Aug 14, 2026
Evidence Packet for Every Budget Move
An evidence packet is a timestamped record of performance data, decision rationale, and rollback conditions attached to each budget change. It documents the 7-day and 28-day metrics that justified the move and the threshold at which the change should be reversed.

Why Evidence Packets Matter
Budget moves without documentation create two problems: decision amnesia and undefended spend. Six weeks after increasing a channel's budget, the operator who approved it may not remember why. The new operator reviewing spend sees only the current allocation, not the performance that justified it.
Evidence packets solve this by creating a paper trail. They force clarity at the moment of decision and provide a reference point for future audits. When a channel underperforms, the packet shows whether the original thesis still holds or whether conditions have changed.
The 7-Day Snapshot
The 7-day snapshot captures immediate performance after a budget move. It answers: did the channel respond as expected in the first week?
- ROAS or CPA at day 7 (or day 3 if testing paid social)
- Spend velocity - did the budget get deployed as planned?
- Volume metrics - orders, sessions, or conversions depending on channel
- Any platform-specific flags - account health, policy warnings, creative fatigue signals
The 28-Day Baseline
The 28-day metric is the decision standard. It smooths out weekly noise and reflects true channel economics. This is the number that determines whether the move was correct.
- ROAS or blended CPA at 28 days
- Cumulative spend and revenue
- Repeat purchase rate or LTV proxy if available
- Comparison to the channel's historical 28-day performance (if scaling up) or to benchmark (if new channel)
The Decision Metric and Trigger
Every budget move must name the single metric that triggered it. This is not the full business case - it is the one number that crossed a threshold.
Examples: 'Increased Google Shopping budget because 28-day ROAS hit 3.2x (threshold: 3.0x).' Or: 'Reduced TikTok spend because 7-day CPA rose to $18 (rollback trigger: $17).' Naming the metric prevents post-hoc rationalization and makes rollback automatic.
The Rollback Rule
A rollback rule is a pre-set condition that reverses the budget move without debate. It removes emotion from underperformance and forces action on drift.
Set the rollback threshold at the 28-day mark. If the channel does not hit the target metric by day 28, revert the budget and document why it failed. Common rollback rules:
- ROAS falls below X for 28 days - reduce budget by Y%
- CPA rises above X for 28 days - pause channel entirely
- Spend velocity is below 80% of plan - reduce daily budget and investigate
- Creative fatigue detected (CTR down >15% week-over-week) - pause and refresh creatives before re-testing
The Evidence Packet Template
Document each move in a single record. This can live in a spreadsheet, Slack thread, or ops tool - the format matters less than consistency.
- Date of move
- Channel and campaign
- Old budget vs. new budget (dollar amount and % change)
- Decision metric and threshold that triggered the move
- 7-day performance snapshot (date captured, ROAS/CPA, spend, volume)
- 28-day performance snapshot (date captured, ROAS/CPA, spend, volume, repeat rate if tracked)
- Rollback rule (the condition that reverses this move)
- Owner name
- Status: Active, Rolled Back, or Graduated (moved to permanent allocation)
Execution Checklist
Before approving any budget change, verify:
- The decision metric is named and the threshold is written down
- 7-day and 28-day baselines are captured at the time of the move (or within 24 hours)
- The rollback rule is specific enough that a different operator can execute it without interpretation
- The packet is stored in a shared location and linked to the budget change in the ad platform or finance tool
- A calendar reminder is set for day 28 to review the 28-day metric and decide: hold, scale, or rollback
Questions
FAQ
What if a channel is new and has no historical baseline?
Use the benchmark from similar channels or competitors. If scaling Google Shopping from $500/day to $2000/day, the baseline is the current 28-day ROAS at $500/day. If launching a new channel like Pinterest, set the rollback threshold at the blended brand ROAS minus 20% (e.g., if brand ROAS is 2.5x, rollback if Pinterest hits below 2.0x by day 28).
Should the 7-day snapshot trigger rollback decisions?
No. The 7-day snapshot is diagnostic only. It flags problems early (e.g., a platform outage, creative underperformance) but does not trigger rollback. Rollback decisions are made at 28 days. Use 7-day data to decide whether to pause and investigate before day 28 completes.
How often should budget moves be documented?
