100 Candidate Sub-Metrics → Ranked → Top 30 → Proposed 20

Starting from all available data sources (production DB, Stripe, PostHog, GSC, PSI, Rollbar, Postmark, GitHub), generating 100 candidates, ranking by decision utility × collectibility × uniqueness.

Scoring Criteria

Each metric scored 1-5 on three axes:

Decision utility (D): Would this metric change a decision this month? (5 = definitely, 1 = nice to know)
Collectibility (C): Can we automate collection today? (5 = SQL/API query, 1 = requires new infrastructure)
Uniqueness (U): Does it tell us something the 8 heroes don't? (5 = completely new signal, 1 = redundant)

Composite score = D × 0.5 + C × 0.3 + U × 0.2 (weighted toward decision utility)

The 100 Candidates

Revenue & Pricing (15)

#	Metric	Source	D	C	U	Score	Current value
1	New MRR added this month	Stripe	5	5	5	5.0	—
2	Churned MRR lost this month	Stripe	5	5	5	5.0	—
3	Revenue per listing (MRR / paid listings)	Stripe+SQL	4	5	4	4.3	~$14.14
4	Plan tier distribution (% monthly/quarterly/annual/premium)	Stripe	4	5	3	4.1	15.6%/13.9%/44.2%/17.9%
5	Net new subscriptions/week	Stripe	4	5	3	4.1	—
6	Checkout completion rate	PostHog	4	3	4	3.9	—
7	Premium upgrade rate (standard → premium)	Stripe	4	4	4	4.0	—
8	Coupon redemption count/month	Stripe	3	5	3	3.6	1
9	Average revenue per user (ARPU)	Stripe	3	5	3	3.6	~$18/mo
10	Failed payment amount/month	Stripe	4	5	4	4.3	—
11	Failed payment recovery rate	Stripe	4	4	5	4.2	—
12	Expansion MRR (upgrades)	Stripe	3	4	4	3.5	—
13	Contraction MRR (downgrades)	Stripe	3	4	5	3.7	—
14	Free-to-paid conversion rate	SQL	3	4	4	3.5	—
15	Trial conversion rate	Stripe	2	4	3	2.8	—

Churn & Retention (15)

#	Metric	Source	D	C	U	Score	Current value
16	No-inquiry churn rate (0 inquiries → canceled)	SQL	5	4	5	4.7	~80.1% of 0-inq listers churn
17	Got-inquiry churn rate (≥1 inquiry → canceled)	SQL	5	4	5	4.7	~66.3%
18	Churn by billing cycle (monthly/quarterly/annual)	Stripe	4	5	4	4.3	87.7%/65.2%/49.3%
19	30-day cohort retention	SQL	4	4	4	4.0	—
20	90-day cohort retention	SQL	4	4	4	4.0	—
21	Cancellation count/month (from subscription_cancellations)	SQL	3	5	2	3.4	353
22	Reactivation rate (canceled → re-subscribed)	Stripe	3	3	5	3.4	—
23	Days to cancellation (median sub lifetime)	Stripe	3	4	4	3.5	—
24	Involuntary churn rate (payment failure → canceled)	Stripe	3	4	4	3.5	—
25	Lister NPS or satisfaction proxy	Manual	5	1	5	3.8	—
26	Support ticket volume/week	Manual	3	2	4	2.9	—
27	Churn by listing age (first 30d vs 30-90d vs 90d+)	SQL+Stripe	4	3	4	3.7	—
28	Churned lister win-back rate (re-signup within 90d)	SQL	2	3	4	2.7	—
29	Subscription pause rate	Stripe	2	4	3	2.7	—
30	% of churned listers who never got an inquiry	SQL	5	4	4	4.5	—

Demand & Searcher Experience (15)

