Metrics & Money Framework

Pre-work for Session 8 | Prepared: 2026-02-18

Summary (3 sentences for Danny)

We are pre-revenue with zero production data, so the metrics system needs to be dead simple: one North Star metric (properties under active management), a weekly dashboard Danny checks in 5 minutes, and a monthly review that the engineering team runs from Stripe + Supabase data — no analytics platform until we hit 50 paying customers. The unit economics look strong on paper (GBP 0.43-0.99 cost per property, GBP 3-5 revenue per unit, 67-86% gross margin) but every assumption needs validation from the first 10 customers, so the framework includes a concrete validation plan with specific data points to collect and when. Stripe handles billing with per-unit pricing and 14-day free trials, ChartMogul (free tier) gives us MRR/churn dashboards from day one, and we track LLM costs per customer using Langfuse or a simple Supabase logging table — total tooling cost: GBP 0-80/month.

The Metrics That Matter

For a pre-revenue SaaS, what should we actually track? The answer is not everything — it is the smallest set of numbers that changes decisions. If a metric does not cause you to do something different, stop tracking it.

North Star Metric

Properties under active management (PAM) — defined as properties where at least one tenant has sent a message through Envo in the last 30 days.

Why this and not MRR?

MRR is a lagging indicator — it tells you what happened. PAM is a leading indicator — it tells you whether the product is delivering value. A landlord who pays but whose tenants never use the system will churn. A landlord whose tenants are actively messaging will stay. PAM correlates with retention, expansion (landlords add more properties when it works), and referrals (they tell other landlords).

At our stage, growing PAM from 0 to 100 is the only thing that matters. Revenue follows.

Target trajectory:

Month 1-3: 0 to 50 PAM (first 5-8 customers)
Month 4-6: 50 to 200 PAM (product-market fit signal)
Month 7-12: 200 to 500 PAM (scaling signal)

Weekly Dashboard

What does Danny look at every Monday morning? What do Bilal and Deen look at?

Danny’s Monday Dashboard (5 minutes, no login required — delivered via email or Slack)

Metric	Target	Source	Why It Matters
New sign-ups this week	3-5/week by month 3	Supabase Auth	Is the top of funnel working?
Active trials	Track trend	Supabase (trial_started_at)	How many people are evaluating us right now?
Trial-to-paid conversions this week	>15% of expiring trials	Stripe	Are trials turning into revenue?
Hot leads in pipeline	3-5 at any time	Notion lead tracker	Is there enough to close this month?
Properties added this week (new + existing customers)	Increasing week-on-week	Supabase	Are customers expanding?
Customer health flags	Zero red flags	Supabase (no login >5 days, zero issues >14 days)	Who is about to churn?

Engineering Dashboard (5 minutes, from Supabase + provider dashboards)

Metric	Target	Source	Why It Matters
Total AI conversations this week	Track trend	Supabase conversation logs	Is usage growing? Are we approaching cost thresholds?
AI resolution rate (conversations that did not need landlord intervention)	>60%	Supabase (issue status)	Is the AI good enough? This is our core product quality metric
LLM spend this week (GBP)	< 50% of revenue	Anthropic/OpenAI dashboards	Are costs tracking to plan?
System uptime	99.5%+	UptimeRobot	Did anything break?
Support tickets opened	<1 per customer/month	Crisp or email inbox	Is the product confusing?
Error rate (Sentry)	<0.1% of requests	Sentry	Stability check

Monthly Review Metrics

The team sits down for 30 minutes on the first Monday of each month.

