PPC Mission Control

The PPC Specialist is an autonomous Amazon Advertising optimization agent built by unybrands. Its intelligence is governed exclusively by the "SOPs by Candace" — 7 Standard Operating Procedures that define every bid change, keyword harvest, budget adjustment, and placement optimization.

What it does:
• Monitors SP, SB, and SD campaign performance continuously
• Executes bid adjustments, keyword promotions, ASIN negations, and budget changes automatically — only when confidence exceeds your defined threshold
• Flags all uncertain actions for your review before touching anything
• Logs every single action with a written rationale and SOP citation
• Adapts its strategy based on inventory levels, organic rank, and competitive positioning

What it never does:
• Act without data to back it up
• Make a decision that isn't covered by a Candace SOP (it stops and asks instead)
• Execute silently — every action is logged and traceable
• Override the confidence threshold you've set

Data sources: Amazon Advertising API (GET /ads/v3/reports), Intentwise Platform, Helium 10, Jungle Scout, Brand Analytics, Data Dive, Amazon Ads Console.

Login: Enter your email and password on the login screen. Your session persists until you click Sign Out or close the browser.

Change Password: Click the 🔑 Change Password button in the top header. Enter your current password, your new password (min. 8 characters), confirm it, and click Update. Passwords are stored locally in your browser.

User Roles:
• Admin (T3): Full access — manage users, change agent mode, edit thresholds, update methodology, approve/reject queue, view all logs, rollback actions
• Manager (T2): Can approve/reject the queue, view audit trail and methodology, access SOP schedule. Cannot manage users or change agent settings.
• Viewer (T1): Read-only access. Can view dashboard, audit trail, methodology, and schedule. Cannot approve, reject, or change anything.

Default accounts:
• ary.selener@unybrands.com — Admin
• candace.wilson@unybrands.com — Admin

⚠️ Change your password immediately after first login. Never share credentials.

The Dashboard is the first thing you see after login. It gives you a real-time snapshot of the agent's state.

Stat Cards (top row):
• Pending Approvals: Number of actions sitting in the queue waiting for your decision. If this is not zero, review the Approval Queue immediately.
• Actions Today: Total actions the agent executed today (auto-approved above confidence threshold).
• Active Alerts: KPI anomalies detected — budget issues, ACoS spikes, sudden spend drops. Investigate immediately.
• Agent Mode: Current Permission Tier (T1/T2/T3). This is critical — know what your agent is allowed to do on its own.

KPI Alerts Panel: Real-time notifications when campaign data deviates from your defined thresholds (e.g. ACoS exceeds your TACoS Max, inventory hits DOS Critical, budget runs out). Green = all clear. Yellow = warning. Red = critical — take immediate action.

Recent Agent Activity: The last 5 actions taken by the agent (approved or executed). Each shows the action, rationale, SOP reference, and confidence score. Click any entry for full detail in the Audit Trail.

Active Strategy — Live README: A plain-English summary of how the agent is currently thinking. Shows your active confidence threshold, key thresholds (TACoS, ACoS buffer, negate rules, DOS triggers), and the SOP Conflict setting. This updates automatically when you change thresholds in Settings.

Every action the agent is not confident enough to execute on its own lands here. Actions are color-coded by confidence:

• 🟢 Green border — Confidence ≥ 85% (high). The agent was close to auto-executing this. Usually safe to approve.
• 🟡 Yellow border — Confidence 65–84% (medium). The agent has some uncertainty. Read the rationale carefully before approving.
• 🔴 Red border — Confidence < 65% (low). The agent is flagging significant uncertainty. Do not approve without your own analysis. This is a human decision.

Each queue item shows:
• The proposed action (e.g. "Increase bid on 'blood pressure supplement' by $0.15")
• Full rationale explaining what data triggered it and which SOP rule applies
• Timestamp and SOP reference badge
• Confidence score with color indicator

Approve: Click ✅ Approve. The action is logged to the Audit Trail as APPROVED with your name and timestamp. The agent proceeds to execute it.

Reject: Click ❌ Reject. The action is logged as REJECTED. The agent does not execute it. Consider logging a note in the Methodology tab explaining why, so the agent learns the pattern.

