Klaviyo Analyst

Klaviyo Marketing Analyst

Expert-level guidance for Klaviyo email and SMS marketing from the marketing operations and analyst perspective — auditing, building, and optimizing flows, segments, campaigns, and integrations.

> For API integration, SDK usage, event tracking implementation, webhooks, and developer patterns, see the klaviyo-developer skill.

Install

Step 1 — Install the skill

git clone https://github.com/thatrebeccarae/claude-marketing.git && cp -r claude-marketing/skills/klaviyo-analyst ~/.claude/skills/

Step 2 — Connect the Klaviyo MCP (recommended)

This skill is designed around Klaviyo's official MCP server. Connect Claude to your Klaviyo account through OAuth — no local API key, no Python install. The setup path depends on which Claude surface you're using.

For Claude Chat or Claude Cowork — use the Connector Directory

Klaviyo is listed in Claude's Connector Directory (announced as part of the expanded Klaviyo + Anthropic integration on 2026-05-07). Setup takes about two minutes:

1. Open Claude → Settings → Connectors → Browse Connectors
2. Search for Klaviyo
3. Click Connect and authenticate

Plan requirement: Connectors are available on Claude Pro, Max, Team, and Enterprise plans. Free plan users will need to use Claude Code or the local install path below.

For Claude Code — register the remote MCP

claude mcp add klaviyo --transport http https://mcp.klaviyo.com/mcp

For audit-only sessions, append ?read-only=true to disable all write tools at the protocol layer:

claude mcp add klaviyo --transport http "https://mcp.klaviyo.com/mcp?read-only=true"

For local install (CI containers, offline development)

claude mcp add klaviyo -e PRIVATE_API_KEY=pk_... -e READ_ONLY=true -- uvx klaviyo-mcp-server@latest

Klaviyo-side requirement: Owner, Admin, or Manager role to authorize the connection.

The MCP exposes 40+ tools across Accounts, Campaigns, Catalogs, Events, Flows, Groups, Profiles, Reporting, Templates, and Translations. See [REFERENCE.md](REFERENCE.md#mcp-server-reference) for the full tool inventory.

Step 3 — Fallback: local CLI scripts (optional)

If you need scripted CLI access for CI, cron, or headless reporting — install the Python requirements and provide a KLAVIYO_API_KEY. See the [Scripts (Fallback)](#scripts-fallback-local-cli) section below.

Core Capabilities

Flow Auditing & Optimization

• Audit existing flows against best practices (welcome series, abandoned cart, post-purchase, winback, browse abandonment, sunset)
• Identify revenue leakage from missing or underperforming flows
• Recommend split tests, timing adjustments, and conditional logic improvements
• Review flow filters and trigger conditions for accuracy

Segmentation Strategy

• Build RFM-based segments (Recency, Frequency, Monetary)
• Design engagement tiers: Active (0-30d), Warm (31-90d), At-Risk (91-180d), Lapsed (180d+)
• Create predictive segments using Klaviyo's predictive analytics (CLV, churn risk, next order date)
• Suppress unengaged contacts to protect deliverability

Campaign Strategy

• Plan campaign calendars balancing promotional and value content
• A/B testing frameworks for subject lines, send times, content blocks
• Dynamic content personalization using profile properties and catalog data
• SMS campaign compliance (TCPA, quiet hours, opt-in requirements)

Deliverability Management

• Monitor and diagnose deliverability issues (bounce rates, spam complaints, inbox placement)
• Warm-up strategies for new sending domains/IPs
• Authentication setup: SPF, DKIM, DMARC
• List hygiene practices and sunset flow design

Revenue Attribution & Reporting

• Interpret Klaviyo's attribution model (click-based, 5-day email / 24-hour SMS default windows)
• Build custom dashboards for flow revenue, campaign ROI, list growth
• Benchmark KPIs against industry standards

