Want fast, reliable answers to “What’s our LTV?”, “How healthy is retention?”, and “Where do users drop in our funnel?” This guide gives you copyable Mixpanel recipes with exact UI steps, guardrails, and QA checklists. You’ll set up reusable Saved Metrics/Formulas for LTV/ARPU, build trustworthy Retention reports, and bulletproof your Funnels with a pragmatic QA flow.
- Difficulty: Intermediate (assumes basic Mixpanel familiarity)
- Time: 15–20 minutes per section
- What you’ll build:
- LTV/ARPU via Revenue Analytics templates and a fallback method
- Retention reports with correct windows/frequency and segmented cohorts
- A standard eCommerce funnel with a hardened QA checklist
Before you start, ensure you can look up and validate events in your project’s Events or Live View. From experience, this single habit prevents most misreads.
1) Data prerequisites and governance (do this once)
Get your event schema and governance right so your metrics don’t drift. Mixpanel’s Saved Metrics, Behaviors, and Formulas let you standardize definitions for re-use across reports; they’re managed in Lexicon and behave like first-class metrics (see Mixpanel’s 2023–2025 updates to saved assets in the Saved Metrics and Behaviors docs and the Formulas behave as metrics changelog).
Recommended events and properties (copyable checklist):
Events
- Sign Up (or Activate)
- App Open (or Session Start)
- View Product
- Add to Cart
- Checkout Started
- Purchase (or Order Completed)
Properties (on relevant events)
- revenue_amount: Number (e.g., 39.99)
- currency: String ISO 4217 (e.g., USD)
- order_id: String (for dedup)
- product_id / plan_tier
- channel/source/utm_* (e.g., utm_source, utm_medium)
- device/app_version/platform
- is_test_user: Boolean
Identity
- distinct_id: stable across sessions
- identify() on login/registration; alias() or Simplified ID Merge for anonymous→known
Why it matters:
- Inconsistent revenue types or mixed currencies will corrupt ARPU/LTV.
- Missing identity merges fragment funnels and retention (anonymous vs logged-in treated as separate users).
References: Mixpanel documents governance and reuse in Saved Metrics and Behaviors, and identity strategy in the 2024–2025 ID management docs.
Tip: Maintain a “Sandbox” project for QA and keep is_test_user = true flagged. Mixpanel’s docs note that excluding internal/test users improves report accuracy; see the 2025 guidance in Funnels FAQ.
2) LTV and ARPU recipes
You have two solid paths: a template-driven flow with Revenue Analytics, and a fallback using Saved Metrics + Formulas. The template is fastest; the fallback is flexible when templates aren’t available. Mixpanel introduced Revenue Analytics to tie product usage to outcomes like LTV/ARPU/ROAS; see the 2024 guide in Building Revenue Metrics and the 2023 feature announcement in Introducing Mixpanel Revenue Analytics.
2A) Recipe A — Revenue Analytics template (fastest)
Prerequisites: Your Purchase event carries numeric revenue_amount and currency; acquisition properties (e.g., utm_*) exist if you plan to break down by channel.
Steps:
- Go to Boards → Create New Board → choose a Revenue template (eCommerce or SaaS).
- When prompted, map your revenue event(s). Confirm:
- revenue_amount is a Number
- currency uses ISO codes (e.g., USD)
- In the template, enable metrics:
- LTV (per user over time)
- ARPU (Total Revenue / Active Users)
- ROAS (requires ad spend ingestion via events/integration/warehouse)
- Add breakdowns/filters:
- Cohort by signup month
- Break down by utm_medium, plan_tier, or platform
- Save to a Board and convert key metrics into Saved Metrics for reuse in other reports.
Verification:
- Spot-check a few users in Events/Live View for correct revenue_amount and currency.
- Compare total revenue to a finance/bookings export for the period. Expect small differences from timing/attribution; large deltas mean missing events or duplicates.
Troubleshooting:
- ROAS shows 0: Verify ad spend ingestion or connectors per the 2024–2025 Warehouse Connectors docs.
- Odd LTV swings: Check identity merges and that refund/cancel events aren’t mistakenly included as positive revenue.
2B) Recipe B — Fallback with Saved Metrics + Formulas (flexible)
Use this when templates aren’t available or your schema is custom.