Adding actions manually: Use the form at the bottom of the Queue tab to add an action yourself (useful when you spot something the agent hasn't flagged). Fill in the action description, select the SOP reference, write the rationale, and enter a confidence score. The action will appear in the queue for approval like any other.

The Audit Trail is an immutable, timestamped record of every action ever taken by the agent or approved by a human. It is your compliance ledger and accountability system.

Each entry contains:
• Timestamp (exact date and time)
• Action description
• Full rationale ("Action X taken because Data Point Y met SOP Rule Z")
• SOP reference badge
• Confidence score with color coding
• Status: APPROVED, REJECTED, or ROLLBACK
• Who approved it (your name or "Agent" for auto-executions)

One-Click Rollback: Every APPROVED action has a ↩ Rollback button. Clicking it logs a ROLLBACK entry to the audit trail and initiates the reversal of that change (bid restored to previous value, budget reversed, etc.). You will be asked to confirm before the rollback executes. The rollback itself is also logged — nothing is ever silently undone.

Export CSV: Click ⬇ Export CSV to download the full audit trail as a spreadsheet. Use this for reporting, compliance reviews, or sharing with stakeholders.

⚠️ The audit trail cannot be deleted or edited — it is append-only by design. Every action leaves a permanent record.

The Methodology tab is the agent's "Live README" — a real-time explanation of the logic, rules, and parameters it is currently using. Think of it as the agent's thinking out loud.

Current Agent State: Shows active mode (T1/T2/T3), active SOP version, confidence threshold, and data sources.

Active Mathematical Models: Plain-English description of every formula the agent is applying right now — placement bid multipliers, CPC benchmarks, budget utilization rules, and bid recommendation logic.

SOP Gaps Log: Whenever the agent encounters a scenario not covered by any Candace SOP, it stops and logs the gap here. This is a red alert — the agent is paralyzed and waiting for a human to either add a rule or give a one-time directive. Review these immediately and update the SOPs or provide guidance.

Log a Gap Manually: If you identify a scenario gap yourself (before the agent hits it), use the + Log a Gap Manually button to document it proactively.

Update Methodology (Admin only): When you change the agent's strategy (e.g. shifting from Growth Mode to Profitability Mode, or updating which keywords are "Hero" targets), log it here. Every methodology update must include:
• What changed
• Why it changed
• Which campaigns are affected
• Effective date

This is your operational changelog. Future team members can read the Methodology Log to understand exactly how and why decisions were made.

The SOP Schedule shows you the cadence of every automated task the agent runs. Each SOP has a defined frequency — the agent will execute (or queue for approval) on this schedule.

Status badges:
• 🟡 DUE SOON — This SOP is coming up within the next 7 days. Prepare any data or approvals in advance.
• 🟢 SCHEDULED — This SOP is scheduled and not yet due.
• ⚪ ON DEMAND — This SOP runs when manually triggered (SOP 6 — CPC Review).

Full SOP schedule:
• SOP 1 (ASIN Negation) — Every 2 weeks
• SOP 2 (Keyword Mapping) — Every 2 weeks (keywords), Monthly (ASIN targets)
• SOP 3 (Budget Review) — Every 2 weeks
• SOP 4 (Intentwise Bidding Recs) — Every Wednesday
• SOP 5 (Placement Bids) — Monthly
• SOP 6 (CPC Lifetime Review) — On demand / optimization cycles
• SOP 7 (Campaign Mapping) — Monthly

When the agent runs a scheduled SOP, all resulting actions appear in the Approval Queue (if below confidence) or are auto-executed and logged to the Audit Trail (if above confidence).

Only Admin (T3) users can access the Users tab. This is where you control who has access to the Mission Control and what they can do.

Adding a user: Click + Add User. Fill in their full name, email address, role (Admin/Manager/Viewer), and a temporary password. Send them the temporary password securely — they should change it on first login using the 🔑 Change Password button in the header.

Roles:
• Admin (T3): Complete access. Use for PPC team leads and account owners only.
• Manager (T2): Can approve/reject queue, view all logs. Ideal for PPC specialists and campaign managers.
• Viewer (T1): Read-only. Ideal for clients, stakeholders, or junior team members who need visibility but not control.

Deactivating a user: Click Deactivate next to any user. They immediately lose login access. Their historical entries in the audit trail are preserved.

Removing a user: Click Remove to permanently delete the account. This cannot be undone. Their audit trail entries remain for compliance.

