Total monthly billing across all users — tier subscription price × user count, plus any overage credit purchases when users exceed their monthly credit allowance.
?
Monthly COGS
$24,016 CAD
Vendor + Stripe + infra
Cost of Goods Sold — what we pay vendors to deliver the product: Anthropic (Claude API), Google Gemini, FAL, Resend, Cloudflare, plus Stripe payment-processing fees. Excludes overhead (founder time, marketing, support).
?
Gross Profit
$74,984 CAD
After COGS · pre-overhead
Revenue minus COGS — the money left after paying the variable costs to deliver the product. Doesn't include overhead like founder salaries, marketing spend, or office costs. The "raw" profitability of each customer.
?
Gross Margin
76%
AI-Native target: 70–80%
Gross Profit as % of Revenue. Traditional SaaS targets 80–90%. AI-native SaaS (like Clarify) targets 70–80% because inference costs scale with usage. Above 80% is excellent, 60–70% is concerning, below 60% needs structural fix.
Anthropic offers 90% off cached input tokens. Not wired anywhere yet. Every repeated system prompt across Listings, Chat, Email, Social, Website etc. compounds the saving — payoff scales with volume. Caching the Listing prompt alone is the easiest first pass.
Lever · Available
Model Routing
Route cheap deliverables (social captions, subject lines, chat) to Haiku 4.5 ($1/$5 per M); reserve Opus 4.7 ($5/$25 per M) for CMAs and Carousels. Agent never sees the difference — same UX, ~30% cheaper aggregate COGS at typical mix.
Risk · Watch
Vendor Concentration
Anthropic = ~52% of variable COGS. Any pricing change directly compresses margin. Opus 4.7 just dropped 3× from Opus 4.5, but a future hike has full pass-through risk. Mitigation: prompt caching + model routing absorb a 20-30% Anthropic hike before it touches margin.
Risk · Future
Customer Support at Scale
Today: founders (Braden + Kelley) handle every ticket — free labour, doesn't hit COGS. Once volume requires a dedicated support hire (probably ~500+ paid agents), customer support becomes 7-10% of revenue, dropping Pro margin from ~76% to ~68%. Plan: hire from revenue, not from raise.