--- id: "13.2" title: "Unit economics" dept: finance status: draft owner: Param branded: false last_verified: 2026-10-26 sources: - "brew origin/main a559de6: drafts-bg.js:142, rewrite-pot.js:19-24, public/demo-picker.html:201, docs/handover/product/08-demo-guide.md:8, live-watch.js:22-23, social-feed.js:274, ai.js:8, gemini-client.js:16" - "brew working tree docs/pricing.md:20-35, untracked" - "handbook content/ai-and-models/cost-per-brand.md Today and Numbers" - "memory: brew-model-compare-2026-10-20, brew-lean-pipeline, brew-jam-v2, brew-sloosh-finals, brew-gemini-provider, treg-tool-catalog, keep-all-moves" --- # Unit economics ## Answer What one read, one brand a month, one make and one final cost Brew, against what the pricing proposal would charge; single pieces are measured, a brand's month is not, so margin per brand is a range until a real month is counted. ## Today ### Cost per unit | Unit | Cost | Measured or estimated | Source | |---|---|---|---| | A live read | about $6 | Estimate in the demo guide | [src: brew/docs/handover/product/08-demo-guide.md:8] | | A full demo build, read plus every Play | about $80 | Staff note | [src: brew/public/demo-picker.html:201] | | A Play's own call | median $0.18 on Pro, $0.83 on Opus, about $0.10 on Flash | Measured on Arata and Cozy | [src: memory brew-model-compare-2026-10-20] [src: brew/rewrite-pot.js:18-19] | | A Play run with search, check and fix | about 1.5 times its own call | Code constant | [src: brew/rewrite-pot.js:23-24] | | A Play before the lean pipeline | $1.51 | Measured, 15 Oct | [src: memory brew-lean-pipeline] | | A draft picture | $0.04 | Code default | [src: brew/drafts-bg.js:142] | | A draft video | $0.90 | Code default | [src: brew/drafts-bg.js:142] | | A draft voice clip | $0.01 | Code default | [src: brew/drafts-bg.js:142] | | A Studio make on Arata, 23 Oct | product shot $0.10, preset $0.10, Kling 5 s $0.40, Seedance UGC 10 s $3.02 | Measured | [src: memory brew-jam-v2] | | A Sloosh final | image 2.01 credits, reel 45.071 credits | Measured | [src: memory brew-sloosh-finals] | | An AI answers check | about $0.50 a run | Code note | [src: brew/live-watch.js:22-23] | | Listening and ideas | listening about $0.30 a run, ideas about $1 a run, $3 a brand a day cap | Code estimates | [src: brew/social-feed.js:274] | | Outside data a month | about $10 of treg a brand | Estimate | [src: memory treg-tool-catalog] | Model list prices are in the code: Opus $4 and $20 per million tokens in and out, Sonnet $2 and $10, Flash $0.5 and $3, 3.1 Pro $2 and $12 [src: brew/ai.js:8] [src: brew/gemini-client.js:16]. Vertex is paid by Google credits today [src: memory brew-gemini-provider]. ### Cost per brand a month - Not measured for any brand [src: content/ai-and-models/cost-per-brand.md Open] - Proposed ceiling from the caps: about $180 before drafts [src: content/ai-and-models/cost-per-brand.md Numbers] - The pricing proposal's estimate: about $15 brain cost in v1 [src: brew working tree docs/pricing.md:23] ### Against the proposed prices | Item | Price | Our cost | Gross margin, on paper | Source | |---|---|---|---|---| | Post ready image | $0.75 | about $0.03 to $0.06 | about 92% to 96% | [src: brew working tree docs/pricing.md:28] | | Carousel | $2 | about $0.12 to $0.30 | about 85% to 94% | [src: brew working tree docs/pricing.md:29] | | Ad ready video | $4 | about $0.55 to $1.30 | about 68% to 86% | [src: brew working tree docs/pricing.md:30] | | Full rewrite | $6 | about $1.80 to $3.00 | about 50% to 70% | [src: brew working tree docs/pricing.md:32] | | v1 plan | $79 | about $15 estimated, up to about $180 allowed | from about 81% down to below zero | [src: brew working tree docs/pricing.md:20-23] [src: content/ai-and-models/cost-per-brand.md Numbers] | Margins are our arithmetic on the proposal's figures, before payment fees and tax. ## Tomorrow | Item | Version | Tag | |---|---|---| | Every Play kept; each made cheaper | v1 | Stays [src: memory keep-all-moves] | | One measured cost line per brand per month | v1 | New | | A monthly ceiling per brand tied to its plan | v2 | New | | Wallet ledger with cost and price per piece | v1, after pricing | New | | Posting, results and ad runs add their own cost lines | v2 to v4 | New | ## How we do it 1. Read a brand's outside data spend for a month from its spend file, as shown in [cost per brand](/ai-and-models/cost-per-brand). 2. Add the AI tags for that brand from the meter. 3. Divide by the plan price; anything above a third is a flag for Param. 4. Never cut a Play to save cost; cut the cost of the Play. ## Numbers | Metric | Definition | Where measured | Status | |---|---|---|---| | Cost per read | Dollars from paste to a finished read | AI tags `reader-` plus treg | Proposed; about $6 estimated | | Cost per brand a month | All AI and outside data for one brand | Spend files plus AI tags | Proposed, not measured | | Cost per make | Dollars per Studio take | Studio quotes | Measured, $0.10 to $3.02 | | Cost per final | Sloosh credits per final, converted to dollars | Sloosh quotes | Measured in credits only | | Plan margin | Plan price less cost per brand a month | Both above | Proposed | ## Open 1. **What is a Sloosh credit in dollars?** Proposal: Param adds the Pro plan's price and credit count to vendors and bills, and finals are reported in dollars from then on. 2. **Which number do we plan on until a month is measured?** Proposal: plan on $45 a brand a month, the v4 brain estimate, as a cautious middle; replace it with the measured figure in December.