This is the most important tab for configuring how the agent thinks. Every numerical value the agent uses in its logic is set here — nothing is hardcoded.

⚠️ Critical rule: The agent reads these thresholds before every single action. Changing a value here takes effect immediately on the next action cycle. Think carefully before editing live thresholds.

Agent Mode:
• T1 (Manual) — The agent analyzes but never acts. Every single action requires your approval. Use this during onboarding or when you don't yet trust the agent's judgment.
• T2 (Semi-Autonomous) — The agent auto-executes actions above your confidence threshold and flags the rest. This is the recommended production mode.
• T3 (Full Autonomy) — The agent executes all SOP-driven actions automatically. Use only when you have high confidence in the thresholds and have reviewed the agent's track record for 30+ days.

Confidence Threshold: The cutoff score (0–100%) above which the agent can act autonomously in T2/T3 mode. Default is 85%. Raising this to 90–95% makes the agent more conservative. Lowering it to 70–75% gives the agent more autonomy but increases risk. Never set below 70%.

SOP Conflict Resolver: When a threshold you set here directly contradicts a rule in the Candace SOPs, the agent needs to know which wins:
• SOP Wins — The Candace SOP always takes precedence. Your threshold is ignored for that decision.
• Settings Win — Your threshold overrides the SOP. The agent follows your number.
• Pause & Alert (Recommended) — The agent stops, logs the conflict in the Methodology tab, and waits for a human to resolve it manually. This is the safest option.

Financial Guardrails:
• TACoS Max % — Your ultimate ceiling. If total ad spend / total revenue exceeds this, the agent enters panic mode and pauses spend increases until it recovers.
• ACoS Buffer % — How much slack above your target ACoS is acceptable. E.g. if target ACoS is 25% and buffer is 5%, the agent starts defensive action at 30%.
• Max Daily Budget Shift % — Prevents runaway budget increases. If set to 20%, the agent cannot increase any campaign's daily budget by more than 20% in a 24-hour period without approval.
• NTB Priority Weight (1–10) — How much extra value to assign to New-to-Brand customer conversions vs. repeat purchases. Higher = agent is more willing to spend on NTB-focused campaigns.

Inventory-Adaptive Bidding:
• DOS Critical Threshold — Days of FBA stock remaining. If a SKU drops below this number, the agent cuts bids by the Critical Reduction % to slow sales velocity and prevent stockout.
• DOS Critical Bid Reduction % — How aggressively to cut bids when stock is critical.
• DOS Liquidation Trigger — If stock exceeds this many days, the agent increases the ACoS target to push more aggressive spend and clear inventory.
• DOS Liquidation ACoS Boost % — How much to raise the ACoS target during clearance mode.
• Lead Time Buffer (days) — Factory-to-FBA shipping time. Added to DOS calculations to prevent premature bid cuts (e.g. if you have 15 days of stock but 45-day lead time, the agent treats it as a critical situation).

Keyword & Campaign Lifecycle:
• Harvest Min Sales — A keyword must generate at least this many sales in an Auto or Broad campaign before the agent promotes it to an Exact Match campaign. Too low = premature promotion of unproven keywords. Too high = slow keyword development.
• Harvest Max ACoS % — Even if a keyword hits the sales threshold, it cannot be promoted if its ACoS exceeds this. Ensures only profitable keywords advance.
• Negate Click Floor — If a search term accumulates this many clicks with zero sales, the agent flags it for negation. Prevents wasted spend on non-converting terms.
• Negate Spend Cap $ — Alternatively, if a term spends this much with zero sales (regardless of clicks), it gets flagged. Whichever threshold is hit first triggers the negation.
• Max CPC SP / SB $ — Hard bid ceilings. The agent will never bid above these values for Sponsored Products or Sponsored Brands, even if Intentwise recommends it. If there is a conflict with a Candace SOP, the SOP Conflict Resolver applies.