Metric	Formula	Benchmark (Early Stage B2B SaaS)	Our Target
MRR	Sum of all active subscription values	N/A pre-revenue; 10-20% MoM growth once live	15% MoM growth after month 3
MRR Growth Rate	(MRR this month - MRR last month) / MRR last month	10-20% MoM for early stage	15% MoM
Logo Churn Rate	Customers lost / customers at start of month	<5% monthly for early B2B SaaS; 3.5% is average	<5% monthly
Revenue Churn Rate	MRR lost from churned + contracted / MRR at start	<3% monthly (best-in-class B2B SaaS)	<5% monthly
Net Revenue Retention	(MRR at start + expansion - contraction - churn) / MRR at start	>100% is the gold standard	>100% by month 6
CAC Payback Period	CAC / (ARPU x Gross Margin %)	<12 months; early stage median is 4.8 months	<6 months
Gross Margin	(Revenue - COGS) / Revenue	70-85% for B2B SaaS	>70%
Trial-to-Paid Conversion	Paid conversions / trial starts in same cohort	10-25% for B2B SaaS free trials	>15%
Properties per Customer	Total properties / total customers	N/A	Increasing trend (expansion signal)
Tenant Adoption Rate	Tenants who messaged / total tenants in system	N/A (our unique metric)	>30% within 30 days of go-live

Revenue Metrics

Metric	Definition	How to Calculate	Tool
MRR	Monthly Recurring Revenue — the predictable revenue normalised to a monthly amount	Sum of all active subscriptions’ monthly value. Annual plans divided by 12	Stripe Dashboard or ChartMogul
ARR	Annual Recurring Revenue	MRR x 12	ChartMogul
ARPU	Average Revenue Per User (per customer)	MRR / number of paying customers	Stripe + spreadsheet
ARPA	Average Revenue Per Account	Same as ARPU for us (1 account = 1 org)	Stripe
Revenue per property	Revenue generated per property under management	MRR / total properties across all paying customers	Supabase + Stripe
MRR Growth Rate	Month-on-month percentage increase in MRR	(Current MRR - Previous MRR) / Previous MRR	ChartMogul
New MRR	MRR from brand new customers this month	Sum of first-month subscriptions	ChartMogul
Expansion MRR	MRR increase from existing customers (added properties, upgraded tier)	Sum of MRR increases from existing accounts	ChartMogul
Contraction MRR	MRR decrease from existing customers (removed properties, downgraded)	Sum of MRR decreases from existing accounts	ChartMogul
Churned MRR	MRR lost from customers who cancelled	Sum of MRR from cancelled subscriptions	ChartMogul
Net New MRR	Total MRR change this month	New MRR + Expansion MRR - Contraction MRR - Churned MRR	ChartMogul

Customer Metrics

Metric	Definition	Target	How to Measure
Total paying customers	Organisations with an active, non-trial subscription	Growing month-on-month	Stripe active subscriptions
Total properties under management	All properties across all paying orgs	This is our scale metric — target 500 by month 12	Supabase `properties` table, filtered to paying orgs
Trial-to-paid conversion rate	Percentage of trial sign-ups that convert to a paid plan within 30 days	>15% (industry benchmark 10-25%)	Stripe: count conversions / count trial starts per cohort
Activation rate	Percentage of sign-ups who complete the aha moment (first real tenant conversation)	>60% within 7 days of sign-up	Supabase: first non-test message timestamp exists within 7 days of account creation
Time to activation	Days from sign-up to first real tenant conversation	<3 days (Solo Landlord), <7 days (Portfolio)	Supabase: timestamp diff between account creation and first tenant message
Logo churn rate	Percentage of customers who cancel per month	<5% monthly	Stripe cancelled subscriptions / active at start of month
Revenue churn rate	Percentage of MRR lost to cancellations and downgrades	<5% monthly	ChartMogul
Net Revenue Retention (NRR)	Whether existing customers generate more or less revenue over time	>100% (means expansion outweighs churn)	ChartMogul
Properties per customer	Average number of properties per paying org	Track trend — increasing means expansion	Supabase
NPS / CSAT	Customer satisfaction	NPS >40, CSAT >4.0/5.0	Manual survey (Typeform or Google Form) at day 30 and quarterly
Customer concentration	Revenue from top customer / total MRR	<25% (risk flag if one customer is too large)	Stripe