Every move, including small daily adjustments. If the budget for a campaign changes by more than 10% or the daily spend moves by more than $100, document it. Micro-adjustments below that threshold can be logged in batch weekly.
What happens if the rollback rule is met but business context has changed?
Execute the rollback as planned. Document the reason for the underperformance in a separate note (e.g., 'Rolled back due to iOS update impact on pixel tracking'). Then run a new test with updated assumptions and create a new evidence packet. Overriding rollback rules erodes the system.
More from the blog
- Did the action actually work?
- One number a day
- Sunday night reporting is a product bug
- Never let AI change ad spend without a yes
- Stop optimizing platform ROAS alone
- Write-Access Matrix for AI on Meta and Google
- Reverse Platform ROAS Dependency Before It Reverses You
- AI Agents for Ecommerce: Scheduled Loops, Tools, and Approval Gates
- Data Requirements for AI in Ecommerce
- The AI Ecommerce Stack for DTC Brands
- AI for Ecommerce Agencies: Automate Execution, Keep Craft
- Reconciling Attribution Conflict with AI
- AI for Ecommerce During BFCM: What to Freeze, Monitor, and Automate
- AI for Ecommerce Creative Testing Workflows
- AI for Ecommerce Customer Support That Protects Brand
- AI for Email and SMS Operations: Detection, Fatigue, and Segmentation
- Recovering Revenue from Failed Payments: AI Retry Logic for DTC
- What Ecommerce Founders Should Never Automate
- AI for Ecommerce Fraud and Chargeback Signals
- AI for Ecommerce Growth Teams: Roles and Rituals
- AI for Ecommerce Inventory: Demand Signals from Ads and Cohorts
- AI for Ecommerce Pricing and Promo Calendars
- AI for Ecommerce Reporting: Kill the Sunday Deck
- Security and Access Control for Ecommerce AI
- AI for Ecommerce Unit Economics Decisions
- Winback Campaigns: Prioritize High-Value Lapsed Customers and Ladder Offers
- Prevent PMax Cannibalization and Reclaim Brand Search ROI
- AI for Meta Ads in Ecommerce: Operator Checklist
- AI for Multichannel Ecommerce: Connecting Inventory, Pricing, and Ads Across Channels
- AI for Shopify Merchandising and Margin
- AI for Subscription Ecommerce: Dunning, Churn Prevention, and Revenue Stacking
- AI for TikTok Ads: Solving Creative Volume Without Losing Control
- AI Operator vs Growth Agency: What Each Covers and Costs
- AI Operator vs In-House Analyst: Cost and Task Split
- AI Operator vs Klaviyo AI: When to Choose Each
- AI Operator vs Meta Advantage+ - Where Each Solves
- AI Operator vs Northbeam: Measurement vs Execution
- AI Operator vs Shopify Sidekick: Scope and Operational Fit
- AI Operator vs Triple Whale: Measurement Layer vs Execution Layer
- AI Will Not Fix Bad Creative
- AI Will Not Negotiate Your Suppliers
- Analyst vs Operator: Split the Job Before You Hire
- AOV Checklist for Growth Leads
- AOV for Multi-Channel DTC
- AOV Thresholds Worth Writing Down
- Approval-Gated AI Is a Feature, Not a Missing Feature
- ASC Campaigns and Contribution Margin
- Attribution Checklist for Growth Leads
- Attribution for Multi-Channel DTC
- Attribution Thresholds Worth Writing Down
- Best AI Tools for Ecommerce in 2026 (By Job, Not Hype)
- Black Friday Automation Freeze: What Stays Manual
- Never Mix Brand Search and Prospecting Efficiency
- Building an AI-First Ecommerce Ops Team
- CAC Checklist for Growth Leads
- CAC for Multi-Channel DTC: Definitions, Thresholds, and Failure Modes
- CAC Thresholds Worth Writing Down
- Cancel Flow Metrics That Matter
- ChatGPT Cannot See Your Ad Account
- Churn Checklist for Growth Leads
- Churn for Multi-Channel DTC
- Churn Thresholds Worth Writing Down
- Cohort Analysis: The Gate Before Scaling Spend
