03 - Cost & Unit Economics

All figures are from EFFEN_Cost_Model.xlsx at placeholder inputs (to be confirmed with the team). MYR. The model is fully formula-driven; every number below updates when the real inputs land.

The headline

Net profit impact  =  (labor displaced + revenue lift)  -  (Track A + B + C spend)

Metric (Base, 36 months)	Value
Payback	month 5
Cumulative net impact	~RM 3.69M
Cumulative labor avoided	~RM 4.10M
Cumulative A+B+C spend	~RM 0.41M
Leverage at month 36 (labor / cost)	~17x

The widening gap - the whole thesis in one table

Cumulative cost barely moves while cumulative labor-avoided compounds. That divergence is the profit.

Month	Cumulative A+B+C cost	Cumulative labor avoided	Cumulative net
0	RM 8,930	RM 0	(RM 8,930)
6	RM 65,017	RM 82,503	RM 17,485
12	RM 123,953	RM 350,135	RM 226,182
24	RM 260,910	RM 1,768,029	RM 1,507,119
36	RM 414,961	RM 4,100,365	RM 3,685,405

Cost grows ~2x over three years; labor-avoided grows from zero to millions. Same output, flat headcount.

Where the monthly spend goes (month 36)

pie showData
    title Monthly A+B+C at month 36 (RM/mo)
    "Track C - AI usage (reporting, creative, DM, video)" : 7071
    "Track B - data infra (warehouse, VPS, MCP)" : 3459
    "Track A - org tooling (Notion, Workspace seats)" : 2865

Total is about RM 13,400/month at month 36 - against RM 220,000/month of labor it avoids at that point.

Unit economics - AI vs the human it replaces

This is the “if we run 1,000 ads at 100% AI, what does it cost?” answer.

Static ad creatives

	AI	Human designer
Cost per ad	RM 1.36	RM 50.00
1,000 ads / month	RM 1,363	RM 50,000 (10 designers)
Monthly saving on 1,000 ads		RM 48,637
AI as % of human	2.7%	100%

Per ad: 4 image variants x (USD 0.04 gen + 0.02 render) + USD 0.05 copy = USD 0.29, x FX 4.70 = RM 1.36.

Customer DMs

	AI	CS agent
Cost per DM	RM 0.33	RM 4.00
10,000 DMs / month	RM 3,290	RM 40,000 (10 agents)
AI as % of human	8.2%	100%

Why A+B+C stays sub-linear (the honest version)

flowchart LR
    OUT["Output grows<br/>(brands, ads, DMs, markets)"] --> LAB["Labor a normal org needs<br/>grows LINEARLY"]
    OUT --> SPEND["A+B+C spend<br/>grows SUB-LINEARLY"]
    LAB --> GAP{{"Widening gap = profit"}}
    SPEND --> GAP
    classDef g fill:#C6EFCE,stroke:#2E7D32,color:#111;
    class GAP g;

• Track A (per-seat) and Track B (flat/step infra on a shared VPS) barely move with output.
• Track C usage lines (video generation, then DMs, then ads, plus API and scraping) do scale with output - but at a per-unit cost that is 2-8% of the human they replace. So even the linear part has a tiny slope.
• The one line to watch: video/image generation and DM volume are the biggest Track C drivers. If creative volume per brand explodes or you lean on premium video at scale, that is where spend could grow. The model’s Unit Economics tab is where you pressure-test it.

The scale-without-cost proof

flowchart TD
    NB["Add one brand / market / ad account"] --> W["Rides EXISTING workflows<br/>(reporting, ideation, creative, social, DM)"]
    W --> MC["Marginal cost: a few hundred RM/mo"]
    W --> ML["Labor it would otherwise need: tens of thousands RM/mo"]
    MC --> NET["Net: each new unit widens the gap"]
    ML --> NET
    classDef g fill:#C6EFCE,stroke:#2E7D32,color:#111;
    class NET g;

Scaling the business becomes adding brands to workflows, not adding headcount. That is lean scaling, and it is measurable - the cost model is the scoreboard.