analytics-interpretation

Analytics Interpretation

Read product metrics, diagnose where the funnel leaks, and decide what to do about it. Works on App Store Connect data, third-party analytics (Mixpanel, Amplitude, Firebase, RevenueCat), or raw numbers the user shares.

Responsibility

Owns: Metric definitions, benchmarks by app type, App Store funnel interpretation, AARRR diagnostics, cohort retention reading, LTV / payback math, invest-iterate-pivot-sunset decision.

Does NOT own: Pricing changes (see monetization-strategy), ASO and product-page copy fixes (see app-store-marketing), implementing analytics SDKs, A/B test infrastructure.

North-Star Metrics by Model

The right metric to obsess over depends on monetization. Don't average across models.

Model	North star	Supporting
Free + ads	DAU, sessions/day, ad revenue per DAU	D1/D7/D30 retention, session length
Freemium (one-time unlock)	Free→paid conversion rate	D7 retention of free users, time-to-conversion
Subscription	Trial→paid conversion, monthly churn, LTV	Trial start rate, M1/M3/M12 retention, payback period
Paid upfront	Downloads, refund rate (<5%), rating	Organic vs paid mix

App Store Funnel (App Store Connect)

Impressions
  ↓ Tap-Through Rate (TTR) = Product Page Views / Impressions
Product Page Views
  ↓ Conversion Rate (CVR) = Downloads / Product Page Views
Downloads
  ↓ D1 retention
Day 1 Active
  ↓ D7
Day 7 Active
  ↓ D30
Day 30 Active
  ↓ paywall conversion
Paying users

Step-level reads

Stage	Good	Average	Poor	First fix when poor
TTR	> 8%	4-8%	< 4%	Icon, title, subtitle, search rank
CVR	> 40%	25-40%	< 25%	First 3 screenshots, App Preview, ratings
D1 retention	> 35%	20-35%	< 20%	Onboarding length, value-on-first-launch, screenshot honesty
D7 retention	> 20%	10-20%	< 10%	Reason to come back (notifications, streaks), core loop depth
D30 retention	> 10%	5-10%	< 5%	Feature depth, progression, content cadence
Free→paid	> 5%	2-5%	< 2%	Paywall placement, trial design, price

These ranges are rough industry medians. Your category may differ — use your own historical baseline as the real benchmark.

AARRR Funnel — Diagnostic Order

Walk the funnel top-to-bottom. Fixing a leaky bucket below a leakier one upstream gives nothing.

1. Acquisition — top sources, organic vs paid mix, CPA vs LTV.
2. Activation — % of new installs reaching the "aha action" in session 1.
3. Retention — D1/D7/D30, DAU/MAU ratio (15-30% is healthy daily engagement).
4. Revenue — paywall touch rate, trial start rate, trial→paid, ARPU vs ARPPU.
5. Referral — organic multiplier, share rate, rating prompt acceptance.

If retention is broken, do not invest in acquisition. You will pour water into a sieve.

Cohort Reading

              M0     M1     M2     M3     M4     M5
Jan cohort   100%   62%    55%    50%    48%    46%
Feb cohort   100%   58%    51%    46%    44%     —
Mar cohort   100%   65%    59%    54%     —      —
Apr cohort   100%   70%    63%     —     —      —
May cohort   100%   68%     —      —     —      —

What to look for:

• M0→M1 drop — biggest single drop. Industry-typical 30-50% churn on subscription apps. >50% means trial experience or onboarding is broken.
• Curve flattening — should plateau by M3-M5. That plateau is your "natural retention floor" and drives LTV.
• Cohort improvement — compare same column across cohorts. Apr M1 (70%) > Jan M1 (62%) means recent product changes worked.
• Scheduled cliffs — M1 cliff often means annual non-renewals; M12 cliff is annual renewal decisions.

Significance rule of thumb (no formal stats)

• < 3 percentage points cohort-over-cohort: noise, do nothing.
• 3-10 pp: meaningful, keep the change.

• 10 pp: major signal, double down.

For real significance use a proper test — but don't paralyze on small samples either.