- Cohorts Checklist for Growth Leads
- Cohorts for Multi-Channel DTC
- Cohorts Thresholds Worth Writing Down
- Common AI Ecommerce Mistakes Brands Make
- Common AOV Mistakes on Shopify
- Common Attribution Mistakes on Shopify
- Common CAC Mistakes on Shopify
- Common Churn Mistakes on Shopify
- Common Cohorts Mistakes on Shopify
- Common Creative Mistakes on Shopify
- Dunning Failures on Shopify: Definitions, Thresholds, and Recovery
- Common LTV Mistakes on Shopify
- Margin Mistakes That Kill Shopify Unit Economics
- Common MER Mistakes on Shopify
- Common Retention Mistakes on Shopify
- ROAS Mistakes That Kill Shopify Profitability
- Contribution Margin: The One Finance Number Paid Social Needs
- Copilot vs Autopilot: Approval Gates for Ecommerce AI
- Creative Checklist for Growth Leads
- Detecting Creative Fatigue: Operational Signals That Matter
- Creative for Multi-Channel DTC
- Creative Kill Criteria You Can Write Down
- Creative Thresholds Worth Writing Down
- Credits and Honest Metering: How Usage-Based Pricing Should Work
- Dashboards Do Not Pause Ads
- Dayparting Is Usually Wrong for Ecommerce
- Demo Theater vs Production AI: Why Read-Only Proofs Matter
- Dunning Checklist for Growth Leads
- Dunning for Multi-Channel DTC
- Dunning Thresholds Worth Writing Down
- Email Fatigue from Growth Teams: When Send Volume Kills LTV
- Email Revenue Collapsed Overnight: Flow Break Detection
- Failed Payment Alert Design for Operators
- Failed Payments Are Not Churn
- Finance Rejects Marketing Numbers
- First Week With an AI Operator: Read-Only, Briefings, Then Gated Writes
- Why Your CAC Just Moved: A Diagnostic Framework
- Frequency Cap as Brand Protection
- GA4 Is Not Your P&L
- Google Ads Brand vs Nonbrand Split: Reporting Rule
- Brand Cannibalization: Measuring When Paid Brand Search Destroys ROI
- Health Score Inputs for DTC: RFM + Support + Payments
- Why Horizontal AI Employees Don't Move Shopify Store Metrics
- How Operators Think About AOV
- Attribution as a Measurement System
- How Operators Think About CAC
- How Operators Think About Churn
- Cohort Analysis for DTC Operators
- How Operators Think About Creative
- How Operators Think About Dunning
- How Operators Think About LTV
- How Operators Think About Margin
- How Operators Think About MER
- How Operators Think About Retention
- How Operators Think About ROAS
- How Operators Think About Subscription
- MER as a Daily Operating Metric
- Run a Two-Week Read-Only AI Pilot
- How to Use AI for Ecommerce Ads Without Blowing the Budget
- How to Use AI for Ecommerce Retention and Lifecycle
- Human SLA for AI Proposals: Same-Day Approvals or the Queue Is Theater
- Implementing AI in Ecommerce in 30 Days
- Who Owns Involuntary Churn
- Connect Shopify, Meta, and Klaviyo Without a Data Team
- Klaviyo Flows the Operator Watches Weekly
- Learning Phase Budget Mistakes: Why Ad Restarts Waste Spend
- LTV Checklist for Growth Leads
- LTV for Multi-Channel DTC: Calculation, Thresholds, and Failure Modes
- LTV Thresholds Worth Writing Down
- Margin Checklist for Growth Leads
- Margin Floor by Collection: Gate Media Spend on Unit Economics
- Margin for Multi-Channel DTC
- Margin Thresholds Worth Writing Down
- Measuring AI ROI in Ecommerce: Hours, Revenue, and Avoided Spend
- MER Checklist for Growth Leads
- MER Down After a Creative Win
- MER for Multi-Channel DTC: Thresholds and Failure Modes
- MER Thresholds Worth Writing Down
- Meta Ads Manager Is Not Enough
- What to do when Meta Pixel stops firing
- Ecommerce AI Operator vs Generic AI Employee: Vertical Depth and Operational Ownership
- The Eight Fields Every Monday Brief Needs
- Multi-Channel Complexity Is the Prerequisite
