ACoS control | Most sellers, good all-around starting point | | **Aggressive / Growth** | Maximize impressions, clicks, and sales volume | Visibility and market share, even at higher ACoS | Product launches, category expansion, ranking pushes | | **Conservative / Profit** | Minimize ACoS and protect margins | Low-waste spending, pausing underperformers | Mature products with stable demand, margin-sensitive sellers | | **Profit Maximizer** | Maximize net profit per keyword (revenue minus ad cost) | Absolute profit, not percentages | Sellers who know their exact COGS (Cost of Goods Sold) and want dollar-optimized returns | Select your strategy in the hero section at the top of the page. The AI adjusts its optimization logic, bid aggressiveness, and keyword management behavior accordingly. ### Strategy Comparison: Detailed Breakdown | Behavior | Balanced | Aggressive / Growth | Conservative / Profit | Profit Maximizer | | -------------------------------- | ----------------------------- | ------------------------------------------------ | ----------------------------------------- | ------------------------------------------------ | | **Bid increase frequency** | Moderate | High | Low | Variable (profit-dependent) | | **Bid decrease frequency** | Moderate | Low | High | Variable (profit-dependent) | | **Keyword pause threshold** | ACoS > 2x target for 14+ days | Rarely pauses | ACoS > 1.5x target for 7+ days | Negative profit for 14+ days | | **Keyword harvesting** | Active | Very active | Selective | Selective (high-margin terms only) | | **Budget reallocation** | Balanced across campaigns | Shifts toward highest-impression campaigns | Shifts toward lowest-ACoS campaigns | Shifts toward highest-profit campaigns | | **Negative keyword suggestions** | Moderate | Conservative (avoids blocking potential traffic) | Aggressive (blocks anything above target) | Profit-based (blocks anything unprofitable) | | **Risk tolerance** | Medium | High | Low | Medium | | **Typical ACoS outcome** | Near target | Above target (by design) | Below target | Variable, optimizes for dollars, not percentages | > **Pro Tip:** You can change your strategy at any time. The AI re-evaluates all campaigns within one optimization cycle (typically within 24 hours). A common approach: start with **Aggressive** during a product launch to build keyword data, then switch to **Balanced** once you have 30+ days of data, and finally move to **Conservative** or **Profit Maximizer** for mature products. {% hint style="warning" %} **Important:** Switching strategies does **not** undo previous changes. If the Aggressive strategy increased bids on 50 keywords yesterday, switching to Conservative today does not automatically roll those bids back. The Conservative strategy will evaluate those keywords fresh and may suggest decreases in the next cycle. {% endhint %} *** ## AI Status Panel The status panel at the top of the page gives you a real-time view of the AI's operational state. | Indicator | Description | What to Watch For | | --------------------------------- | ------------------------------------------------- | -------------------------------------------------------- | | **Current Mode** | Manual Review / Auto-Apply / Paused | Ensure this matches your intended level of automation | | **Active Strategy** | Which optimization strategy is currently selected | Verify this aligns with your current business goals | | **Last Optimization** | Timestamp of the most recent AI analysis run | If stale (24+ hours), check sync schedule and API status | | **Next Scheduled Run** | When the AI will analyze data next | Helpful for planning review sessions | | **Pending Actions** | Number of recommendations awaiting your decision | High numbers mean the Approval Center needs attention | | **Actions Applied (Last 7 Days)** | Count of changes executed in the past week | Gives you a pulse on AI activity level | | **System Health** | Green (healthy), Yellow (delayed), Red (error) | Yellow or Red requires attention, see Troubleshooting | {% hint style="info" %} **Quick Win:** Bookmark a time each morning to check the AI Status Panel. A 2-minute glance tells you whether everything is running smoothly or if something needs attention. {% endhint %} *** ## ๐Ÿ“Š Portfolio Health Score The AI Optimizer summarizes your account's overall PPC health as a single **0--100 score**, color-coded from red (needs work) through amber and blue to green (healthy). It is an **equal-weighted blend (25% each)** of four sub-scores: | Sub-score | What it measures | | ------------------ | --------------------------------------------------------- | | **ACoS vs target** | How your achieved ACoS compares to your Target ACoS | | **Conversion** | Your portfolio conversion rate (CVR) | | **Efficiency** | The share of your spend that is profitable | | **Data quality** | How much of your portfolio has enough click data to judge | The score is shown alongside (and is distinct from) the per-keyword **average score** and the period **momentum** trend. ### The Score Is Graded *Relative* to Your Target ACoS This is the single most important thing to understand about the score: the ACoS component is graded **against your Target ACoS, not against an absolute benchmark.