Product Metrics

Metric	Definition	Why It Matters
AI conversations handled	Total conversations where the AI responded to a tenant without human intervention	Core product value metric — proves the automation thesis
AI resolution rate	Percentage of conversations resolved without landlord intervention (issue created, information provided, or troubleshooting completed by AI alone)	The single best measure of product quality. If this is low, landlords are still doing the work manually
Tenant adoption rate	Percentage of imported tenants who have sent at least one message	If tenants do not use it, landlords see no value. Track per-property and per-customer
Average response time	Time from tenant message to AI first response	Should be <30 seconds. If degraded, indicates system performance issues
Conversations per property per month	Average number of AI conversations per active property	Drives cost calculations. Our model assumes ~2/property/month. Validate this urgently
Channel distribution	Percentage of conversations by channel (web chat, WhatsApp, voice, email)	Determines cost mix (voice is 5-10x more expensive). Informs channel investment decisions
Issue creation rate	Percentage of conversations that result in a new issue being created	Not all conversations are issues (some are information queries). Track the split
Document AI queries	Number of times the AI retrieves information from uploaded property documents	Measures whether the RAG pipeline is being used. Low numbers mean either documents are not uploaded or the AI is not retrieving well
Onboarding completion rate	Percentage of sign-ups that complete all activation milestones	Per the onboarding matrix: >85% should reach fully-live status
Feature adoption breadth	Number of features used per customer in first 30 days (issue management, vendor assignment, compliance tracking, document upload, reporting)	Customers using 3+ features in the first month are significantly less likely to churn
Vendor assignment rate	Percentage of issues that get a vendor assigned	Measures whether landlords are using the full workflow or just using Envo as a notification system

Cost Metrics

Metric	Definition	Target
Cost per property per month	Total infrastructure costs / total active properties	<GBP 1.00 (from capacity analysis: GBP 0.43-0.99 at scale)
Cost per conversation	(LLM cost + messaging cost) / total conversations	<GBP 0.20 (from estimates: GBP 0.06-0.18 LLM + GBP 0.06-0.18 WhatsApp)
LLM cost per customer per month	Total Anthropic + OpenAI spend / paying customers	Track and ensure it stays <30% of ARPU
Twilio cost per customer per month	Total Twilio spend (WhatsApp + SMS) / paying customers	Track and ensure it stays <20% of ARPU
Voice cost per customer per month	Total VAPI spend / customers using voice	This is the wildcard — voice is 5-10x more expensive per interaction
Total COGS per customer	All variable costs attributable to serving one customer (LLM + Twilio + VAPI + SendGrid + allocated Supabase/Vercel)	<30% of ARPU to maintain 70%+ gross margin
Gross margin	(Revenue - COGS) / Revenue	>70% (SaaS industry norm for B2B vertical software is 70-85%)
Infrastructure cost (fixed)	Supabase + Vercel + SendGrid base fees	<GBP 150/month until 500+ properties
Infrastructure cost per property	Fixed infra costs / total properties	Decreasing as we scale (economies of scale on fixed costs)
Support cost per customer	Danny + Bilal/Deen time on support x hourly rate / customers	<GBP 10/customer/month (otherwise hire a support person)
Burn rate	Total monthly spend (infra + tools + any contractor costs, excluding salaries)	Track monthly; target <GBP 500/month pre-revenue

