Marketing Spend Analysis

Marketing Spend Analysis

Overview

Marketing spend is typically the largest variable cost in a DTC ecommerce business — often 15–40% of revenue. Unlike most costs, marketing spend is directly controllable in near-real-time: you can increase or decrease budgets on paid channels within minutes. This creates both opportunity (scale what works) and risk (waste capital on what does not).

The core goal is to maximize total contribution profit from your marketing investment — not just revenue. A channel with high ROAS but thin margins, high return rates, or low AOV may generate less actual profit than a channel with lower ROAS and stronger unit economics.

This skill guides you through building a unified view of marketing spend and performance across all channels, using tools designed specifically for ecommerce merchants.

When to Use This Skill

• When managing marketing budgets across multiple platforms (Meta, Google, TikTok, Amazon Ads)
• When wanting to identify which channels generate the most profitable customers
• When needing a unified marketing performance dashboard fed by multiple ad platforms
• When hitting diminishing returns on a key channel and deciding how to reallocate spend
• When comparing platform-reported ROAS against first-party attributed ROAS
• When building a marketing efficiency report for a board or investor update

Core Instructions

Step 1: Choose a unified marketing analytics tool

The biggest problem in marketing spend analysis is that each platform (Meta, Google, TikTok) reports its own ROAS using its own attribution window — and they all claim 100% credit. You need a tool that pulls data from all platforms into one view and compares against your actual order data.

Key metrics to track in your chosen tool:

Step 2: Set up your marketing analytics stack

---

Shopify

Triple Whale setup (recommended for $50K+/mo ad spend):

1. Install Triple Whale from the Shopify App Store 2. Connect all ad accounts under Settings → Integrations: Meta, Google, TikTok, Pinterest, Snapchat 3. Triple Whale installs a first-party pixel on your store that tracks the full customer journey 4. Go to Triple Whale → Summary Dashboard — view daily spend, attributed revenue, ROAS, MER, new customer CAC, and gross profit by channel in one dashboard 5. Go to Triple Whale → Attribution → Channel to see revenue under different attribution models side by side (first-click, last-click, Triple Whale's blended model) 6. Set up Morning Digest emails — Triple Whale sends a daily performance summary automatically 7. Set Budget Alerts under Settings → Alerts — get Slack/email notifications when ROAS drops below threshold or spend is pacing to overshoot monthly budget

Polar Analytics setup (mid-market, more affordable):

1. Install Polar Analytics from the Shopify App Store 2. Connect ad accounts under Integrations 3. Go to Polar → Channels — view spend, attributed revenue, ROAS, and gross profit by channel 4. Go to Polar → Blended Dashboard — view overall MER and blended spend vs. revenue trend 5. Use Polar → AI Insights for automated anomaly detection and spend recommendations

---

WooCommerce

Metorik + GA4 setup:

1. Install Metorik and connect to your WooCommerce store 2. Metorik captures UTM parameters on every order — go to Metorik → Reports → UTM to see orders and revenue by utm_source, utm_medium, utm_campaign 3. Install Google Analytics 4 via the Site Kit or MonsterInsights plugin 4. In GA4, go to Advertising → Attribution → Model comparison — select Google, Paid Social, Email as your channels and compare last-click vs. data-driven attribution 5. For ad spend data from non-Google platforms (Meta, TikTok), manually import spend CSVs into a Google Sheet and connect to Looker Studio alongside GA4 data for a unified view

Alternative: Northbeam or Rockerbox

• Both support WooCommerce via JavaScript pixel + order API integration
• Provide the same cross-platform attribution capabilities as Triple Whale but with WooCommerce compatibility

---

BigCommerce

1. Install Glew.io from the BigCommerce App Marketplace 2. Connect ad accounts (Meta, Google, Amazon) in Glew under Integrations 3. Glew shows spend, attributed revenue, ROAS, and blended MER by channel 4. For more advanced first-party attribution: install Rockerbox via script injection (BigCommerce → Storefront → Script Manager)

---

Step 3: Define your ROAS break-even and targets

Before analyzing whether a channel is performing well, calculate your minimum viable ROAS.

ROAS break-even formula:

Break-even ROAS = 1 / (1 - COGS rate - variable cost rate)

Example:
  COGS rate: 45% of revenue
  Fulfillment + payment fees: 12% of revenue
  Break-even ROAS = 1 / (1 - 0.45 - 0.12) = 1 / 0.43 = 2.33x

Any channel with true ROAS below 2.33x is losing money on every sale.

