What is Unit Economics?

Unit economics is profit and payback measured per unit, not for the company as a whole. The unit is whatever the model turns on: a paying customer, a first-time deposit, an activated subscriber, a delivered order, an approved lead. Until the unit is named, CAC and LTV formulas cannot be compared across teams: they are counting different people.

The pieces already live in this glossary and easily drift apart. CAC answers “what it costs to acquire one unit.” LTV answers “what one unit returns over its life.” Profit margin answers “what share of revenue remains after cost.” Unit economics binds the three numbers into one rule: a unit must pay back in an acceptable time and leave slack for holds, fraud, and approval drop. This is not a monthly P&L and not yesterday’s campaign ROI: P&L mixes old and new cohorts; day-0 in the tracker misses the rebill tail.

What counts as a unit

Define the unit before the spreadsheet. “Customer” in the CAC formula is a business definition, not a click or an install — the CAC article already says that. The point here is that the same unit sits in the CAC denominator and the LTV numerator. Otherwise LTV/CAC is meaningless: installs below, deposits above.

  • New payer / FTD — subscriptions, gambling, ecommerce. Mistake: treating every lead as a “customer.”
  • Approved conversion — CPA, nutra, lead gen. Mistake: confusing a submitted lead with approved.
  • Activated subscriber — trial + rebill. Mistake: putting unpaid trials in the LTV denominator.
  • Order / line item — marketplaces, goods. Mistake: mixing first order and repeats into CAC.

In arbitrage a buyer often treats the unit as “offer conversion.” For the advertiser the unit is a paying user after antifraud. The mismatch is a normal report conflict, not “broken stats.”

Formulas in the frame

Variable cost per unit (shipping, partner payout, acquiring, COGS) is subtracted from unit revenue. What remains is contribution toward ads and fixed cost.

  • Contribution per unit = unit revenue − variable cost per unit.
  • CAC = (ad spend + marketing overhead) ÷ number of new units.
  • LTV = sum of margin from the unit over its life (or Σ period ARPU × retention × gross margin).
  • LTV / CAC — how many times lifetime covers acquisition cost.
  • Payback = CAC ÷ average contribution per unit per period (week / month).
  • Unit margin = contribution ÷ unit revenue. Not ROI: ROI = profit ÷ cost.

Thresholds depend on the model, not on a magic “3×.” SaaS slides often cite LTV/CAC > 3 and payback under 12 months — someone else’s benchmark. In arbitrage, LTV > CAC with slack for hold and chargeback is enough; short payback allows more aggressive scale. In COD nutra the life is short and approval cuts revenue — a unit without approve rate is a lie.

Example. Spend $10,000, get 200 approved first purchases. CAC = $50. Average order $40, repeats over 60 days add $35 revenue per customer, variable costs 20%. LTV ≈ ($40 + $35) × 0.8 = $60. LTV/CAC = 1.2; payback with an even tail is tens of days, not “green on day-0.” A tracker day-0 ROI near zero can sit next to positive unit economics — and the reverse: yesterday in the black, negative LTV after rebill cancels.

How the frame ties CAC, LTV, and margin

CAC is wider than ad-account CPA: overhead includes creatives, team, tools. Effective CPA in the tracker ≈ a lower bound on CAC when other costs are zero. Affiliate LTV is often invisible: the network pays day-0, rebills arrive later — forecast from start-date cohorts, not from an all-time average. Margin says whether the model survives volume: 10% on a large cap can beat 50% on 20 conversions. A unit that ignores COGS and network fees inflates LTV.

The funnel shows where units are lost (CR, approve). Unit economics says whether survivors are worth their cost. Cutting only CPC while chargebacks eat contribution is optimizing the wrong variable. Attribution hits both sides of the ratio: if the ad account and the network credit conversions differently, CAC and LTV are computed on different sets. Reconciliation runs on click ID and postback, not on average order value from web analytics tagged with UTMs.

Mistakes and limits

Wrong unit. Leads in the denominator, payments in the numerator. Or CAC on all visits and LTV on payers only — the ratio inflates.

Mixed cohorts. “Average LTV for the year” blends cheap old traffic and expensive traffic from yesterday. Scale decisions use recent cohorts, or you scale a dead funnel.

CPA = unit economics. Cabinet CPA = spend ÷ offer conversions without overhead, hold, or ops. For a buyer it is a useful day-0 anchor; for the product it is not.

Ignoring time-to-cash. Hold shifts cash payback versus “accounting” LTV. Chargebacks and rejects after payout cut contribution that was already booked.

Unit economics does not replace a campaign report and does not require a full P&L on every sub ID. It sets the stop/scale rule: which unit, which contribution, which payback. Without that, CAC, LTV, and margin stay three disconnected slides.

See also: CAC, LTV, cohort analysis, funnel, ROI.