LTV and Payback

ARPU      = total revenue / total users (free + paid)
ARPPU     = total revenue / paying users only
LTV       = ARPPU × average subscriber lifetime (months)
Payback   = CPA / monthly net ARPPU
Healthy   = LTV ≥ 3× CPA, payback ≤ 6 months for indie subscription apps

Common mistakes:

• Using ARPPU when you mean ARPU (or vice versa) — they differ by 5-20×.
• Computing LTV from a 30-day window — long-tail subscriptions accrue most LTV after month 6. Use cohort-based LTV, not snapshot.
• Forgetting Apple's cut. $9.99/mo at 15% Small Business Program = ~$8.49 net.
• Treating annual subscribers as "12× monthly" — they churn differently and pay upfront.

Common Misinterpretations

Symptom	What people say	What it usually is
DAU rising, retention flat	"We're growing!"	You're acquiring faster than churning. Fragile. Check D30 cohort trend.
MAU > DAU × 30	"Lots of monthly users"	Most are zombies who launched once. Track DAU/MAU ratio.
Conversion up after price cut	"Lower price worked"	Revenue per visitor may be down. Compute price × CVR, not CVR alone.
Retention "improved" after release	"Feature X works"	Could be cohort mix change, seasonality, or marketing push. Compare same-source cohorts.
Paid acquisition CPA looks low	"Channel is profitable"	Compare CPA to paid-user LTV, not blended LTV. Paid users churn faster than organic.
Refund rate under 1%	"Users love it"	App Store hides involuntary refunds; check chargebacks separately for subscription.
Vanity install number	"We hit 100k installs"	Installs without retention is a cost line, not an asset.

Diagnostic Decision Trees

Low impressions (established app)

Ranking for any keywords?
├── No  → ASO problem. See app-store-marketing.
└── Yes → High-volume keywords?
    ├── No  → Target higher-volume terms.
    └── Yes → Top 10 placement?
        ├── No  → Improve rank: more ratings, better CVR.
        └── Yes → Expand keyword set or add locales.

High impressions, low TTR (< 4%)

Icon distinctive at thumbnail size?
├── No  → Redesign, A/B test 3 variants.
└── Yes → Title clear and keyword-rich?
    ├── No  → "Brand - Value Keyword" format.
    └── Yes → Subtitle specific?
        ├── No  → Concrete benefit, not generic.
        └── Yes → Differentiated from category competitors?

Good downloads, bad D1 retention (< 25%)

Onboarding completion > 70%?
├── No  → Cut steps, add skip.
└── Yes → User reaches "aha" in session 1?
    ├── No  → Restructure first run to show core value first.
    └── Yes → Crashes / slow launch?
        ├── Yes → Fix stability before anything else.
        └── No  → Did screenshots overpromise vs reality?

Good retention, low revenue (< 3% conversion)

Users see the paywall?
├── No  → Add paywall touchpoints (feature gates, usage limits).
└── Yes → Compelling paywall (value, comparison, social proof)?
    ├── No  → Redesign.
    └── Yes → Price right?
        ├── Too high  → Test lower or cheaper tier.
        ├── Too low   → Users may not perceive value; test higher.
        └── Right     → Trial showcasing premium features?

Invest, Iterate, Pivot, Sunset

After walking the funnel, pick one path:

Path	Signals	Action
Invest	D7 > 40%, organic growth, requests for features, conversion improving, 4.5+ rating	Increase build cadence, consider paid acquisition, expand platforms
Iterate	D7 20-40%, mixed feedback, stable but unspectacular conversion	Find what retained users do differently; make all users do that. A/B onboarding and paywall.
Pivot	D7 < 20% after 3+ iterations; engagement concentrated on an unexpected feature	Rebuild around what users actually do, not what you planned
Sunset	Declining across the board, no organic growth despite iteration, opportunity cost too high	Maintenance mode, consider open-sourcing or selling, redirect energy

Sunsetting is not failure. Most successful indie portfolios shipped several apps before one worked.

Output Format (when asked for a health report)

# Analytics Health Report — [App]

App type: [free/freemium/subscription/paid]
Stage: [pre-launch/early/growing/mature]
Period: [date range]

## Funnel
| Stage | Metric | Value | Status | First-action |
| ... | ... | ... | green/yellow/red | ... |

## Primary bottleneck
[Stage] — [one-sentence cause].

## Recommendations (priority order)
1. [critical fix] — expected impact
2. [high] — expected impact
3. [medium] — expected impact

## Verdict
Invest / Iterate / Pivot / Sunset — [2 sentences]

References

• app-store-marketing — fixing TTR/CVR, screenshot rework, ASA readiness gates.
• monetization-strategy — pricing/tier changes that follow from low-revenue diagnosis.
• storekit — instrumenting trials and renewals if data is missing.