Target ROAS by channel type:

Step 4: Build a weekly channel performance scorecard

Set up a weekly review with this structure. Pull data from your analytics tool (Triple Whale, Glew, GA4):

CHANNEL PERFORMANCE SCORECARD (Last 7 Days vs. Prior 7 Days)
─────────────────────────────────────────────────────────────
Channel    | Spend  | 1P Revenue | 1P ROAS | vs LW | Status
Meta Ads   | $12,400| $38,400    | 3.1x    | +0.3x | OK ✓
Google Ads | $8,200 | $31,100    | 3.8x    | -0.2x | OK ✓
TikTok Ads | $3,100 | $5,700     | 1.8x    | -0.8x | REVIEW ⚠
Amazon Ads | $4,500 | $19,800    | 4.4x    | +0.1x | OK ✓
Email/SMS  | $800   | $22,000    | 27.5x   | +2.0x | OK ✓
─────────────────────────────────────────────────────────────
Total      | $29,000| $117,000   | 4.0x    |       | MER: 4.0x
Break-even ROAS: 2.33x | Monthly Budget: $125,000 | Paced to spend: $126,500 (+$1,500)

TikTok is flagged: ROAS dropped from 2.6x to 1.8x (below 2.33x break-even). Action: Investigate creative fatigue; pause underperforming ad sets; hold budget before cutting entirely.

Step 5: Make budget reallocation decisions

Use this decision framework when channels are over/under-performing:

When to reduce spend on a channel:

• True ROAS has been below break-even for 2+ consecutive weeks
• First-party ROAS has declined more than 30% with no operational explanation
• Creative has not been refreshed in 30+ days and ROAS is declining (creative fatigue)

When to increase spend on a channel:

• True ROAS is 2x or more above break-even
• First-party ROAS has been stable or increasing for 3+ weeks
• Marginal ROAS at current spend level is still above break-even (not in diminishing returns territory)

Before cutting spend, investigate: 1. Is a top SKU out of stock? Ads running for out-of-stock products waste spend with no ability to convert 2. Did a creative rotation happen? Performance often dips temporarily during algorithm learning phases (50+ optimization events needed) 3. Did a competitor run a major promotion? Temporary CPM/CPC spikes are not structural channel problems 4. Is the tracking pixel firing correctly? A sudden ROAS drop can be a tracking issue, not a performance issue

Best Practices

• Track true ROAS (profit-based), not platform-reported ROAS — platform ROAS using view-through attribution is almost always higher than reality; compute ROAS from your own first-party data
• Monitor MER (Marketing Efficiency Ratio) as a holistic metric — blended MER (total revenue / total marketing spend) is harder to game than individual channel ROAS; a healthy ecommerce MER is typically 3–6x
• Separate new customer acquisition from retention spending — remarketing and email/SMS to existing customers have very different economics than prospecting; track CAC and ROAS separately for new vs. returning customer campaigns
• Build a weekly spend pacing report — track cumulative spend vs. monthly budget by day; a channel at 80% of monthly budget by day 15 will either overspend or dramatically under-deliver in the second half
• Track creative performance, not just campaign performance — one creative often drives 70% of a campaign's ROAS; surface creative-level performance and prioritize winning formats
• Test incrementality before doubling down — high retargeting ROAS may be capturing sales that would have happened anyway; run holdout tests to measure true incremental impact

Common Pitfalls

Related Skills

• @attribution-modeling
• @cost-allocation-analysis
• @unit-economics-tracking
• @ecommerce-budgeting-forecasting
• @financial-analytics-dashboard

Meta Ads Scaling Analysis

Problem Description

A DTC skincare brand has been scaling their Meta advertising budget aggressively over the past six months. Each week they increased spend slightly and tracked how much attributed revenue came back. Now the growth team is debating whether to push spend up another 30% this quarter or hold steady — the CEO wants data to back up the decision.

The analytics lead has compiled weekly snapshots of Meta spend and the revenue attributed to it by their internal model. She suspects they may be hitting diminishing returns but can't quantify it, and needs to present a concrete recommendation with supporting numbers at the next board meeting.

Your job is to write a Python analysis script that models the relationship between spend and revenue, determines where the channel sits on the efficiency curve today, and produces a clear spend recommendation.

Output Specification

Write a Python script named scaling_analysis.py that: 1. Loads the data below as hard-coded inputs 2. Fits a curve to the spend vs. revenue relationship 3. Computes current efficiency and the theoretically optimal spend level 4. Outputs a recommendation (reduce / maintain / increase) with supporting numbers 5. Saves results to scaling_output.json

Run with: python scaling_analysis.py

Input Data

The following 24 weeks of Meta spend and first-party attributed revenue data should be used as input (already in USD):

Marketing Profitability Analysis

Problem Description

A DTC apparel brand runs paid ads on Meta and Google. Their finance lead has flagged that the numbers reported by the ad platforms look suspiciously good, and she suspects the company is actually losing money on some channels when you account for product costs and fulfillment expenses.