** That means a healthy, profitable account can read "Critical" purely because its Target ACoS is set far below what it is actually achieving. {% hint style="danger" %} **A "Critical" score is not always bad, check your Target ACoS first.** If your Target ACoS is set implausibly low (for example 5%) while your account is genuinely achieving a healthy 11%, the score grades every profitable keyword as "too expensive" and drags the whole portfolio down. The AI surfaces a warning in exactly this case, telling you the target you set versus the ACoS you are achieving and suggesting you raise the target if that low value was not intentional. **Before reacting to a low score, confirm your Target ACoS reflects your real margins.** {% endhint %} {% hint style="info" icon="lightbulb" %} **Read the score with confidence and data volume in mind.** A score built on thin data is less trustworthy than one built on thousands of clicks. The individual recommendations carry their own **confidence** levels and click-count notes, weigh those alongside the headline score rather than treating the number as an absolute verdict. {% endhint %} *** ## Smart Recommendations ![PPC Approval Center: review AI suggestions before applying](/files/ghZ5dsuNnNuCA5IOZrI5) ![PPC AI Optimizer - Automated campaign optimization](/files/RD3WgV2Zv1627zJsG05F) The AI generates several types of recommendations, each shown as an actionable card in the Approval Center (Manual Review mode) or executed automatically (Auto-Apply mode). ### Recommendation Types | Type | Description | Typical Frequency | | -------------------------- | ---------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------- | | **Bid increases** | Raise bids on high-converting, profitable keywords that can capture more traffic | Daily to weekly | | **Bid decreases** | Lower bids on high-ACoS or underperforming keywords to reduce waste | Daily to weekly | | **Keyword pauses** | Pause keywords with consistently poor performance (high spend, zero or near-zero conversions) | Weekly | | **Keyword harvesting** | Promote high-performing search terms from auto campaigns into manual campaigns as exact or phrase match keywords | Weekly | | **Budget reallocation** | Shift budget from low-ROI campaigns to high-ROI campaigns to maximize overall portfolio efficiency | Weekly to biweekly | | **Negative keywords** | Block search terms that generate clicks but no conversions, wasting ad spend | Weekly | | **Placement adjustments** | Modify Top of Search or Product Pages bid modifiers based on placement-level performance data | Biweekly | | **Campaign state changes** | Enable or disable entire campaigns based on aggregate performance trends | Rare, only when a campaign is clearly failing or succeeding | ### Recommendation Card Details Each recommendation card displays: * **Type icon** and descriptive title (e.g., "Increase bid for 'wireless charger stand'") * **Affected items count** (e.g., "12 keywords across 3 campaigns") * **Estimated impact:** Projected ACoS change, estimated additional sales, or estimated spend savings * **Confidence score:** Data reliability percentage (higher = more historical data supports the recommendation) * **Reasoning:** Plain-language explanation of why the AI is suggesting this change * **Action buttons** : **Approve** (apply the change), **Preview** (expand to see all affected keywords/campaigns), **Reject** (dismiss with optional feedback) > **Pro Tip:** When you reject a recommendation, use the optional feedback field to tell the AI *why*. For example: "This keyword is for a seasonal product, keep bid as-is until Q4." The AI uses rejection feedback to refine future suggestions. ### Scenario: Reviewing a Bid Increase Recommendation > You open the Approval Center and see a recommendation: **"Increase bid on 8 keywords in Campaign 'Wireless Chargers - DE'"**. You click **Preview** and see: > > The AI explains: "These keywords are converting well below your Target ACoS of 25%. Increasing bids should capture more impressions