Organic & Competitive Positioning:
• Organic Rank Target — When a keyword achieves this organic position (e.g. top 3), the agent begins "PPC Pullback" — gradually reducing PPC bids since organic traffic is handling visibility. This saves budget on keywords that no longer need heavy PPC support.
• PPC Pullback Increment % — How much to reduce PPC bid each cycle once the organic target is met.
• Conquesting Max ACoS % — Competitor brand keywords are intentionally allowed a higher ACoS target since their CPC is typically higher and conversion rate lower. Set this separately from your main ACoS target.
• SOV Target % — Share of Voice target for your Hero keywords (top 5–10 most important terms). The agent monitors impression share and adjusts bids to maintain this level of visibility.
• Placement ROI Ratio — Minimum ROAS required before the agent applies placement bid multipliers. Prevents multiplier increases on placements that aren't yet profitable.
• CVR Target % — Conversion rate threshold a keyword must meet to be considered for promotion or bid increases.

Conflict detection: When you click Save All Thresholds, the system automatically checks for logical contradictions (e.g. Harvest Max ACoS lower than TACoS Max, DOS Critical overlapping with Liquidation Trigger, Max CPC exceeding SOP guidance). Conflicts are flagged inline and logged to the Methodology tab. Resolve them before the next agent cycle.

These 7 SOPs are the agent's entire rulebook. It will not take any action not covered by one of these SOPs.

SOP 1 — Irrelevant ASIN Negation from AUTO Campaigns (Bi-weekly)
The agent logs into Intentwise, sets date range to Last 60 Days, and filters search terms containing "B0" (ASIN format) with zero orders. It cross-checks each ASIN on the Amazon SERP to confirm irrelevance, then adds confirmed irrelevant ASINs as Negative Product Targets in auto campaigns. Important: ASINs that may be seasonally relevant or have low conversion for non-relevance reasons are NOT negated.

SOP 2 — Keyword Mapping & ASIN Targeting Research (Bi-weekly for keywords, Monthly for ASINs)
Full keyword discovery using Helium 10, Jungle Scout, Brand Analytics, and Data Dive. The agent identifies high-volume, long-tail, competitor, and category keywords. It maps them to a structured spreadsheet with campaign type (SP/SB/SD), intent (Branded/Non-Branded/Competitor/Category), and assigns them to campaigns. Branded and non-branded keywords are always kept in separate campaigns. New opportunities are flagged. High PPC rank / weak organic rank keywords get bid increases.

SOP 3 — SP, SB & SD Budget Review (Bi-weekly)
The agent reviews the Amazon Ads Console Budgets tab filtered to Active campaigns, Last 30 Days. It sorts by Avg. Time in Budget (low to high) and focuses on campaigns under 100% utilization. If a recommended budget exists → apply it. If not → increase by the % defined in your Settings (Max Daily Budget Shift). For SB/SD: sorted by spend high to low, recommendations applied if ROAS meets or exceeds your target.

SOP 4 — Intentwise Bidding Recommendations (Every Wednesday)
The agent reviews Intentwise's Bidding Optimization section only (Keyword Targeting recommendations are explicitly ignored). It reviews current vs. recommended bid for each keyword/ASIN, selects appropriate targets, applies recommendations, and submits. If any recommendation appears aggressive or contradicts performance data, the action is queued for human approval instead of auto-applying.

SOP 5 — Placement Bid Optimization (Monthly)
The agent reviews Top of Search, Product Detail Page, and Rest of Search placement ROAS via Intentwise. It applies ROAS-banded multipliers:
ROAS 5–10 → +5% | ROAS 10–20 → +10% | ROAS >20 → +20%
ROAS 2–3 → -5% | ROAS 1–2 → -10% | ROAS 0–1 → -20% or 0%
No spend → +10%
These rules apply equally to TOS, PDP, and Rest of Search. The Placement ROI Ratio threshold in Settings must be met before any increase multiplier is applied.

SOP 6 — CPC Review Using Lifetime Data (On demand)
The agent sets the date range to Lifetime, filters to Enabled campaigns and targets with ROAS above your brand-specific goal, and sorts by Orders (highest first). It reviews the top 15–20 targets and compares current bid to CPC: Bid ≈ CPC → leave it. Bid 2–3x CPC → review for reduction. Bid 4x+ CPC without justification → reduce toward CPC. Every change is documented with original bid, CPC, new bid, and rationale.

SOP 7 — Campaign Mapping (Monthly)
Full export and audit of all campaigns. The agent reviews for duplicate targets, overlapping keyword intent, and coverage gaps. Every keyword and ASIN is mapped to the correct campaign type and intent. Branded and non-branded keywords must never share a campaign. Action statuses are updated (Active/Paused/Planned).