Unit Economics

The Model

Variable	Estimate	Source	Confidence
ARPU (per customer)	GBP 150-250/month (assumes 30-50 units at GBP 3-5/unit)	Pricing analysis — GBP 3/unit Basic, GBP 5/unit Professional	Medium — untested with real customers
Revenue per property	GBP 3-5/month	Pricing analysis	Medium
COGS per property	GBP 0.43-0.99/month (all-in, Envo pays everything)	Capacity analysis	Medium — based on 2 conversations/property/month
COGS per property (BYOAK Premium)	GBP 0.25-0.55/month	Capacity analysis	Medium
Gross margin per property	67-86% at GBP 3/unit; 80-90% at GBP 5/unit	Calculated: (price - cost) / price	Medium
Gross margin per customer	GBP 100-215/month	ARPU x gross margin %	Medium
CAC	GBP 200-500 (estimated: Danny’s time + demos + content)	Estimate: 5-10 hours of Danny’s time per closed customer	Low — no data yet
LTV	GBP 1,800-5,100 (at 12-18 month average lifetime)	ARPU x gross margin x average customer lifetime in months	Low — no retention data
LTV/CAC ratio	3.6-25.5x	LTV / CAC	Low — both inputs are estimates
CAC payback period	1-3 months	CAC / (ARPU x gross margin %)	Low — optimistic because early adopters convert easily
Monthly burn rate (infra only)	GBP 100-200 pre-customers	Current Supabase + Vercel + domain costs	High
Break-even customer count	1-2 customers (infra only, excluding salaries)	Burn rate / gross margin per customer	High

Key Assumptions in the Model

2 conversations per property per month — This is the biggest cost assumption and the least validated. HMOs with 6+ tenants could be 3-5x this.
30-50 units per customer — Based on ICP (15-50 property HMO operators, averaging 2-3 units per property). Could be lower if we attract smaller landlords.
12-18 month customer lifetime — Industry average for SMB B2B SaaS is 12-24 months. We have zero data.
GBP 200-500 CAC — Assumes founder-led sales (Danny’s time). No paid acquisition. This is low by SaaS standards but realistic for a warm-network-first GTM.
70%+ gross margin — Requires costs per property to stay under GBP 1.00. If voice adoption is high or conversation volume per HMO exceeds estimates, margin compresses.

Validation Plan

What data do we need from the first 10 customers to validate or invalidate the model?

Assumption	What We Assumed	How to Measure	When We Will Know	Action if Wrong
Conversations per property per month	~2	Count distinct conversations per property per month from Supabase logs	After 30 days of live usage per customer	If >4: raise per-unit price or introduce conversation caps. If <1: investigate tenant adoption
Channel mix (WhatsApp vs voice vs web chat)	70% WhatsApp, 10% voice, 20% web chat	Count conversations by channel from Supabase	After 60 days of usage	If voice >25%: cost model breaks, must either cap voice minutes or increase price
Tenant adoption rate	>30% of tenants use system within 30 days	Count unique tenant senders / total tenants per org	30 days after each customer go-live	If <15%: onboarding problem — tenants are not being told about Envo. Fix welcome message and landlord communication
Average units per customer	30-50	Average property count across paying customers from Supabase	After 10 paying customers	If <15: ARPU is lower than modelled, need more customers to reach same MRR. Consider lowering minimum commitment
CAC (Danny’s time to close)	5-10 hours	Danny tracks hours per closed customer in Notion	After 10 closes	If >15 hours: sales process is too heavy. Invest in self-serve. If <5 hours: scale outreach
Trial-to-paid conversion	>15%	Stripe: conversions / trial starts	After 20 trial sign-ups	If <10%: aha moment is not landing. Review onboarding flow, sandbox demo, and Danny’s follow-up timing
Churn (monthly)	<5%	Stripe cancellations / active at month start	After 6 months (need enough data)	If >8%: product-market fit problem. Conduct exit interviews. Check if AI quality is the issue
LLM cost per conversation	GBP 0.06-0.18	Anthropic/OpenAI billing / conversation count	After 30 days	If >GBP 0.25: optimise prompts, reduce token usage, consider cheaper model for simple queries
Expansion revenue	Existing customers add properties	Track unit count changes per org in Supabase over time	After 90 days	If zero expansion: product is not sticky enough or landlords have fixed portfolios. Adjust growth model
Gross margin	>70%	(Revenue - all variable costs) / Revenue	After 3 months of billing	If <60%: pricing is too low or costs are higher than modelled. Review per-customer cost breakdown

Billing Setup

Stripe Configuration

Based on ADR-004 (Billing Integration Strategy), Stripe is the assumed provider. Here is how to configure it for per-unit billing with tiers.