Step 1 — Create Saved Metrics:
Copyable definitions
Saved Metric: Total Revenue
- Report type: Insights
- Query: Sum of revenue_amount on event = Purchase
- Filters: Exclude is_test_user = true; currency = USD (or standardize as needed)
Saved Metric: Active Users
- Report type: Insights
- Query: Unique users (DAU/WAU/MAU as needed)
- Filters: Exclude is_test_user = true
Step 2 — Create an ARPU Saved Formula:
Copyable formula
ARPU = Total Revenue / Active Users
Step 3 — Approximate LTV (two ways):
- Cohort method (preferred): Create cohorts by signup month, then visualize Cumulative Sum of revenue per user over N days/weeks/months. This reflects realized LTV, not a model.
- Ratio proxy (use cautiously): LTV ≈ ARPU ÷ monthly churn rate (or ARPU × average lifespan). This is a model approximation and sensitive to churn accuracy.
Tip: Save these as Saved Metrics/Formulas so they behave like first-class metrics across reports (see the 2023–2024 notes in the Saved Formulas changelog).
Verification & QA:
- Validate revenue_amount’s type is numeric and currency is standardized; see 2024 Revenue Analytics guide.
- Sanity-check ARPU by comparing to Total Revenue / unique users from your source-of-truth period.
3) Retention recipes
Retention hinges on three choices: initial event, return event, and window type. Mixpanel’s 2025 docs cover these semantics in the Retention report.
Copyable configuration (baseline):
Report: Retention
Initial event: Sign Up (or Activate)
Return event: App Open (or Purchase for monetization retention)
Interval: Week (choose Day/Week/Month based on your cadence)
Window: Rolling (use Calendar when you need true calendar months)
Frequency: at least 1 interval
Breakdown: platform or plan_tier
Filters: Exclude is_test_user = true
How to build it:
- Go to Reports → Retention.
- Set Initial = Sign Up; Return = App Open (or Any Event for broad engagement, but be consistent).
- Choose Interval = Week; Window = Rolling.
- Set Frequency = “at least 1 interval.”
- Add a breakdown by platform or plan_tier; save.
Key decisions and pitfalls:
- Initial vs Return: Changing Return from App Open to Any Event will materially shift numbers; compare deliberately. Mixpanel explains these levers in the 2025 Retention docs.
- Rolling vs Calendar: Rolling “Month” approximates 30-day windows; use Calendar windows if you require true calendar months. A 2024 Mixpanel community thread clarifies the “<1 month” semantics as the first 30 days from initial event; treat it as approximate unless Calendar is selected; see Understanding “<1 month” in Retention.
- Frequency: “At least N intervals” measures multi-period stickiness; increasing N tightens the bar for how many periods users return.
Verification:
- Drill into example users to validate they performed the Return event within the chosen window.
- Toggle Rolling vs Calendar and confirm the change behaves as expected.
Troubleshooting:
- Retention too high/low: Confirm Return event choice; switching to Any Event often inflates retention.
- Cohort sizes seem off: Check test-user filters and identity merges to avoid splitting the same person.
4) Funnel QA recipes (with a production-ready checklist)
We’ll build a standard eCommerce funnel and then harden it with QA. Mixpanel’s 2024–2025 funnel references explain steps, conversion windows, exclusions, and comparisons; start with the Funnels overview/advanced/quickstart and the consolidated Funnels FAQ.
4A) Build the funnel
Copyable configuration (baseline eCommerce funnel):
Report: Funnels
Steps (Specific order)
1) View Product
2) Add to Cart
3) Checkout Started
4) Purchase
Conversion window: 7 days (tune to your business; test 1d/7d/30d)
Ordering: Specific order (switch to Any order if flow allows)
Exclusions: Exclude is_test_user = true
Breakdowns: device, app_version, utm_medium, or cohort
Time compare: Previous period (or same period last year)
Annotations: Release or promo dates
How to build it:
- Reports → Funnels → Add steps in specific order.
- Set Conversion window = 7 days.
- Exclude is_test_user = true (Filter). Optionally exclude refund/cancel events between steps using “did not do” conditions; note exclusions can’t be the first/last step, per the 2025 Funnels FAQ.
- Add breakdowns by device/app_version or channel.
- Enable Time to convert and Time compare; annotate release dates.
4B) QA checklist (run this every time)
Event correctness
- For each step, confirm the event actually fires with required properties (e.g., product_id on View Product, order_id and revenue_amount on Purchase) using Events or Live View.
Window and ordering sensitivity
- Re-run with 1d vs 7d vs 30d windows; note sensitivity. If results swing wildly, your buying cycle might be long or late events are missing.
- Test Specific order vs Any order; if Any order jumps, your real world flow is not strictly linear.
Exclusions and bots
- Filter out is_test_user = true. If bots are suspected, filter via user agent/IP patterns if you capture them.
