SuperAPI AI Directory

metrics-briefing

General & other

View repository

Updated 10-15 20:13

About this skill

Metrics Briefing

Metrics analysis through CEO Bible Section G — 9 categories. PostHog covers some; the rest come from conversations, CRM, and the founder's observations.

Principles (Bible G)

  1. All metrics by target wedge. Segment first, aggregate second.
  2. Behavior > opinions. Usage, money, time spent > compliments.
  3. Pre-PMF metrics answer 5 questions:
    • Are we talking to the right people?
    • Do they share the same painful problem?
    • Do they get value quickly?
    • Do they come back?
    • Can we win them again without heroics?
  4. MRR, CAC, LTV are guardrails, not the steering wheel. Until there is repeatability.

9 CEO Metrics Categories

1. Market Learning (MANUAL — ask the founder)

Metric Definition PostHog?
Qualified conversations/week 20+ min conversations with target roles; explored problem, trigger, workaround, buyer, urgency No — track manually, target 8-12/week
Repeated pain rate Same problem + trigger + workaround recur / total qualified conversations No — from conversation notes
Objection concentration Losses/stalls by top 3 reasons / all losses No — from CRM/notes

If conversations < 8/week -> flag hard. This is the main leading indicator.

2. Product Value (PostHog: partial)

Metric Definition PostHog
Median TTFV Time from signup to first meaningful success user_journey_milestone WHERE milestone='First Value Moment', property timeSincePrevious
Activation rate Target accounts hitting critical event sequence within X days funnel_next_stage_available WHERE currentStage='first_value_delivered' / total signups
Outcome attainment Activated accounts achieving promised business result funnel_next_stage_available: first_value_delivered -> return_usage conversion rate

3. Usage / Engagement (PostHog: yes)

Metric Definition PostHog
Retained active target accounts (week 4/8) Accounts still performing core workflow user_journey_milestone WHERE milestone='Weekly Active'
Workflow depth Core workflow at target frequency, multiple use cases core_query_performed frequency, deep_workflow_started, searchMode breakdown
Second-bite rate Users repeating core action within X days funnel_next_stage_available: first_value_delivered -> return_usage

4. Sales / Pipeline (MANUAL — CRM/conversations)

Metric Definition PostHog?
ICP pipeline created New opportunities from target wedge per week No — CRM
Meeting -> opportunity New opps / qualified first meetings No — CRM
Opportunity -> win rate Closed-won / closed in target wedge No — CRM
Median sales cycle Days from first meeting to signature No — CRM

5. Retention / Expansion (PostHog: partial)

Metric Definition PostHog
Healthy account rate Paid accounts with green usage + outcome + active champion funnel_next_stage_available: regular_usage+ stage, organizationId grouping
Expansion pull rate Healthy accounts requesting more seats/team rollout funnel_next_stage_available WHERE nextStage='team_invitation_sent'
Gross logo retention Paying logos retained once cohorts mature Not fully — partially through return/regular/power stages

6. B2C -> B2B Conversion (PostHog: yes — key category)

Metric Definition PostHog
Target-signup rate Signups from target roles/domains / total signups user_journey_milestone milestone='Account Created', filter by email_verified + has_primary_integration_connected
Activated -> PQL rate Activated users hitting enterprise triggers power_user stage + deep_workflow_started + organizationId
PQL -> sales conversation PQLs that book real sales call Partial — track manually
Same-company emergence Active companies with >1 user or team invite funnel_next_stage_available nextStage='team_invitation_sent', organizationId grouping

7. Team Execution (MANUAL)

Metric Definition PostHog?
Experiment cycle time Days from hypothesis to readout No — from hypothesis tracker
Hypothesis-linked roadmap % Work tied to hypothesis / total work No — from backlog
Belief-change rate Experiments ending in keep/kill/reframe / total No — from weekly reviews