Tier	Stripe Product	Price Model	Trial	Notes
Starter	Product: “Envo Starter”	Per-unit recurring price: GBP 3/unit/month. Use `recurring[usage_type=licensed]` with quantity set to number of units. Minimum 10 units enforced via checkout logic	14-day free trial, no card required at sign-up. Card collected at day 12 via email prompt	Quantity updated when landlord adds/removes properties via dashboard. Use Stripe `subscription_items.update` to change quantity
Professional	Product: “Envo Professional”	Per-unit recurring price: GBP 5/unit/month. Same licensed model with quantity. Minimum 20 units	14-day free trial, no card required. Premium features active during trial	Higher conversation cap (8/unit/month vs 4). Voice channel included
Enterprise	Product: “Envo Enterprise”	Custom pricing: GBP 2-3/unit/month negotiated. Manual Stripe subscription creation with custom amount	Pilot programme (30 days, bespoke)	BYOAK means lower Envo costs. Manual invoicing may be needed for large accounts. Minimum GBP 500/month
Add-ons	Separate Stripe Products	One-time or recurring: Extra conversation packs (GBP 15/50 conversations), extra voice minutes (GBP 10/30 minutes), extra team seats (GBP 10/seat/month)	N/A	Attached as additional subscription items

Stripe setup steps (practical):

Create a Stripe account (if not already done). Activate in the UK with GBP as default currency
Create Products for each tier (Starter, Professional, Enterprise)
Create Prices on each Product: per-unit, monthly recurring, GBP denominated
Set up a Customer Portal (Stripe hosted) so landlords can update payment methods, view invoices, and cancel
Implement Stripe Checkout for sign-up conversion (trial end → card collection)
Set up webhooks to sync subscription status back to Supabase (organisations.plan and external_billing_id)
Enable Smart Retries for failed payments (Stripe’s built-in dunning — recovers up to 25% of failed charges)
Configure Stripe Tax for automatic UK VAT calculation

Integration pattern (from ADR-004):

Stripe Webhook --> Supabase Edge Function --> billing_events table
                                          --> Update subscriptions table
                                          --> Update organisations.plan

Billing Flows

Event	What Happens	Stripe Mechanism
Sign-up (trial start)	Customer creates account. Stripe Customer created with `external_billing_id` stored in Supabase. Subscription created with `trial_period_days: 14`. No card required	`stripe.customers.create` + `stripe.subscriptions.create` with trial
Trial day 12	Automated email: “Your trial ends in 2 days. Add a payment method to continue.” Link to Stripe Customer Portal	Stripe Billing automatic trial-ending emails (configurable in Dashboard)
Trial end (card on file)	Subscription transitions to `active`. First invoice generated and charged. `organisations.plan` updated via webhook	Automatic Stripe trial-to-active transition. Webhook: `customer.subscription.updated`
Trial end (no card)	Subscription moves to `past_due` or `canceled` (configurable). Dashboard becomes read-only. Data preserved 30 days	Stripe `subscription_schedule` or manual cancellation logic. Webhook: `customer.subscription.deleted`
Upgrade (Starter to Professional)	Subscription updated to Professional price. Prorated amount charged for remainder of billing period	`stripe.subscriptions.update` with `proration_behavior: 'create_prorations'`
Downgrade (Professional to Starter)	Subscription updated at end of current billing period. Features restricted immediately or at period end (decision needed)	`stripe.subscriptions.update` with `proration_behavior: 'none'`, effective at `current_period_end`
Property count change (add units)	Quantity on subscription item increased. Prorated charge for added units	`stripe.subscriptionItems.update` with new `quantity`. Proration automatic
Property count change (remove units)	Quantity decreased. Credit applied to next invoice	`stripe.subscriptionItems.update` with reduced `quantity`
Failed payment	Stripe Smart Retries kick in (up to 4 retry attempts over ~3 weeks). Customer emailed automatically. After final failure: subscription moves to `past_due`, then `canceled` after grace period	Stripe Smart Retries + configurable dunning emails. Webhook: `invoice.payment_failed`
Successful retry	Subscription restored to `active`. No customer action needed	Automatic. Webhook: `invoice.paid`
Voluntary cancellation	Customer cancels via Customer Portal or contacts Danny. Subscription set to cancel at end of billing period. Access continues until period end	`stripe.subscriptions.update` with `cancel_at_period_end: true`. Webhook: `customer.subscription.updated`
Reactivation	Previously cancelled customer resubscribes. New subscription created (or existing one uncancelled if within grace period)	`stripe.subscriptions.update` to remove `cancel_at_period_end`, or new subscription creation
Annual billing	20% discount applied. Single annual charge. Subscription with `interval: 'year'`	Separate annual Price on same Product. Or use Stripe Coupons for the discount