#	Metric	Source	D	C	U	Score	Current value
31	Thread completion rate (2+ msgs each side)	SQL	5	5	4	4.8	~28%
32	Messages/week	SQL	3	5	2	3.4	4,810
33	New conversations/week	SQL	3	5	3	3.6	~1,750
34	Unique searchers/week	SQL	4	5	3	4.1	~570/wk
35	Searcher return rate (messaged again within 30d)	SQL	4	4	4	4.0	59.3% (if replied to)
36	Mobile vs desktop inquiry rate	PostHog	3	3	4	3.2	—
37	Search-to-message conversion (search → listing view → inquiry)	PostHog	4	2	5	3.6	—
38	Listing view → inquiry rate	PostHog	4	3	4	3.7	—
39	Zero-result search rate	PostHog	3	2	5	3.1	—
40	Avg messages per conversation	SQL	2	5	3	2.9	—
41	Searcher signup → first message (days)	SQL	4	4	4	4.0	—
42	Inquiry-to-booking conversion (if tracked)	SQL	5	2	5	4.0	—
43	Repeat searcher rate (2+ conversations, different rooms)	SQL	3	4	4	3.5	—
44	Conversation abandonment rate (started, never completed)	SQL	3	4	4	3.5	—
45	Mobile listing creation rate (PostHog)	PostHog	3	4	3	3.3	18/wk

Fraud & Trust (10)

#	Metric	Source	D	C	U	Score	Current value
46	Median time-to-block (hours)	SQL	4	4	4	4.0	—
47	Conversations with eventually-blocked accounts	SQL	4	4	4	4.0	41/mo
48	Dispute win rate (won / total disputes)	Stripe	4	5	5	4.5	—
49	Dispute loss amount/month	Stripe	3	5	3	3.6	$35
50	Blocked users/month	SQL	3	5	3	3.6	—
51	Verification request completion rate	SQL	3	4	4	3.5	—
52	Fraudulent listing lifespan (hours from creation to block)	SQL	3	3	4	3.2	—
53	Disposable email signup attempts (when blocking ships)	SQL	3	3	4	3.2	—
54	Reports/flags from users	SQL	2	3	4	2.7	—
55	Stripe Radar risk score distribution	Stripe	2	3	4	2.7	—

Supply & Listing Quality (15)

#	Metric	Source	D	C	U	Score	Current value
56	Time-to-first-inquiry (median days, new listings)	SQL	5	4	5	4.7	27.1% never get one
57	Listing completeness score (title + desc + photos + address)	SQL	4	4	5	4.3	—
58	Paid listings with 0 inquiries in 30d (%)	SQL	4	5	3	4.1	~28%
59	Premium vs standard inquiry rate	SQL	4	4	4	4.0	5.63 vs 4.48/mo
60	Listings per city distribution (concentration)	SQL	3	4	5	3.7	550 cities
61	Multi-listing listers (power hosts)	SQL	3	5	4	3.7	131
62	New listings/week	SQL	3	5	3	3.6	—
63	Avg listing rent	SQL	2	5	3	2.9	$1,470
64	Median listing rent	SQL	2	5	3	2.9	$1,200
65	Listing deactivation rate (active → inactive)	SQL	3	4	4	3.5	—
66	Avg description length	SQL	2	5	3	2.9	—
67	Listings with reviews (%)	SQL	2	4	4	2.9	0 (reviews not used)
68	Room availability fill rate	SQL	3	3	4	3.2	—
69	Stale listings (active >180d, no edits)	SQL	2	4	3	2.7	—
70	Boosted listings (active boosts)	SQL	2	5	3	2.9	0

Organic Growth & SEO (15)