Before every action, the agent assigns itself a confidence score from 0–100%. This score determines whether the action executes automatically or goes to the queue.

What determines the score:
• Data completeness — Does the agent have enough data? (minimum 30 days for bid changes). Thin data = lower confidence.
• SOP match — Does the scenario exactly match a Candace SOP rule, or is it a partial match? Exact match = high confidence. Partial match = medium.
• Risk level — Is the action reversible? A bid change of $0.05 is low risk (high confidence). A 50% budget cut is high risk (lower confidence).
• Anomaly check — Is the performance data within normal range, or is it an unusual outlier? Outlier data = lower confidence.

Score ranges:
🟢 90–100% — Exact SOP match, full data, low risk. Auto-executes in T2/T3.
🟢 85–89% — High confidence, minor data gaps or small risk factors. Auto-executes in T2/T3.
🟡 65–84% — Partial SOP match or limited data. Goes to Approval Queue regardless of tier.
🔴 0–64% — No SOP match, anomalous data, or high-risk action. Blocked. Logs to queue with full explanation. Human must decide.

Your Confidence Threshold setting (default 85%) is the dividing line between auto-execute and queue. You can raise it to be more conservative or lower it carefully if the agent has a strong track record.

T1 — Manual: The agent runs all analysis and generates all recommendations, but executes nothing. Every single action — no matter how routine — lands in the Approval Queue for your explicit decision. Use this during the first 2–4 weeks of deployment to build confidence in the agent's logic. Expect a high volume of queue items.

T2 — Semi-Autonomous (Recommended): The agent auto-executes anything above your confidence threshold and queues everything else. This is the right balance for ongoing operations — the agent handles routine, well-defined tasks automatically while humans retain control over uncertain decisions.

T3 — Full Autonomy: The agent executes all SOP-driven actions automatically, regardless of confidence score (as long as it's above 65% — the absolute floor). All actions are still logged to the Audit Trail. Use T3 only after operating in T2 for at least 30 days with a clean record and when your team has reviewed and is satisfied with the thresholds. Even in T3, actions below 65% confidence are always blocked and queued.

⚠️ Never switch from T1 to T3 directly. Always transition: T1 → T2 (2–4 weeks) → T3 (if needed).

Conflicts arise when a threshold you set in Settings contradicts an explicit rule in the Candace SOPs. For example: the SOPs say never bid over $5.00 on SP, but you've set Max CPC SP to $7.00 in Settings.

Option A — SOP Wins: The Candace SOP always takes precedence. The agent ignores the conflicting threshold and follows the SOP. Your Settings value is overridden silently for that decision (but the conflict is still logged).

Option B — Settings Win: Your threshold overrides the SOP. The agent follows your number. This essentially customizes the SOP for your brand's specific situation. Use with caution — you're knowingly deviating from Candace's guidance.

Option C — Pause & Alert (Recommended): The agent stops, logs the conflict in the Methodology tab as a SOP Gap, and sends an alert to the Approval Queue. A human must resolve the conflict by either updating the SOP, adjusting the threshold, or giving a one-time directive. This is the safest option because no action is taken under uncertainty.

Conflicts are automatically detected when you save thresholds and flagged in the Settings tab. Always resolve conflicts before the next agent cycle.

This is one of the most powerful features. The agent monitors FBA inventory levels and automatically adjusts bidding strategy to prevent stockouts and accelerate clearance.

Stockout Prevention (DOS Critical):
When a SKU's Days of Supply drops below your DOS Critical Threshold (default: 10 days), the agent cuts bids by your DOS Critical Reduction % (default: 30%). This slows sales velocity to prevent running out of stock before the next shipment arrives. The Lead Time Buffer is added to the calculation — if your factory-to-FBA time is 45 days, the agent starts treating 55 days of stock as critical.

Liquidation Mode (DOS Liquidation):
When stock exceeds your DOS Liquidation Trigger (default: 90 days), the agent enters clearance mode. It increases the ACoS target by your DOS Liquidation ACoS Boost % (default: 15%), allowing the agent to spend more aggressively to move excess inventory. This prevents storage fees and capital lockup.

Important: These rules apply per SKU, not per campaign. A campaign running multiple SKUs will have its bids modulated individually based on each SKU's inventory status.