Invoice and Tax

UK VAT requirements for SaaS:

VAT registration threshold: GBP 90,000 taxable turnover. Below this, VAT registration is optional but not required
VAT rate on SaaS: 20% standard rate for UK customers
When to register: When you anticipate reaching GBP 90K in the next 30 days, or when you have exceeded it in the previous 12 months. At our projected pricing, this is roughly 150-250 paying customers
Recommendation: Do NOT register for VAT at launch. You are pre-revenue. Register when you approach the threshold. This keeps pricing simple (no VAT to add or absorb)
When you do register: Use Stripe Tax for automatic VAT calculation and collection. Stripe handles the rate determination and adds VAT to invoices automatically

What needs to be on invoices (once VAT-registered):

Your business name and address
Your VAT registration number
Invoice date and unique invoice number
Customer’s name and address
Description of the service (“Envo property management platform — [X] units, [Tier] plan”)
Quantity and unit price
Total amount excluding VAT
VAT rate and VAT amount
Total amount including VAT

Pre-VAT registration: Stripe invoices will include all the above except VAT fields. Still include your business details and a clear description of the service.

B2B reverse charge: If you sell to VAT-registered businesses (letting agencies), the reverse charge mechanism may apply. Stripe Tax handles this automatically when the customer provides their VAT number.

Record keeping: HMRC requires you to keep records for at least 6 years. Stripe retains all invoice and payment data indefinitely. Additionally, the UK will require mandatory B2B e-invoicing from April 2029 — Stripe is expected to support this.

Financial Tooling

Need	Tool	Cost	Why
Payment processing + subscriptions	Stripe Billing	1.5% + 20p per transaction (UK cards) + GBP 0 monthly	Industry standard. Handles subscriptions, trials, dunning, invoices, tax. Per-unit pricing supported natively. No monthly fee — just transaction costs
SaaS metrics dashboard (MRR, churn, LTV)	ChartMogul (free tier)	GBP 0 (free up to GBP 96K ARR / ~$120K)	One-click Stripe integration. Automatic MRR, ARR, churn, LTV, cohort analysis. Free tier includes unlimited users and core reports. Upgrade to paid only when ARR exceeds threshold
LLM cost tracking	Langfuse (open source, self-hosted) OR simple Supabase logging table	GBP 0 (self-hosted) or GBP 0-25/month (Langfuse cloud)	Tag each LLM call with `org_id` and `conversation_id`. Aggregate daily cost per customer. Essential for validating unit economics. Alternative: log token counts and costs to a `llm_usage` table in Supabase and query with SQL
Uptime monitoring	UptimeRobot (free tier)	GBP 0	50 monitors free, 5-minute checks. Status page included. Alerts via email/Slack/webhook. Upgrade to paid (GBP 6/month) for 1-minute checks if needed
Error tracking	Sentry (free tier)	GBP 0	5K errors/month free. Catches bugs before customers report them. Critical for product quality metrics
API cost dashboards	Anthropic Console + OpenAI Usage + Twilio Console	GBP 0	Each provider has its own usage dashboard. Check weekly. No additional tooling needed until costs are complex enough to warrant aggregation
Spreadsheet for manual tracking	Google Sheets	GBP 0	For anything that does not yet justify a tool: CAC tracking, Danny’s time per customer, manual cohort analysis, board reporting. One shared sheet, updated monthly
Lead tracking	Notion (free tier)	GBP 0	As defined in the lead tiers playbook. Not a CRM — a structured database. Upgrade to paid when >200 leads
Customer feedback / NPS	Typeform or Google Forms	GBP 0	Send at day 30 and quarterly. 3-5 questions max. Do not overthink this — a simple “How likely are you to recommend Envo?” with a free-text “Why?” is enough
Accounting	Xero (Starter) or FreeAgent	GBP 15-25/month	Not needed until revenue starts. When it does: connect Stripe to Xero for automatic invoice reconciliation. Essential for VAT returns when you register
Forecasting	Google Sheets	GBP 0	A simple model: current MRR x growth rate = forecast. No tool needed. Baremetrics has forecasting but costs GBP 80+/month — premature