at profitable CPCs." > > You approve the recommendation. The bids are updated in Amazon Ads within 15 minutes. | Keyword | Current Bid | Suggested Bid | Change | Current ACoS | Projected ACoS | Confidence | | ---------------------- | ----------- | ------------- | ------ | ------------ | -------------- | ---------- | | wireless charger stand | 0.85 | 1.05 | +24% | 18% | 16% | 92% | | qi charger desk | 0.62 | 0.78 | +26% | 22% | 19% | 87% | | fast wireless charger | 1.20 | 1.45 | +21% | 15% | 14% | 95% | ### Estimated Savings & Growth The AI Optimizer summarizes the projected upside of its recommendations as two headline figures, **estimated savings** (mainly from bid decreases and negations that cut waste) and **estimated growth** (additional sales from scaling profitable terms). The savings figure is derived from the size of the suggested bid reductions projected forward, so you can gauge the value of acting on the decrease suggestions before you approve them. {% hint style="warning" %} **Mixed currencies are not directly comparable.** If your account spans multiple marketplace currencies, the savings and growth totals are summed across currencies and are flagged as not directly comparable. Select a single marketplace for an accurate figure. {% endhint %} ### Duplicate Keyword Detection The AI can detect the **same keyword running across multiple campaigns or ad groups**, where your own campaigns end up bidding against each other for the same term. When it finds duplicates, it offers a one-click action to **pause the duplicates and keep the best-performing instance enabled** (the confirmation reads *"N duplicate keywords will be paused. The best-performing instance stays enabled."*). {% hint style="info" icon="circle-info" %} **Why this saves money:** When the same term lives in two campaigns, you can negate a wasteful query in one and still bleed spend on it through the other. Consolidating to the single best-performing instance closes that gap and removes self-competition across your portfolio. {% endhint %} *** ## Performance Insights The AI generates plain-language insights based on your campaign data. These are not actionable recommendations, they are **observations** designed to help you understand what is happening and why. Categories of insights: * **Key Findings:** e.g., "Your top 5 keywords drive 60% of total sales but only account for 35% of spend, these are your most efficient performers" * **Optimization Opportunities:** e.g., "3 keywords have ACoS above 50% and have spent more than 20 combined with zero conversions in the last 14 days" * **Anomalies Detected:** e.g., "CTR dropped 25% on Monday across Campaign 'Bluetooth Speakers - US', investigate potential listing suppression or competitor activity" * **Trend Analysis:** Yesterday vs. today, week-over-week, and month-over-month changes in key metrics (spend, sales, ACoS, impressions, CTR) * **Competitive Signals:** e.g., "Average CPC for 'portable speaker' has increased 18% in the last 7 days, competitor bidding activity may be rising" {% hint style="info" icon="lightbulb" %} **Tip:** Performance Insights update after each optimization run. Read them before reviewing recommendations, they provide the context you need to make faster approve/reject decisions. {% endhint %} *** ## Keyword Optimization Table A detailed, sortable, filterable table of all AI-suggested bid changes for the current cycle. | Column | Description | | ------------------------ | ------------------------------------------------------------------ | | **Keyword** | Keyword text and match type (Exact, Phrase, Broad) | | **Campaign / Ad Group** | Parent campaign and ad group | | **Current Bid** | Your current bid in Amazon Ads | | **Suggested Bid** | AI-recommended bid | | **Change** | Percentage and absolute change (e.g., +15%, +0.12) | | **Current ACoS** | ACoS for this keyword over the lookback period | | **Projected ACoS** | Estimated ACoS after the bid change is applied | | **Clicks (Period)** | Total clicks during the lookback period | | **Conversions (Period)** | Total conversions during the lookback period | | **Confidence** | AI confidence level (0--100%) based on data volume and consistency | | **Reason** | Human-readable explanation of the recommendation | | **Status** | Pending, Approved, Rejected, Applied | **Bulk actions:** Select multiple keywords using checkboxes and approve or reject them in batch. You can also filter by recommendation type, confidence level, campaign, or ACoS range to focus on the most impactful suggestions first. > **Best Practice:** Sort by **Confidence (descending)** and approve high-confidence