#	Metric	Source	D	C	U	Score	Current value
71	Organic sessions/week	PostHog	4	4	3	3.8	6,070
72	Blog impressions/week	GSC	3	5	3	3.6	9,100
73	Blog clicks/week	GSC	3	5	3	3.6	105
74	Blog CTR	GSC	3	5	3	3.6	~1.2%
75	Indexed blog pages	GSC	3	5	3	3.6	66
76	Sublets page impressions/week	GSC	4	5	5	4.5	— (not tracked separately)
77	Homepage LCP (mobile)	PSI	4	5	2	3.9	14,400ms
78	Search page LCP (mobile)	PSI	3	5	2	3.4	13,300ms
79	TN keyword position	GSC	3	5	3	3.6	#12
80	Total site impressions/week	GSC	3	5	3	3.6	—
81	Organic signup rate (organic visitors → account created)	PostHog	5	3	5	4.4	—
82	Blog → signup conversion	PostHog	4	2	5	3.6	—
83	Landing page impressions (non-blog content pages)	GSC	3	5	4	3.7	—
84	New backlinks/month	Ahrefs	2	4	4	3.0	—
85	Domain rating trend	Ahrefs	2	4	3	2.7	—

Marketplace Dynamics (10)

#	Metric	Source	D	C	U	Score	Current value
86	Inquiries per active listing (supply-demand ratio)	SQL	5	5	5	5.0	~3.4/mo
87	Top 10 cities liquidity (% with inquiry in 14d)	SQL	4	4	5	4.3	—
88	Thin markets count (cities with searchers but <5 listings)	SQL	3	4	5	3.7	—
89	Searcher-to-lister ratio	SQL	3	5	4	3.7	2,289 searchers / 1,454 listers
90	Cross-market search rate (searcher looks in 2+ cities)	SQL	2	3	5	2.9	—
91	Avg days listing stays active	SQL	2	4	3	2.7	—
92	Seasonal inquiry volume (MoM change)	SQL	3	4	3	3.3	—
93	Weekend vs weekday inquiry patterns	SQL	1	4	3	2.0	—
94	Price-to-inquiry correlation by city	SQL	3	3	4	3.2	—
95	Avg listings viewed before first inquiry	PostHog	3	2	4	2.9	—

Product & Operations (5)

#	Metric	Source	D	C	U	Score	Current value
96	Verification completion rate (started → approved)	SQL	3	4	4	3.5	839 requests/30d
97	CI test pass rate	GitHub	2	5	2	2.7	0/10
98	Rollbar active error count	Rollbar	2	5	2	2.7	—
99	Email delivery rate	Postmark	2	4	3	2.7	—
100	Page load errors (client-side JS errors)	PostHog	2	3	3	2.4	—

Ranked Top 30

Sorted by composite score, filtered for uniqueness (removing near-duplicates):

Rank	#	Metric	Score	Category	Why it matters
1	1	New MRR added/month	5.0	Revenue	Half of the MRR waterfall — is acquisition working?
2	2	Churned MRR lost/month	5.0	Revenue	Other half — how much are we leaking?
3	86	Inquiries per active listing	5.0	Marketplace	The supply-demand ratio — marketplace efficiency
4	31	Thread completion rate	4.8	Demand	Do conversations turn into real housing discussions?
5	16	No-inquiry churn rate	4.7	Churn	The biggest churn driver decomposed
6	56	Time-to-first-inquiry (median days)	4.7	Supply	How fast do new listers see value?
7	30	% churned who never got inquiry	4.5	Churn	Validates the inquiry→retention link at churn time
8	48	Dispute win rate	4.5	Fraud	Is the evidence system working?
9	76	Sublets page impressions	4.5	SEO	Core product pages, separate from blog
10	81	Organic signup rate	4.4	SEO	Does organic traffic convert to accounts?
11	3	Revenue per listing	4.3	Revenue	Volume growth vs value growth
12	10	Failed payment amount/month	4.3	Revenue	Revenue at risk from payment failures
13	18	Churn by billing cycle	4.3	Churn	Validates annual tilt strategy
14	57	Listing completeness score	4.3	Supply	Quality predicts inquiry rate
15	87	Top 10 cities liquidity	4.3	Marketplace	Geographic health check
16	11	Failed payment recovery rate	4.2	Revenue	Can we save failed charges?
17	4	Plan tier distribution	4.1	Revenue	Portfolio mix monitoring
18	5	Net new subscriptions/week	4.1	Revenue	Acquisition velocity
19	34	Unique searchers/week	4.1	Demand	Demand-side volume
20	58	Paid listings 0 inquiries 30d (%)	4.1	Supply	Inverse of liquidity, longer window
21	7	Premium upgrade rate	4.0	Revenue	Is Premium delivering enough value to upgrade?
22	19	30-day cohort retention	4.0	Churn	Are new users retaining better over time?
23	35	Searcher return rate	4.0	Demand	Platform stickiness for demand side
24	41	Signup → first message (days)	4.0	Demand	Activation speed
25	46	Median time-to-block	4.0	Fraud	Fraud detection speed
26	47	Conversations with blocked accounts	4.0	Fraud	Harm exposure before detection
27	59	Premium vs standard inquiry rate	4.0	Supply	Does Premium deliver value?
28	77	Homepage LCP	3.9	SEO	Page speed → rankings
29	71	Organic sessions/week	3.8	SEO	Traffic volume trend
30	83	Landing page impressions	3.7	SEO	Non-blog content pages (TN, FAQ, etc.)