Total tooling cost at launch: GBP 0-25/month (Langfuse cloud is the only possible non-zero cost, and even that can be self-hosted for free).

When to upgrade:

Trigger	Action	Estimated Cost
ARR exceeds GBP 96K (~80 customers at GBP 100/month)	ChartMogul paid tier	GBP 80-100/month
>50 customers and cost attribution becomes complex	Dedicated cost tracking (Portkey or LiteLLM proxy)	GBP 0-50/month
Revenue starts and VAT registration approaches	Xero or FreeAgent for accounting	GBP 15-25/month
>100 customers	Baremetrics (adds dunning, forecasting, cancellation insights)	GBP 80-200/month

Reporting Cadence

Report	Frequency	Audience	Contains
Danny’s Dashboard	Weekly (Monday morning, automated email)	Danny	New sign-ups, active trials, conversions, pipeline count, customer health flags, properties added
Engineering Dashboard	Weekly (Monday morning, Supabase query or script)	Bilal, Deen	AI conversations, resolution rate, LLM spend, uptime, errors, support tickets
Monthly Business Review	Monthly (first Monday)	Danny + Bilal + Deen	Full metrics: MRR, growth, churn, CAC, unit economics, cost per property, product metrics, NPS. 30-minute meeting
Customer Health Check	Monthly (alongside business review)	Danny	Per-customer: properties, usage, last login, issues handled, support tickets, expansion/contraction signals
Cost Audit	Monthly (Bilal or Deen runs, 15 minutes)	Bilal, Deen	Provider-by-provider spend: Anthropic, OpenAI, Twilio, VAPI, Supabase, Vercel, SendGrid. Per-customer attribution where possible
Quarterly Board/Investor Update (when relevant)	Quarterly	External stakeholders	ARR, growth rate, customer count, churn, LTV/CAC, gross margin, product highlights, next quarter goals
Validation Report	After 10 customers (one-time, then quarterly)	Whole team	Comparison of assumed vs actual: conversations/property, channel mix, tenant adoption, CAC, churn. The “were we right?” report

How to Build the Dashboards (Practically)

For a 3-person team, do not build dashboards. Query data and send it somewhere people already look.

Danny’s weekly email: Write a Supabase Edge Function or a simple cron job (n8n workflow) that runs every Monday at 8am, queries key metrics from Supabase + Stripe API, and sends a formatted email to Danny. Total build time: 2-4 hours.

Engineering weekly check: A SQL script saved in the repo that Bilal or Deen runs against Supabase. Returns a table of key metrics. Takes 5 minutes to run and read. Build time: 1-2 hours.

Monthly review: Google Sheets template with manual data entry from ChartMogul, Stripe, Supabase, and provider dashboards. Pre-built formulas calculate derived metrics (gross margin, CAC payback, NRR). Build time: 2 hours initially, 30 minutes to update monthly.

Do not build a custom analytics dashboard. It is a trap. Use existing tools (ChartMogul, Stripe Dashboard, provider consoles) and a spreadsheet. Build a custom dashboard only when you have 100+ customers and the manual process takes >1 hour per week.