suggestions first. These are backed by the most data and carry the lowest risk. Review lower-confidence suggestions more carefully, they may be based on limited clicks or short time periods. *** ## Budget Allocation Suggestions The AI analyzes your entire campaign portfolio and recommends budget shifts to maximize overall efficiency. * **Current Allocation:** Pie or bar chart showing how your daily budget is distributed across campaigns today * **Recommended Allocation:** Where the AI suggests reallocating spend, with specific dollar amounts * **Projected Impact:** Estimated sales increase, ACoS improvement, or profit gain from the reallocation * **Reallocation Details:** Table showing each campaign's current budget, suggested budget, and the net change **Actions:** View Details, Apply Recommendation, Dismiss ### Scenario: Budget Reallocation Across EU Campaigns > You have 5 campaigns across DE and FR with a total daily budget of 200. > > The AI projects that this reallocation would increase total portfolio sales by 12% while reducing overall ACoS from 24% to 21%. | Campaign | Current Budget | Suggested Budget | Change | Rationale | | ---------------- | -------------- | ---------------- | ------ | ----------------------------------------------- | | DE - Exact Match | 60 | 80 | +20 | Highest ROAS campaign (4.2x), limited by budget | | DE - Broad Match | 50 | 35 | -15 | High ACoS (38%), diminishing returns | | DE - Auto | 30 | 25 | -5 | Sufficient for discovery; excess goes to Exact | | FR - Exact Match | 40 | 45 | +5 | Growing market, strong conversion rate | | FR - Auto | 20 | 15 | -5 | Low volume, adequate at reduced budget | {% hint style="warning" %} **Important:** Budget reallocation suggestions are based on recent performance trends. If you are planning a promotion, product launch, or seasonal push for a specific campaign, consider that context before applying the recommendation. The AI does not know about your future marketing plans. {% endhint %} *** ## Anomaly Detection The AI continuously monitors your campaign data for unusual patterns that may indicate problems requiring immediate attention. | Anomaly Type | Example | Typical Severity | | ------------------------ | ---------------------------------------------------------------------------- | ---------------- | | **CTR drop** | Sudden click-through rate decrease on a campaign or keyword set | Medium to High | | **Spend surge** | Unexpected spike in daily spend (e.g., 2x normal level) | High to Critical | | **Impression drop** | Significant decrease in ad impressions (possible listing or keyword issue) | Medium | | **Budget exhaustion** | Campaign hitting daily budget cap before midday | High | | **Conversion rate drop** | Sharp decline in conversion rate (possible listing, pricing, or stock issue) | High | | **Quality issues** | Impression share dropping for key terms (competitors outbidding you) | Medium | | **CPC spike** | Average CPC jumps significantly (possible competitor bid war) | Medium | Each anomaly shows: * **Severity:** Critical (red), High (orange), Medium (blue), Low (gray) * **Timestamp** and affected entity (campaign, keyword, or ad group) * **Metric details:** What changed, by how much, and compared to what baseline * **Suggested remediation:** AI-recommended action to address the anomaly * **Approve / Dismiss** buttons {% hint style="danger" %} **Critical anomalies require immediate attention.** A spend surge or budget exhaustion flagged as Critical means your advertising budget may be draining faster than expected. Review these within hours, not days. {% endhint %} {% hint style="info" icon="lightbulb" %} **Tip:** Not all anomalies are bad. A sudden impression surge after launching a Lightning Deal is expected behavior. Use the **Dismiss** button to tell the AI this anomaly is not a concern, which helps it calibrate future detection sensitivity. {% endhint %} *** ## ๐Ÿ’ก Opportunities: Products Worth Advertising Beyond optimizing existing campaigns, the AI Optimizer surfaces **Opportunities**, products in your catalog that are not yet being advertised (or are under-advertised) but look like strong candidates for a campaign. Each opportunity is scored **0--100** and sorted into a tier: | Tier | Meaning | | ---------- | ------------------------------------------- | | **STAR** | Strongest candidates, advertise these first | | **GROWTH** | Solid candidates with room to scale | | **TEST** | Worth a small, controlled test | | **SKIP** | Not recommended right now | Each opportunity card shows the supporting evidence, sales rank, margin, profit per unit, units sold, sales velocity, and days of