- Add an exclusion step for refund/cancel between Checkout Started and Purchase if those exist and would invalidate purchases.
Breakdowns and regression checks
- Break down by app_version or platform to localize issues to a release.
- Use Time compare and annotations to spot regressions tied to launches.
Identity merges
- If counts shift unexpectedly after login changes, review identity merge behavior. Misconfigured identify()/alias() splits pre- and post-login steps. Mixpanel’s 2024–2025 ID management docs explain merge effects.
Latency expectations
- Mixpanel reports are near real-time for UI exploration, but export/Data Pipelines have separate SLAs. Treat UI as near live and exports as potentially delayed; see 2024–2025 details across Data Pipelines and replay SDK docs.
5) Reusable Saved assets: Metrics, Behaviors, Formulas
Lock in your best work so teams don’t rebuild logic differently.
What to save:
- Metrics: Total Revenue, Active Users, Purchases, Checkout Starts.
- Formulas: ARPU = Total Revenue / Active Users; Conversion Rate = Purchases / Sessions (if you track sessions).
- Behaviors: “Engaged 3+ times in 7 days,” “Viewed ≥3 products then Purchased.”
How to save (click path):
- In a report, open the metric or behavior → ellipsis → Save Metric / Save Behavior. Name it clearly; manage in Lexicon → Metrics/Behaviors. Saved formulas behave like normal metrics across reports per the 2023 update in Formulas behave as metrics. See also the 2023–2024 Saved Formulas changelog.
Governance tips:
- Add descriptions with definitions and owners in Lexicon.
- Version changes: append a suffix (v2) and note what changed.
6) Troubleshooting playbook (by scenario)
If ARPU is zero or tiny
- Check that revenue_amount exists and is numeric; verify currency standardization. Revisit inputs per the 2024 Revenue Analytics guide.
- Confirm Active Users denominator isn’t unexpectedly large (e.g., including bots or internal users).
If LTV looks implausibly high/low
- Inspect cohorts by signup month with cumulative revenue; the curve should grow then flatten. If flat from day 0, revenue events aren’t linking to users (identity or missing properties).
- Ensure refunds/credits are handled appropriately (excluded or negative amounts as intended).
If retention jumps after a config tweak
- Review Return event choice (App Open vs Any Event) and Window (Rolling vs Calendar). The 2025 Retention docs outline how each lever behaves.
If funnel conversion dropped “overnight”
- Use breakdowns by app_version/platform; check annotations around release times.
- Verify each step event still fires (SDK changes can rename payloads). Re-run sensitivity tests on conversion window and ordering.
If numbers don’t match finance
- Expect timing and attribution differences, but >10–15% gaps warrant a deep dive. Compare Mixpanel Total Revenue vs a finance export for the same period. Look for missing order_id, duplicate events, or currency mix-ups.
7) Copyable dashboard blueprint (put it all together)
Build a Board that execs and PMs can trust.
Board: Growth & Monetization Health
Top row (KPIs)
- ARPU (Saved Formula)
- LTV (from Revenue template or cohort cumulative view)
- Purchases (Saved Metric)
- Active Users (Saved Metric)
Middle row (Retention)
- Retention: Sign Up → App Open, Week, Rolling, at least 1 interval; breakdown by platform
- Retention (Monetization): Sign Up → Purchase, Month, Calendar; breakdown by plan_tier
Bottom row (Funnel & Diagnostics)
- Funnel: View Product → Add to Cart → Checkout Started → Purchase; 7d window; Specific order; exclude is_test_user
- Time to convert: same funnel, distribution chart
- Annotations: release and promo dates
Save each chart; pin the Board. Use Lexicon descriptions so “what this means” is always visible.
Sources and further reading
- Mixpanel Docs, 2024–2025: Revenue setup and definitions in the Building Revenue Metrics guide
- Mixpanel Blog, 2023: Product announcement for templates in Introducing Mixpanel Revenue Analytics (LTV, ARPU, ROAS)
- Mixpanel Docs, 2025: Retention semantics and configuration in the Retention report documentation
- Mixpanel Docs, 2024–2025: Funnels configuration and edge cases in the Funnels overview and Funnels FAQ
- Mixpanel Docs, 2024–2025: Reuse and governance in Saved Metrics and Behaviors and Saved Formulas changelog
- Mixpanel Docs, 2024–2025: Identity strategy in ID management and export/latency context in Data Pipelines
You now have a reproducible path to LTV/ARPU, retention, and funnel QA in Mixpanel—with saved assets, clear verification, and guardrails against common pitfalls. Bookmark this and adapt the copyable blocks to your schema and cadence.