- New CMO Wants Another Dashboard: What to Buy Instead
- Connected Operator vs Chat With a CSV
- Pause Rules That Fire on Noise
- Pixel Broke on Friday Night: Incident Response Playbook
- Freeze AI Automation During Promo Weeks
- Prompting vs Connecting: Two Modes of Ecommerce AI
- Reading Failed Billing Signals in Your Morning Brief
- Refund Rate as Acquisition Quality Signal
- Fix Retention Before Buying More CAC
- Retention Checklist for Growth Leads
- Retention for Multi-Channel DTC
- Retention Thresholds Worth Writing Down
- ROAS Checklist for Growth Leads
- ROAS for Multi-Channel DTC: Channel Benchmarks and Reallocation Rules
- ROAS Thresholds Worth Writing Down
- ROAS Up, Cash Down: The Pattern
- Rules Engine vs Approval-Gated AI: When If-Then Logic Fails
- Scale Signals That Are Fake
- Second Purchase Campaign Timing by Category
- Shopify Plus Operator Checklist: Connection Sequence
- Skio, Loop, Bold: Subscription Stack Comparison for Operators
- Slack Approval Button Design
- Slack as the Ecommerce Ops Console
- Software Does Not Replace Brand Taste
- Stop Guessing on AOV
- Stop Guessing on Attribution
- Stop Guessing on CAC
- Stop Guessing on Churn
- Cohort Analysis for DTC: Definitions, Thresholds, and Failure Modes
- Stop Guessing on Creative
- Dunning: Definition, Thresholds, and Failure Modes
- Stop Guessing on LTV
- Stop Guessing on Margin
- Stop Guessing on MER
- Stop Guessing on Retention
- Stop Guessing on ROAS
- Subscription Billing Decline Codes Operators Must Know
- Why Subscription Churn Spikes on Monday
- Surface MRR Risk and Dunning Status Daily
- Subscription Pause as Retention
- Support Tickets as a Churn Signal
- The 11pm Slack Question That Should Be a Scheduled Job
- TikTok Creative Volume Problem: Ops Capacity Limits
- TikTok Testing Budget Rules for DTC
- Using AI to Increase Ecommerce LTV
- Reduce Ecommerce CAC by Automating Waste Detection and Creative Cycles
- UTM Hygiene as Ops Debt
- Vanity Automation Scoreboards: Actions Taken vs Revenue Moved
- Voluntary Churn Reasons Taxonomy
- Weekly AOV Review Template
- Weekly Attribution Review Template
- Weekly CAC Review Template
- Weekly Churn Review Template
- Weekly Cohorts Review Template
- Weekly Creative Review Template
- Weekly Dunning Review Template
- Weekly LTV Review Template
- Weekly Margin Review Template
- Weekly MER Review: Thresholds and Failure Modes
- Weekly Retention Review Template
- Weekly ROAS Review: Thresholds, Diagnostics, and Decision Rules
- What Is a Scheduled Growth Brief?
- What Is an Ad Audit Agent?
- What Is an Ecommerce AI Operator?
- Approval-Gated Automation: Definition and Implementation
- Blended CAC for Operators
- Churn Risk Ranking: Prioritized Customer Intervention Lists
- Contribution Margin ROAS: The Profitability-First Ad Metric
- Cross-Tool Reconciliation: Matching Data Across Shopify, Meta, and Klaviyo
- Operator Memory Across Tools: Why Chat Tabs Fail
- Read-Only Pilot Mode: Definition and Implementation
- What We Will Not Automate in Ecommerce Ops
- Adjudicating Meta ROAS vs Shopify MER Without Politics
- When AOV Is the Wrong Metric
- When Attribution Is the Wrong Metric
- When CAC Is the Wrong Metric
- When Churn Is the Wrong Metric
- When Cohort Analysis Hides What You Need to Fix
- Creative Is Not a Metric
- When Dunning Is the Wrong Metric
- When LTV Is the Wrong Metric
- When Margin Is the Wrong Metric
- When MER Is the Wrong Metric
- When Not to Buy an AI Operator
- When Retention Is the Wrong Metric
- When ROAS Is the Wrong Metric
- When to Kill the Weekly Deck
- When to Pause vs Cut Budget
- Why Every Write Action Is Gated
- Build an Offer Ladder for Lapsed Customers
- You Still Need a Human Who Owns the P&L
- All guides