cover, along with a projected outcome (suggested budget, estimated sales, ACoS, implied ROAS, orders, and bid) and a **confidence** level. Products are automatically excluded with a clear reason when they are not eligible, for example: *Already advertised*, *Listing inactive*, *Missing price*, *No stock*, *No organic sales (30d)*, or *Negative margin*. You can also export the opportunities list (CSV or Excel) with the tier, score, suggested template, suggested budget, and the daily sales/orders/ACoS estimates. ### Pre-Flight Estimate (Before You Launch) When you choose to create a campaign from an opportunity, a **Pre-flight estimate** modal opens *before* anything is created, so you see the economics first. The pre-flight estimate covers: * **Economic impact** : projected daily spend, the 30-day spend range, projected 30-day revenue lift, implied ROAS, and estimated orders and clicks per day * **Profitability** : projected net profit over 30 days (revenue lift ร— product margin, minus ad spend) and a comparison of your **break-even ACoS vs Target ACoS**, with a headroom badge that warns when your target sits above break-even * **Risk and confidence** : Low / Medium / High labels, plus the evaluation window the estimate is based on * **Campaign setup preview** : the template, daily budget, default bid, and Target ACoS that would be used * **Keyword seeds** : the proposed keywords (a mix of Amazon recommendations and proven winners) * **Conflict detection** : keywords that already appear in a sibling campaign are flagged ("keyword bleed") so you do not create internal competition From the modal you can **Continue to review**, **Pick a different preset**, or **Cancel**. {% hint style="success" %} **Why the pre-flight step matters:** It turns "should I advertise this product?" into a numbers-backed decision. Check that the projected net profit is positive and that your Target ACoS sits *below* break-even (headroom) before continuing, and watch the keyword-bleed flags so a new campaign does not cannibalize an existing one. {% endhint %} *** ## Optimization History A time-series chart showing the cumulative impact of AI optimizations over time: * **Total Spend** (line): How your daily/weekly spend has changed * **Total Sales** (line): Revenue attributed to PPC campaigns * **ACoS Trend** (line): Advertising cost of sale over time * **Orders** (bars): Number of orders per period * **AI Actions Applied** (markers): Dots on the timeline showing when the AI made changes This visualization helps you answer the critical question: **"Is the AI making my campaigns better over time?"** ### Before/After: AI Optimizer Impact Examples #### Example 1: Mature Product Portfolio (Conservative Strategy, 60-Day Window) | Metric | Before AI Optimizer | After 60 Days | Change | | --------------------------------------- | ------------------- | ---------------------- | --------------------- | | Average ACoS | 32% | 22% | -10 percentage points | | Monthly Ad Spend | 4,200 | 3,600 | -14% | | Monthly PPC Sales | 13,100 | 16,400 | +25% | | Wasted Spend (zero-conversion keywords) | 890/month | 210/month | -76% | | Active Keywords | 620 | 485 (135 paused by AI) | Quality over quantity | #### Example 2: Product Launch (Aggressive Strategy, 30-Day Window) | Metric | Before AI Optimizer | After 30 Days | Change | | ---------------------------- | ------------------- | --------------------------- | ------------------------------ | | Average ACoS | 55% | 42% | -13 percentage points | | Daily Impressions | 8,500 | 22,000 | +159% | | Monthly PPC Sales | 1,200 | 4,800 | +300% | | Keywords Harvested from Auto | 0 | 47 new exact match keywords | Expanded manual targeting | | Organic Rank (main keyword) | Page 4 | Page 1, position 8 | PPC-driven ranking improvement | {% hint style="info" icon="circle-info" %} **Note:** Results vary based on product category, competition level, and starting campaign quality. The examples above are based on aggregated anonymized data from SellerMagnet users. Your results may differ. {% endhint %} *** ## Configuration These settings are also accessible from the [**PPC Settings**](/ppc-management-beta/ppc-manager-overview/ppc-settings.md) page. Changing them here has the same effect. | Setting | Description | Options | | -------------------------- | -------------------------------------------------------------- | ---------------------------------------------- | | **Strategy** | Optimization approach: affects all AI decisions | Balanced, Aggressive, Conservative, Profit Max | | **Lookback Period** | How far back the AI looks when analyzing performance | 7, 14, 30, 60, or 90 days | | **Run Schedule** | How often the AI analyzes data and generates recommendations | Every 6 hours, Daily, Every 3 days, Weekly | | **Blackout Dates** | Days when no automated changes should be made | Date picker | | **Daily Change Limit** | Maximum percentage a single bid can change per day | 5%--50% (default: 20%) | | **Minimum Data Threshold** | Minimum clicks before the AI will suggest changes to a keyword | 5--100 clicks (default: 10) | > **Best Practice:** Set the **Minimum Data Threshold** to at least 10 clicks. Suggesting bid changes on keywords with only 2--3 clicks is statistically unreliable and can lead to premature pauses of potentially good keywords. *** ## Recommended Strategy by Seller Type | Seller Profile | Recommended Strategy | Lookback | Run Schedule | Notes | | ---------------------------------------------------------------------------- | --------------------------------------------- | ----------- | ------------- | ----------------------------------------------- | | **New seller, first product** | Aggressive | 7 days | Daily | Prioritize data collection and visibility | | **Established seller, stable catalog** | Balanced or Conservative | 14 days | Daily | Focus on efficiency and incremental improvement | | **High-volume seller (100+ daily orders)** | Profit Maximizer | 14--30 days | Every 6 hours | Dollar-optimized with frequent re-evaluation | | **Seasonal seller (holiday/Q4 focus)** | Aggressive pre-season, Conservative in-season | 7--14 days | Daily | Switch strategies as the season progresses | | **Agency managing 10+ clients** | Per-client (typically Balanced start) | 14 days | Daily | Use Manual Review for all client accounts | | **Private label launching new ASIN (Amazon Standard Identification Number)** | Aggressive for 30--60 days, then Balanced | 7 days | Daily | Build keyword data fast, then optimize | ***
โš ๏ธ Common Mistakes to Avoid | Mistake | Why It Hurts | How to Fix | | ------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------- | | Never reviewing AI suggestions (leaving hundreds pending) | The AI cannot learn your preferences; suggestions become stale and irrelevant | Set a daily or weekly review cadence; approve or reject every suggestion | | Using Aggressive strategy on a low-margin product | AI prioritizes volume over profitability, you may drive sales at a loss | Switch to Conservative or Profit Maximizer for low-margin products | | Setting Lookback Period too short (7 days) on stable products | Short windows amplify noise, a single bad day skews all suggestions | Use 14--30 days for products with consistent demand | | Ignoring anomaly alerts | A spend surge or CTR drop can waste hundreds in hours if left unchecked | Enable notifications and respond to Critical/High anomalies within 24 hours | | Approving everything without reading the reasoning | Teaches the AI that all suggestions are good, even ones based on limited data | Read the confidence score and reasoning; reject low-confidence suggestions you are not sure about | | Running Auto-Apply without setting a Daily Change Limit | Bids can swing dramatically in a single day, destabilizing campaign performance | Always set a Daily Change Limit (20% is a safe default) before enabling Auto-Apply | | Expecting results in 48 hours | The AI needs at least 1--2 optimization cycles (and sufficient historical data) to generate meaningful recommendations | Give the AI at least 14 days of campaign data and 2--3 optimization cycles before evaluating its effectiveness |
\## ๐Ÿ”ง Troubleshooting ### AI Not Generating Recommendations | Symptom | Likely Cause | Solution | | --------------------------------------- | ---------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | | "Insufficient data" message | Less than 14 days of campaign data available | Wait for more data to accumulate; ensure sync is running | | Zero pending actions for 7+ days | Approval mode may be set to Paused | Check PPC Settings, switch to Manual Review or Auto-Apply | | Recommendations only for some campaigns | Other campaigns have too few clicks to meet the Minimum Data Threshold | Lower the threshold or wait for more traffic; consider increasing bids to generate clicks | | Strategy card shows "Inactive" | AI Optimizer may be disabled at the account level | Check the main toggle at the top of the AI Optimizer page | ### Unexpected AI Behavior | Symptom | Likely Cause | Solution | | ------------------------------------------------- | -------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- | | AI suggesting bids above your comfort level | Max Bid in Settings is too high for your risk tolerance | Lower the Max Bid ceiling in PPC Settings | | AI pausing keywords you want to keep | Keywords have ACoS far above your Target ACoS and the AI's pause threshold | Either increase your Target ACoS or manually un-pause the keyword and reject the suggestion (the AI will learn) | | AI not suggesting changes for a specific campaign | Campaign may be excluded, paused in Amazon, or have insufficient data | Verify the campaign is active in Amazon Ads and has enough clicks in the lookback period | | Projected impact estimates seem too optimistic | Projections are based on historical trends and assume stable competition | Treat projections as directional estimates, not guarantees, actual results may vary by 20--30% | ### System Health Issues | Symptom | Likely Cause | Solution | | ---------------------------------------------- | --------------------------------------------------------------------------- | -------------------------------------------------------------------------------- | | System Health shows Yellow (delayed) | Sync delay or high processing load | Usually resolves within a few hours; check API Status in PPC Settings | | System Health shows Red (error) | API connection issue, expired token, or Amazon Ads outage | Check PPC Settings > API Status for error details; reconnect if token is expired | | "Last Optimization" timestamp is 48+ hours old | Run schedule may be set to Weekly, or a sync error is blocking the pipeline | Verify sync schedule and API status; click Sync Now to force a refresh | *** ## โ“ FAQ
How long does it take for the AI Optimizer to show results? Most sellers see meaningful impact within 2--4 weeks. The AI needs at least 14 days of historical data to generate reliable suggestions. The first 1--2 weeks are a "learning phase" where the AI is calibrating to your campaigns.
Can the AI Optimizer make things worse? The guardrails (bid boundaries, daily change limits, blackout dates) let you prevent destructive changes. In Manual Review mode, nothing changes without your approval. In Auto-Apply mode, the Daily Change Limit caps how much any single bid can move per day. That said, no optimization tool is perfect, monitoring the Optimization History chart regularly is recommended.
Does the AI learn from my approve/reject decisions? Yes. Every approve or reject decision feeds back into the AI's model. Over time, the AI generates suggestions that increasingly align with your preferences. Consistent feedback (especially rejections with reasons) accelerates this learning.
Can I run different strategies for different campaigns? Currently, the strategy is set at the **profile level** (one strategy per marketplace). Campaign-level strategy overrides are on the roadmap. For now, if you need different strategies for different campaigns, focus on using the Keyword Optimization Table to selectively approve or reject suggestions per campaign.
What happens during Blackout Dates? The AI still runs analysis and generates recommendations, but no changes are applied (even in Auto-Apply mode). Recommendations accumulate in the Approval Center and can be reviewed and applied after the blackout period ends.
How does the Profit Maximizer strategy differ from Conservative? Conservative minimizes ACoS (percentage). Profit Maximizer maximizes absolute profit (dollars). A keyword with 30% ACoS but 500 in monthly attributed sales might be kept or increased by Profit Maximizer (because it generates strong profit) but decreased by Conservative (because 30% ACoS exceeds a typical conservative target). Profit Maximizer requires accurate COGS data for best results.
Can I undo a change the AI made? Yes. In the Optimization History, you can view every change the AI has made. Each change has a **Revert** option that restores the previous bid or setting. Note that reverting does not undo the market impact, if a bid was increased for 3 days before reverting, the impressions and spend from those 3 days are not refundable.
How does keyword harvesting work? The AI monitors search terms in your auto campaigns that are converting well (high CTR, good ACoS, multiple conversions). It then suggests adding these search terms as exact or phrase match keywords in your manual campaigns, giving you more control over bids for proven performers. The original auto campaign continues to discover new search terms.
Is there a limit to how many recommendations the AI generates per cycle? There is no hard limit, but the AI prioritizes recommendations by expected impact. In a typical cycle, you might see 10--50 suggestions for an account with 20 campaigns and 500 keywords. Larger accounts may see more.