8. Unit Economics (MANUAL)

Metric Definition PostHog?
Net burn / runway Monthly cash burn; cash / burn No — finance
Core workflow contribution margin (Revenue - variable delivery costs) / revenue No — finance
Revenue concentration Top 3 customers / total ARR No — finance

9. PMF Detection (MIX)

Metric Definition PostHog
Wedge repeatability score Last 10 wins matching same ICP/problem/buyer/onboarding / 10 No — manual assessment
Durable revenue ratio Revenue from accounts past trial + healthy >90 days / total Partial — through regular_usage/power_user stages
Organic pull index Target-cohort referrals, unsolicited inbound, expansion requests Partial — public_output_viewed, output_sharing_enabled as signals

PostHog Implementation Layer

Core Funnel (funnel_next_stage_available events)

signup_started -> account_created -> email_confirmed
  -> workspace_created -> first_value_moment
  -> first_primary_resource_added -> primary_resource_processed
  -> first_key_action_executed -> first_value_delivered
  -> return_usage -> regular_usage -> power_user
  -> team_invitation_sent

4 Analytics Layers

  1. Anonymous visitor: $initial_utm_source, $initial_referring_domain, $initial_current_url
  2. User identity: $identify, email_verified, has_primary_integration_connected
  3. Product value: workspace_create_started, primary_resource_action, core_query_performed, core_action
  4. Account/company: organizationId (on all events) — B2B grouping

Attribution Model

  • First-touch (person.$initial_*): "What created initial interest?" -> content, SEO, founder content
  • Last-touch: "What triggered conversion?" -> pricing page, CTA performance
  • Right question: "Which source produces the most retained companies?" — apply to activation, PQL, paid, retained

PostHog Query Recipes

CEO Morning Briefing:

# Signups yesterday
posthog__query-generate-hogql-from-question question="How many user_journey_milestone events with milestone Account Created happened yesterday?"

# Funnel transitions yesterday
posthog__query-generate-hogql-from-question question="Show all funnel_next_stage_available events from yesterday, count by currentStage to nextStage"

# Active companies this week
posthog__query-generate-hogql-from-question question="How many distinct organizationId values had any events in the last 7 days?"

Weekly Dashboard:

# Funnel health (all stages)
posthog__query-generate-hogql-from-question question="Show funnel_next_stage_available events in the last 7 days, count unique users for each currentStage to nextStage pair"

# TTFV
posthog__query-generate-hogql-from-question question="What is the average timeSincePrevious for user_journey_milestone where milestone is First Value Moment, last 30 days?"

# Attribution: source -> activation
posthog__query-generate-hogql-from-question question="For users who reached first_value_delivered in funnel_next_stage_available last 30 days, what are the top initial UTM sources?"

# Account-level: companies with multiple users
posthog__query-generate-hogql-from-question question="Which organizationId values have more than 1 unique user performing events in the last 30 days?"

Vanity Metrics (DO NOT report as progress)

  • Total traffic, total signups, total "active users" mixed
  • Feature count shipped
  • Total pipeline across all segments
  • PR hits, likes, followers
  • Total MRR without durability/health filters
  • B2C DAU that never clusters into accounts
  • "AI usage" with no evidence of business outcome

Metrics That Matter LATER (after PMF)

CAC payback, LTV/CAC, quota attainment, forecast accuracy, mature NRR, Rule of 40, hiring efficiency.

Telegram Format (max 15 lines)

Metrics for [period]:

Learning: X conversations (target 8-12), pain rate Y%
Product: TTFV Xh, activation Z%
Usage: retained W4 X%, second-bite Y%
B2C->B2B: X signups -> Y activated -> Z PQL, same-co X
Pipeline: X opps, win rate Y%
PMF: repeatability [low/growing], organic pull [none/emerging]
Burn: $X/mo, runway Y months

Conclusion: [what this means for the current stage]
Action: [concrete action]

Follow output preferences from USER.md (language, format, platform constraints).