PPC Pullback (Organic Flywheel):
As your products gain organic rank, you need less PPC support for those keywords. The agent monitors organic position and when a keyword achieves your Organic Rank Target (e.g. top 3), it begins gradually reducing PPC bids by your Pullback Increment % each cycle. This saves budget that can be redirected to keywords still needing PPC support. The pullback stops when organic rank slips below the target again.

Conquesting:
Competitor brand keywords have a separate ACoS allowance (Conquesting Max ACoS). These keywords are inherently less efficient (lower CVR, higher CPC) but strategically valuable. Setting a higher ACoS ceiling here allows the agent to maintain presence on competitor terms without the conservative bid logic applying to them.

Share of Voice:
For your Hero keywords (your top 5–10 most important terms), the agent tracks impression share. If SOV drops below your SOV Target %, the agent will increase bids to reclaim visibility. This ensures you never cede dominance on your most valuable keywords.

Placement Optimization:
Bid multipliers for Top of Search, PDP, and Rest of Search are only applied when the placement's ROAS meets or exceeds your Placement ROI Ratio. This prevents you from paying more for placements that aren't yet delivering a return.

If the agent encounters a situation not covered by any of the 7 Candace SOPs, it will never guess or improvise. It follows this exact protocol:

1. STOP immediately — The action is not taken under any circumstances.
2. Log the gap — A SOP Gap entry is created in the Methodology tab describing the scenario precisely.
3. Alert the queue — An item appears in the Approval Queue flagged as a SOP Gap requiring human input.
4. Wait — The agent does not attempt the action again until a human resolves the gap.

How to resolve a gap:
Option A: Update the Candace SOPs to cover the new scenario (contact Candace or your PPC team lead).
Option B: Add a new threshold or rule in the Settings module that handles the scenario.
Option C: Approve a one-time manual action via the queue for this specific instance, and note that the SOP needs updating.

⚠️ Never ignore a SOP gap. The agent will keep flagging it every cycle until it's resolved.

Every action in the queue and audit trail includes a rationale statement. Learning to read it quickly is essential.

Standard format:
"Action [X] was taken because [Data Point Y] met [User-Defined Threshold Z]."

Examples:
• "Increased bid on 'blood pressure supplement' by $0.15 because CR = 18.2% and ACoS = 21.3% (7.7% below target of 29%) per SOP 5 Rule: ROAS > 10 → +10% bid increase. Confidence: 92%."
• "Negated ASIN B08XKL2932 from AUTO campaign because it accumulated 22 clicks (above Negate Click Floor of 15) with $0 in sales and $31.50 in spend (above Negate Spend Cap of $25). Per SOP 1. Confidence: 97%."
• "Reduced bids by 30% on all SP campaigns for SKU B09X. DOS = 8 days, below DOS Critical Threshold of 10 days. Lead Time Buffer = 45 days applied. Per Inventory Logic v2. Confidence: 94%."

The rationale always tells you: what happened, why it happened (the exact data), which rule triggered it, and how confident the agent was. If any of these seem wrong, use the Rollback button immediately.

✅ DO:
• Check the Dashboard and Approval Queue every morning before 10AM
• Review the Audit Trail weekly to spot patterns in the agent's decisions
• Update thresholds gradually — change one variable at a time and observe impact before changing another
• Start in T1 mode for the first 2–4 weeks, then move to T2
• Log every strategy shift in the Methodology tab immediately
• Export the Audit Trail monthly for your records
• Resolve SOP gaps within 24 hours — the agent is paused on those scenarios
• Change your password from the default immediately
• Review red-bordered queue items personally — never approve without reading the rationale

❌ DON'T:
• Never switch from T1 directly to T3 — always go T1 → T2 → T3
• Never lower the confidence threshold below 70%
• Never approve a red-bordered item (confidence <65%) without your own analysis confirming the action
• Never ignore SOP gaps — they mean the agent is stuck
• Never change multiple thresholds simultaneously — you won't know which change caused a problem
• Never share login credentials — every user needs their own account for the audit trail to be accurate
• Never delete audit trail entries — they exist for compliance and rollback purposes
• Never trust an action without reading its rationale — the agent should always have a reason
• Never use T3 mode during a product launch, sale event, or major inventory change without extra monitoring