*** ## Use Case Scenarios ### Scenario 1: Managing 500 Keywords Across 3 EU Marketplaces You sell kitchen accessories on Amazon DE, FR, and IT. You have 500 active keywords across 12 campaigns. **Challenge:** Manually reviewing bids for 500 keywords across 3 markets is a full-time job. Different marketplaces have different CPCs, conversion rates, and competition levels. **Solution:** {% stepper %} {% step %} ### Balanced Set strategy to **Balanced** (good default for multi-market management). {% endstep %} {% step %} ### 14 days Set lookback to **14 days** per marketplace. {% endstep %} {% step %} ### Auto-Apply Enable **Auto-Apply** with a **20% Daily Change Limit**. {% endstep %} {% step %} ### Optimization History Review the **Optimization History** weekly to ensure the AI is trending your overall ACoS in the right direction. {% endstep %} {% step %} ### Budget Reallocation Use **Budget Reallocation** suggestions to shift spend toward your best-performing marketplace. {% endstep %} {% endstepper %} **Expected Outcome:** The AI handles the 500-keyword bid management across all three markets. You spend 15 minutes per week reviewing the history chart and anomaly alerts instead of 2+ hours per day manually adjusting bids. ### Scenario 2: Launching a New Product with Zero History You are launching a new product in the US marketplace. You have one auto campaign and one manual campaign with 30 seed keywords. **Challenge:** No historical data means the AI has nothing to analyze. You need traffic fast to build keyword data and organic ranking. **Solution:** {% stepper %} {% step %} ### Aggressive / Growth Set strategy to **Aggressive / Growth**. {% endstep %} {% step %} ### 7 days Set lookback to **7 days** (react quickly to early data). {% endstep %} {% step %} ### Manual Review Use **Manual Review** mode for the first 2 weeks (you want to see every suggestion while data is thin). {% endstep %} {% step %} ### Step 4 Set a higher Target ACoS (50--60%), you are investing in visibility, not optimizing for profit yet. {% endstep %} {% step %} ### Keyword Harvesting Pay special attention to **Keyword Harvesting** suggestions, the AI will surface converting search terms from your auto campaign that should be moved to manual. {% endstep %} {% endstepper %} **Expected Outcome:** Within 30 days, the AI has harvested 30--50 new keywords from your auto campaign, identified your top 10 performers, and paused any seed keywords that proved unprofitable. You switch to Balanced strategy for the next phase. ### Scenario 3: Agency Onboarding a New Client You are an agency onboarding a new client who has been running Amazon PPC for a year without optimization tools. Their current ACoS is 45% against a target of 25%. **Challenge:** The client has years of accumulated inefficiency, wasteful keywords, outdated bids, and no negative keyword strategy. You need to show results within 30 days to retain the client. **Solution:** {% stepper %} {% step %} ### Step 1 Connect the client's account and let 14 days of data sync. {% endstep %} {% step %} ### Conservative / Profit Set strategy to **Conservative / Profit**, the priority is reducing waste, not growing volume. {% endstep %} {% step %} ### Manual Review Use **Manual Review** mode, you want to approve every change on a client account. {% endstep %} {% step %} ### Bid Decreases Focus on three recommendation types: **Bid Decreases** (reduce overspending), **Keyword Pauses** (stop bleeding on zero-conversion keywords), and **Negative Keywords** (block wasteful search terms). {% endstep %} {% step %} ### Optimization History Present the **Optimization History** chart to the client weekly to demonstrate progress. {% endstep %} {% endstepper %} **Expected Outcome:** Within 30 days, ACoS drops from 45% to 28--32% through waste elimination. The client sees clear, data-backed results in the weekly reports. You switch to Balanced strategy for month two to start growing profitable sales. *** ## โžก๏ธ What's Next? {% content-ref url="/pages/xUaJFWbcCTCya95L64xQ" %} [PPC Approval Center](/ppc-management-beta/ppc-manager-overview/ppc-approval-center.md) {% endcontent-ref %} {% content-ref url="/pages/pQTsyjXdDCQAMg0Eueka" %} [PPC Analytics](/ppc-management-beta/ppc-manager-overview/ppc-analytics.md) {% endcontent-ref %}