Proposed 20 Sub-Metrics

From the top 30, removing 10 that are either too similar to a hero metric or not actionable enough this month:

Removed from top 30:

#20 (0-inquiry 30d %) — redundant with listing liquidity hero
#28 (Homepage LCP) — already collected as feature-level metric
#29 (Organic sessions) — already collected as feature-level metric
#30 (Landing page impressions) — too granular for sub-metric tier
#27 (Premium vs standard inquiry rate) — redundant with #14 (listing completeness predicts this)
#22 (30-day cohort retention) — requires new cohort infrastructure, deferred
#26 (Conversations with blocked accounts) — redundant with hero #7 (searchers affected)
#25 (Median time-to-block) — important but low action frequency
#21 (Premium upgrade rate) — wait for Pro Plan to ship
#17 (Plan tier distribution) — interesting but not decision-driving weekly

The 20:

#	Metric	Question	Source	Why included
1	New MRR/month	Q1	Stripe	Acquisition health — paired with churned MRR tells the growth story
2	Churned MRR/month	Q1	Stripe	Retention cost in dollars — more visceral than %
3	Revenue per listing	Q1	Stripe÷SQL	Efficiency — growing because more listings or better monetization?
4	Failed payment amount/month	Q1	Stripe	Revenue at risk, currently invisible
5	Net new subscriptions/week	Q1	Stripe	Acquisition velocity
6	No-inquiry churn rate	Q2	SQL	Decomposed churn — the inquiry-starved segment
7	% churned who never got inquiry	Q2	SQL	Validates that inquiry→retention link holds at churn time
8	Churn by billing cycle	Q2	Stripe	Validates annual tilt — monthly at 87.7% vs annual at 49.3%
9	Thread completion rate	Q3	SQL	Quality of conversations — did both sides actually talk?
10	Unique searchers/week	Q3	SQL	Demand-side volume — are searchers growing?
11	Searcher return rate (30d)	Q3	SQL	Platform stickiness — do searchers come back?
12	Signup → first message (days)	Q3	SQL	Activation speed — how fast do new users find value?
13	Dispute win rate	Q4	Stripe	Evidence system effectiveness
14	Failed payment recovery rate	Q1	Stripe	Can dunning save revenue?
15	Inquiries per active listing	Q5	SQL	Supply-demand ratio — marketplace balance
16	Time-to-first-inquiry (median days)	Q5	SQL	New lister experience — how fast do they see value?
17	Listing completeness score	Q5	SQL	Quality floor — title + desc + photos + address
18	Top 10 cities liquidity	Q5	SQL	Geographic health — aggregate masks city problems
19	Sublets page impressions	Q6	GSC	Core product pages separate from blog
20	Organic signup rate	Q6	PostHog	Conversion quality — do organic visitors become users?