Open Questions for Session

Per-property or per-unit billing in Stripe? The pricing analysis recommends per-unit (per room in HMOs). This affects how we set up Stripe Products and Prices. If per-property, the quantity on the subscription = number of properties. If per-unit, quantity = total rooms/units. This must be decided before Stripe setup. Danny — what did prospects respond to?
When do we start charging? The first 3-5 customers are arguably pilots, not paying customers. Do we offer “founding customer” pricing (e.g., 50% off for 12 months in exchange for feedback and testimonials)? Or do we charge full price from day one to validate willingness to pay? There is a strong argument for charging from day one — free pilots do not validate demand.
Conversation caps — do we enforce them or just track them? The pricing matrix includes 4 conversations/unit/month on Starter, 8 on Professional. Do we actually cut off AI responses when the cap is hit, or do we track usage and use overages as an upgrade trigger? Hard caps create bad experiences. Soft caps with alerts are gentler. Decision needed.
Who owns the metrics? Proposed: Danny owns customer and revenue metrics (he is closest to the money). Bilal and Deen own product and cost metrics (they are closest to the system). All review together monthly. Does this split work?
LLM cost tracking — build or buy? Option A: Log token usage to a Supabase table on every LLM call (add org_id, tokens_in, tokens_out, cost, model columns). Zero cost, fully custom, but we have to build the aggregation queries. Option B: Use Langfuse (purpose-built LLM observability). More features but another tool to manage. Recommendation: Option A for launch (15 minutes to build), migrate to Langfuse when we need deeper analysis.
How do we handle the HMO cost problem? A 6-bed HMO generates roughly 6x the conversations of a single-let but might pay the same under per-property pricing. The validation plan will tell us the actual numbers, but we need a contingency plan now. Options: per-unit pricing (recommended), conversation caps, or an HMO surcharge. What feels right to Danny from a sales perspective?
Do we need Xero from day one? If we are pre-revenue and below the VAT threshold, a spreadsheet tracking income and expenses might suffice for the first 6 months. Xero adds ongoing cost and complexity. Counter-argument: connecting Stripe to Xero from day one means clean books from the start with no retrospective data entry. Worth the GBP 15-25/month?
What is our burn rate tolerance? The infrastructure costs are low (GBP 100-200/month), but total burn including tools, domain, and any contractor work needs a number. How long is our runway, and does that affect how aggressively we price or how many free trials we offer?
Annual billing from launch? 20% discount for annual payment improves cash flow and reduces churn (customers are committed for 12 months). But we have no retention data — what if the product is not good enough and we have to issue refunds? Recommendation: monthly-only at launch, introduce annual option after 6 months of retention data.
When is the first “were we right?” review? Proposed: after 10 paying customers have been live for 30+ days. This gives us enough data to validate the unit economics model. Target date should be agreed in the session.

Appendix: Metric Definitions Quick Reference

For anyone unsure about what a metric means:

Term	Plain English
MRR	How much money comes in every month from subscriptions
ARR	MRR times 12 — the annualised version
ARPU	Average revenue per customer per month
Churn	The percentage of customers (or revenue) you lose each month
NRR	Whether your existing customers are paying you more or less over time. Above 100% = growing from existing customers
CAC	How much it costs to acquire one new paying customer
LTV	How much total revenue one customer generates over their entire lifetime
LTV/CAC	How many times over a customer pays back their acquisition cost. 3:1 is the benchmark
Payback Period	How many months until a customer has paid back their acquisition cost
Gross Margin	Revenue minus the direct costs of serving customers, as a percentage. The money left over to pay salaries and grow
COGS	Cost of Goods Sold — for SaaS, this is infrastructure costs (LLM, hosting, messaging)
PAM	Properties under Active Management — our North Star metric

Research Sources

Prepared for Session 8: Metrics & Money. All benchmarks are from 2025-2026 industry data. All cost estimates reference the Capacity Analysis and Pricing Analysis prepared for earlier sessions. No production usage data is available yet — the validation plan is designed to fix that with the first 10 customers.

EHQ Brain

Explorer

metrics-framework