The head of growth has pulled together a weekly data snapshot: Meta says it drove $42,000 in revenue from $10,000 in spend, while Google reports $31,500 in revenue from $9,000 in spend. However, their analytics team's own attribution model (based on post-purchase surveys and data from their order management system) attributes $28,000 to Meta and $19,000 to Google during the same period. Total gross revenue for the week across all sources was $95,000. Total marketing spend was $19,000.

Your job is to produce a clear profitability analysis that the finance lead and CMO can use to understand the true performance of each channel, identify which (if any) channels are below break-even, and understand overall marketing efficiency.

Output Specification

Write a Python script named roas_analysis.py that: 1. Takes the data above as hard-coded inputs 2. Computes a full set of efficiency metrics for each channel and the portfolio as a whole 3. Prints a formatted summary showing all computed metrics for each channel and a portfolio-level view 4. Saves the computed metrics to roas_output.json

The analysis should be runnable with python roas_analysis.py and should not require any external data files or API keys.

Multi-Channel Marketing Performance Dashboard

Problem Description

A mid-size DTC home goods brand runs paid advertising across Meta, Google (search and shopping), and TikTok, plus an email marketing program through Klaviyo. Their current reporting is fragmented: the Meta team looks at Meta Ads Manager, the Google team lives in Google Ads, and no one has a unified view. The CMO is preparing for a monthly board meeting and needs a single, coherent picture of marketing efficiency across all channels.

The marketing ops lead has exported two weeks of raw campaign-level data from each platform. She needs a Python analysis script that ingests this data, normalizes it into a unified structure, and produces a weekly marketing performance report. The report needs to be actionable — the CMO should be able to glance at it and immediately know which channels are performing well, which need attention, and whether the overall budget is on pace for the month.

The brand runs both prospecting campaigns (targeting new audiences) and remarketing campaigns (targeting website visitors and past customers), and the CFO has asked that these be reported separately because the economics are very different.

Google search includes both branded terms (people searching the company's own brand name) and non-branded terms (generic product searches). The CFO suspects the branded campaigns are inflating Google's overall numbers.

Output Specification

Write a Python script named marketing_dashboard.py that: 1. Loads the campaign data provided below as hard-coded inputs 2. Normalizes it into a unified data structure 3. Produces a formatted text report saved as weekly_report.txt 4. Saves structured channel metrics to channel_metrics.json

Run with: python marketing_dashboard.py

The report should cover at minimum:

• Portfolio-level summary (total spend, attributed revenue, overall efficiency)
• Per-channel breakdown comparing this week vs last week
• Prospecting vs remarketing performance split
• Google branded vs non-branded performance split
• Budget pacing status (each channel has a monthly budget cap provided below)
• A note on which channels warrant attention

Input Files (optional)

The following campaign data covers two weeks (Week 1 = days 1–7, Week 2 = days 8–14 of the current month). Extract and use this data before beginning.

=============== FILE: inputs/campaign_data.csv =============== week,platform,campaign_type,campaign_name,spend,impressions,clicks,platform_conversions,platform_revenue,first_party_revenue,monthly_budget 1,meta,prospecting,Meta_Prospecting_Broad,18200,1450000,8700,142,28400,19800,80000 1,meta,remarketing,Meta_Retargeting_Visitors,6400,320000,4100,98,17640,14200,30000 1,google,branded_search,Google_Brand_Search,3800,210000,12400,185,29600,27100,20000 1,google,nonbranded_search,Google_NonBrand_Search,7200,380000,6200,74,14800,9600,35000 1,tiktok,prospecting,TikTok_TopFunnel_UGC,9100,2100000,11200,87,13050,7900,40000 1,klaviyo,remarketing,Email_Weekly_Newsletter,1100,0,0,64,9600,9100,5000 2,meta,prospecting,Meta_Prospecting_Broad,19100,1520000,9100,149,29800,21400,80000 2,meta,remarketing,Meta_Retargeting_Visitors,6700,335000,4300,103,18540,15100,30000 2,google,branded_search,Google_Brand_Search,3900,215000,12700,191,30560,28200,20000 2,google,nonbranded_search,Google_NonBrand_Search,7400,392000,6400,76,15200,9900,35000 2,tiktok,prospecting,TikTok_TopFunnel_UGC,9400,2180000,11600,91,13650,8200,40000 2,klaviyo,remarketing,Email_Weekly_Newsletter,1100,0,0,67,10050,9500,5000