Distribution by question:

Q1 (Revenue): 5 sub-metrics
Q2 (Churn): 3 sub-metrics
Q3 (Demand): 4 sub-metrics
Q4 (Fraud): 1 sub-metric
Q5 (Supply): 4 sub-metrics
Q6 (Organic): 3 sub-metrics

This gives Q1 (revenue) the most sub-metrics because it's the weakest-covered question today, and Q4 (fraud) the fewest because fraud is currently under control (37 affected searchers/month).

What These 20 Add to the 8 Heroes

Hero metric	What it tells you	What the sub-metrics add
MRR ($36K)	Are we making money?	WHERE the money comes from (new vs churn vs failed), HOW efficiently (per listing)
Churn (25%)	Are we losing people?	WHO is churning (no-inquiry vs got-inquiry), WHY (billing cycle, inquiry status)
Reply rate (51%)	Is the marketplace delivering?	HOW deep (thread completion), HOW fast (signup→message), WHO returns
Fraud searchers (37)	Are people being harmed?	HOW effectively we fight back (dispute win rate)
Liquidity (55.7%)	Is supply useful?	HOW fast (time-to-first-inquiry), HOW balanced (per-listing ratio, geographic), HOW quality (completeness score)
Organic funnel	Is growth sustainable?	WHERE from (sublets vs blog), WHAT quality (signup rate)

Generated 2026-04-11. Scores based on decision utility for a 3-person marketplace team at $36K MRR.

Final Sub-Metrics: 20 Below the Fold

After 100 candidates → ranked → reviewed by strategic subagent + Codex adversarial → revised.

Changes From Draft

Change	Reason	Source
Dropped Net new subs/week	Redundant with New MRR (both measure acquisition)	Both reviewers
Dropped % churned who never got inquiry	Redundant with No-inquiry churn rate (same signal, different angle)	Strategic review
Dropped Failed payment amount	Redundant with recovery rate (recovery is the action metric)	Both reviewers
Dropped Sublets page impressions	Low actionability (what would you do differently?)	Codex
Added Median time-to-block	Fraud detection speed — #1 strategic priority needs >1 metric	Both reviewers
Added Conversations with blocked accounts	Fraud exposure — how much harm before detection	Strategic review
Added Search-to-inquiry conversion	THE marketplace efficiency metric — connects demand to outcomes	Codex (critical missing)
Added 30-day cohort retention	Are product changes improving retention for NEW users?	Strategic review
Swapped Dispute win rate for Fraud blocked accounts/month	$35/month disputes not decision-driving; detection volume is	Strategic review

The Final 20

#	Metric	Question	Source	What decision it informs
Q1: Revenue
1	New MRR added/month	Q1	Stripe	Is acquisition working in dollar terms?
2	Churned MRR lost/month	Q1	Stripe	How much are we leaking? Paired with #1 = net growth
3	Revenue per listing	Q1	Stripe÷SQL	Growing from volume or monetization?
4	Failed payment recovery rate	Q1	Stripe	Is dunning saving revenue? (amount shown as context)
Q2: Churn
5	No-inquiry churn rate	Q2	SQL	What % of inquiry-starved listers churn? (the biggest driver)
6	Churn by billing cycle	Q2	Stripe	Is the annual tilt reducing churn? (87.7% monthly vs 49.3% annual)
7	30-day cohort retention	Q2	SQL	Are product changes improving retention for new cohorts?
Q3: Demand
8	Thread completion rate	Q3	SQL	Do conversations turn into real housing discussions?
9	Unique searchers/week	Q3	SQL	Is demand-side volume growing?
10	Searcher return rate (30d)	Q3	SQL	Do searchers come back? (59.3% if replied to vs 37.9% ghosted)
11	Signup → first message (days)	Q3	SQL	How fast do new users find value? (activation speed)
12	Search-to-inquiry conversion	Q3	PostHog	THE marketplace efficiency metric — do searches turn into messages?
Q4: Fraud
13	Median time-to-block (hours)	Q4	SQL	How fast do we detect and neutralize fraud?
14	Conversations with blocked accounts	Q4	SQL	How much exposure do searchers have before we catch fraudsters?
15	Blocked accounts/month	Q4	SQL	Detection volume — is fraud increasing or decreasing?
Q5: Supply
16	Inquiries per active listing	Q5	SQL	Supply-demand ratio — is the marketplace balanced?
17	Time-to-first-inquiry (median days)	Q5	SQL	How fast do new listers see value? (leading retention indicator)
18	Listing completeness score	Q5	SQL	Quality floor — title + desc + photos + address
Q6: Organic
19	Organic signup rate	Q6	PostHog	Do organic visitors become users? (conversion quality)
20	Blog clicks/week	Q6	GSC	Is content driving traffic? (clicks > impressions for actual engagement)