Event Schema Auditing

• Inventory all tracked metrics by source (built-in integration, custom API, Klaviyo-internal, forms)
• Identify duplicate/redundant events, unused events, and missing standard events
• Assess event property structures for segmentation and personalization accessibility
• Diagnose nested object limitations blocking flow splits and segment conditions

Integration Health Review

• Audit active integrations (e-commerce, review platforms, loyalty, ads)
• Identify stale/dead integrations and orphaned event sources
• Assess flow trigger architecture: direct metric triggers (robust) vs segment-entry triggers via API-synced properties (brittle)
• Evaluate catalog sync health: product coverage, variant handling, freshness

Profile Data Utilization Analysis

• Map profile properties: what's collected vs what's used in segmentation/personalization
• Identify stale calculated properties (set once, never updated)
• Flag properties on events but not synced to profiles (limits segmentation)
• Recommend property flattening strategies for nested data accessibility

Key Benchmarks

| Metric | Good | Great | Warning | |--------|------|-------|---------| | Open Rate (email) | 20-25% | 30%+ | 0.5% | | Spam Complaint Rate | 0.1% | | Flow Revenue % of Total | 30-40% | 50%+ | Cowork-ready: the 4-phase audit is built for unattended execution. With the Klaviyo connector active in Cowork (read-only mode), describe the outcome — "Audit my Klaviyo flows and tell me what's missing," "Pull last week's campaign and flow performance and write me a Monday digest," "Flag every flow with open rates under 20% and write a prioritized fix-it list" — then step away. Cowork pulls the data, runs the analysis, writes the doc, and saves it to the right folder while you're in other meetings.

When asked to audit a Klaviyo account, follow this 4-phase framework:

Phase 1: Status Inventory

1. List/Segment/Flow/Campaign Inventory — Count and categorize all objects by status (live, draft, inactive)
2. Event Schema Inventory — Pull all metrics, categorize by source (built-in, custom API, Klaviyo-internal, forms), flag zero-volume events
3. Integration Inventory — List all active integrations, identify stale connections
4. Custom Profile Property Inventory — Document all custom properties with types and usage

Phase 2: Configuration Audit

1. Per-Flow Configuration Teardown — For each active flow:

• Trigger type and conditions
• Smart Send settings (on/off and implications)
• Exclusion filters (segment membership, event conditions)
• Timing between messages (compare to benchmarks)
• Branching logic (conditional splits, A/B splits)
• Flow duration (first message to last)

1. Campaign Targeting Audit — Segment exclusivity, frequency capping, send-time optimization
2. A/B Testing Methodology Audit — Univariate vs multivariate, KPI alignment, statistical significance practices
3. Deliverability Configuration — Authentication (SPF/DKIM/DMARC), dedicated IP, warmup status

Phase 3: Data Structure Audit

1. Event Schema Health — Check for duplicates, missing standard events, nested objects blocking segmentation
2. Profile Data Utilization — What's collected vs what's used in segmentation/personalization
3. Catalog Sync Health — Product feed freshness, coverage, variant handling

Phase 4: Strategic Recommendations

1. Three-Tier Recommendations — For each finding:

• Finding: What's wrong + evidence from audit data
• Recommendation: What to do (client-facing, plain language)
• Implementation Spec: How to build it (internal SOW with triggers, filters, content brief, timing, testing plan)

1. Sequential Testing Plans — Univariate A/B tests with stat sig framework
2. Quantified ROI — Expected uplift % x current revenue baseline
3. Implementation Roadmap — Phased timeline with dependencies mapped

Industry-Specific Benchmarks

B2B / Wholesale E-Commerce

| Metric | Good | Great | Warning | Notes | |--------|------|-------|---------|-------| | Open Rate | 25-35% | 40%+ | 8% | | Reactivation Rate | 5-10% | 15%+ | When to use these: scripted CLI access for CI/cron, headless reporting jobs, or environments where you can't connect the Klaviyo MCP (e.g., automated pipelines using KLAVIYO_API_KEY env vars). For interactive analysis in Claude Code/Chat/Cowork, prefer the MCP path above — it's faster, OAuth-authenticated, and exposes the same data.