Distribution

Question	Heroes	Sub-metrics	Total
Q1: Revenue	2 (MRR, Annual %)	4	6
Q2: Churn	1 (Churn rate)	3	4
Q3: Demand	1 (Reply rate) + 1 (Response time)	5	7
Q4: Fraud	1 (Searchers harmed)	3	4
Q5: Supply	1 (Liquidity)	3	4
Q6: Organic	1 (Funnel)	2	3
Total	8	20	28

What Each Sub-Metric Adds to Its Hero

Hero	Sub-metrics	What they add
MRR ($36K)	New MRR, Churned MRR, Revenue/listing, Payment recovery	WHERE money comes from, HOW efficiently, WHAT leaks
Annual % (53.7%)	(shares Q1 tile)
Churn (25%)	No-inquiry churn, Churn by cycle, Cohort retention	WHO churns, WHY, and is it IMPROVING over time?
Reply rate (51%)	Thread completion, Searchers/wk, Return rate, Activation, Search→inquiry	HOW deep, HOW many, WHO returns, HOW fast, WHERE they convert
Response time (34h)	(shares Q3 tile)
Fraud searchers (37)	Time-to-block, Convos with blocked, Blocked/month	HOW fast detected, HOW much exposure, WHAT volume
Liquidity (55.7%)	Inquiries/listing, Time-to-first-inquiry, Completeness	HOW balanced, HOW fast, HOW quality
Organic funnel	Signup rate, Blog clicks	WHAT converts, WHAT engages

UX Design: Below the Fold

Per reviewer feedback, 20 metrics displayed as a compact table (not cards):

One row per metric: name | value | trend arrow (↑↓→) | status dot (green/yellow/red/gray)
Grouped by question with Q-label headers
Thumbs up/down on each row (lightweight — signals which metrics are useful)
No sparklines below the fold — sparklines live in the hero tiles only
Scannable in 10 seconds — the goal is "anything weird?" not deep analysis

Thumbs up/down gets reviewed monthly: metrics with net-negative votes get replaced. This creates a natural pruning mechanism.

Collectibility Assessment

Source	Metrics	Status
SQL (production RO)	13	✅ `collectSql()` exists
Stripe (paginated)	5	✅ `collectStripe()` with `sgetAll()`
PostHog (HogQL)	1	⚠️ Search-to-inquiry needs event instrumentation
GSC	1	✅ `collectGsc()` exists

19 of 20 collectible today. #12 (search-to-inquiry) needs PostHog event instrumentation for search events — tracked as a dependency on the PostHog Messaging Instrumentation feature.

Reviewed by strategic subagent + Codex adversarial. Both reviews converged on: add fraud metrics, remove redundancies, add search-to-inquiry conversion. 9 changes applied from draft.

gsingal/kpi-100-metrics.md

Select an option

No results found