The skill includes utility scripts for data fetching and analysis:

Fetch Klaviyo Data

# List all flows
python scripts/klaviyo_client.py --resource flows

# List campaigns as table
python scripts/klaviyo_client.py --resource campaigns --format table

# Get flow performance report
python scripts/klaviyo_client.py --resource report --report-type flow --id FLOW_ID

# Export metrics to file
python scripts/klaviyo_client.py --resource metrics --output metrics.json

Analyze and Generate Reports

# Full account audit
python scripts/analyze.py --analysis-type full-audit

# Flow gap analysis
python scripts/analyze.py --analysis-type flow-audit

# Segment health check
python scripts/analyze.py --analysis-type segment-health

# Campaign performance comparison
python scripts/analyze.py --analysis-type campaign-comparison --days 30

# Deliverability diagnostic
python scripts/analyze.py --analysis-type deliverability

# Revenue attribution
python scripts/analyze.py --analysis-type revenue-attribution

The scripts handle API authentication, data fetching, and analysis. I'll interpret the results and provide actionable recommendations.

Troubleshooting

MCP connection issues

MCP not appearing in Claude: Verify the MCP is configured in your client and authenticated. For Claude Code, check ~/.claude.json or your project-level MCP config. For Claude Chat, check Settings → Connectors.

OAuth permission denied: Your Klaviyo user role must be Owner, Admin, or Manager. Lower-permission users cannot authorize the MCP.

Write tools failing: Confirm the connection URL does not have ?read-only=true set. If it does, write tools will be disabled by design — switch to a read-write URL for write actions.

Multi-account confusion: The MCP supports multi-account setups. If Claude pulls data from the wrong account, re-authenticate or check your active account in Klaviyo's UI.

Script (fallback) errors

Authentication Error: Verify that:

• KLAVIYO_API_KEY is set as an environment variable or in a .env file
• The key starts with pk_ (private API key, not public)
• The key has read scopes: profiles:read, flows:read, campaigns:read, segments:read, lists:read, metrics:read

No Data Returned: Check that:

• The API key is associated with the correct Klaviyo account
• The account has active flows, campaigns, or segments to analyze
• Filters are correctly formatted (e.g., equals(status,"live"))

Rate Limit Errors: The SDK handles retries automatically (up to 3 retries with 60s max delay). If you still hit limits:

• Reduce concurrent requests
• Add delays between sequential calls
• Check RateLimit-Remaining header

Import Errors: Install required packages:

pip install klaviyo-api python-dotenv pandas

Security Notes

When using the Klaviyo MCP (recommended)

• The MCP uses OAuth — there's no API key to commit, rotate, or leak
• For audit-only work, gate with ?read-only=true on the connection URL to disable all write tools
• Authorization is scoped to your Klaviyo user role (Owner / Admin / Manager) — least-privilege is enforced upstream
• Re-authentication is required if your role changes or the OAuth token expires
• Multi-account setups: confirm the active account before running tools that include write operations

When using the script fallback

• Never hardcode API keys in code or commit them to version control
• Store keys in environment variables or .env files
• Add .env to .gitignore
• Use read-only scopes for analyst tasks — no write access needed
• Rotate API keys periodically in Klaviyo Settings

Data Privacy

This skill accesses aggregated marketing data only. It does not:

• Access personally identifiable information (PII) beyond email/profile aggregates
• Store Klaviyo data persistently
• Share data with external services
• Modify your Klaviyo configuration (read-only operations)

All data is processed locally and used only to generate recommendations during the conversation.

For detailed Klaviyo data model, flow builder reference, segmentation conditions, deliverability, and SMS reference, see [REFERENCE.md](REFERENCE.md).

For API integration, SDK usage, and developer patterns, use the klaviyo-developer skill.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: thatrebeccarae
• Source: thatrebeccarae/claude-marketing
• License: MIT

Install and usage instructions live in the source repository linked above.