What makes a subscription business actually profitable?
A hands-on walkthrough of the core ideas behind any subscription business — LTV, CAC, payback period, contribution margin — using synthetic data modeled on real paid-acquisition patterns across three channels: Search, Display, and Performance Max.
US onlyPPCNo-trialForecast model
How a subscriber's value is actually calculated
Every dollar a channel brings in passes through the same bridge — sales, then refunds and servicing costs come out, then what it cost to acquire the customer. What's left is the channel's real contribution to the business.
Revenue bridge — latest cohort
Hover any bar for its exact figure and, on the collapsed steps, what it's made of.
CAC — Customer Acquisition Cost
What it costs, on average, to turn ad spend into one paying subscriber.
Ad spend ÷ new subscribers
LTV — Lifetime Value
The total sales a subscriber is expected to generate over their first 3 years, after refunds and servicing costs.
3yr sales − refunds − service costs
LTV : CAC ratio
For every $1 spent acquiring a customer, how many dollars of value comes back over 3 years. Healthy subscription businesses generally target 3× or more.
LTV ÷ CAC
Payback period
How many months of net revenue — sales after refunds and servicing costs, but before subtracting acquisition spend itself — it takes to earn back what was spent acquiring the customer. Shorter is better — it frees up cash to reinvest sooner.
CAC ÷ monthly net revenue per customer
Contribution margin
What's actually left from a cohort after refunds, servicing costs, expert payouts, and acquisition spend — the real economic contribution to the business.
Net revenue − CAC + lead-gen revenue
Channel snapshot
Side-by-side economics for the three channels. Recent cohorts are still young — the upfront payment (join fee and similar) is booked right away, but most of a new cohort's subscription revenue is still projected, not yet collected.
How each channel has moved over time
Monthly acquisition cohorts, Aug 2021 through Aug 2026. Toggle a channel off to isolate the others.
CAC per customer
Acquisition spend ÷ new subscribers, by cohort month
LTV per customer
3-year forecast value per subscriber, by cohort month
LTV : CAC ratioiLTV:CAC compares LTV (revenue after refunds and servicing costs) to spend. Net ROAS (shown on the snapshot cards) compares net revenue plus lead-gen revenue — LTV with expert payouts also subtracted — to spend instead, a more fully-loaded view. They answer related but different questions and won't match.
1.0× = breakeven over 3 years
Estimated payback periodiThe data has no month-by-month revenue curve, only 3-year cohort totals — so this spreads each cohort's 3-year net revenue (sales after refunds and servicing costs, before subtracting acquisition spend) evenly across 36 months as an estimate. It is a modeled approximation, not an observed measurement.
Months to recover CAC, straight-line estimate · ▲ = actual payback is much longer, capped here at 60mo so the chart stays readable (hover for the real number)
Contribution margin per customer
What's left after refunds, servicing, experts and acquisition spend
Channel details
Loading…
Contribution margin by Performance Max cohort
Cohort totals, colored by whether that month's cohort was profitable
Reading these numbers honestly
＊
Recent cohorts are mostly forecast. The one-time upfront payment (join fee and similar) is booked right away, but a subscriber's monthly payments take up to 3 years to fully play out — so for a brand-new cohort, most of that recurring revenue is still projected, not yet collected. For example, — and these numbers will keep updating as those cohorts mature.
＊
Payback period is a modeled estimate, not a measurement. The source data only has 3-year cohort totals, with no month-by-month revenue timeline, so payback assumes each cohort's net revenue (sales after refunds and servicing costs, before acquisition spend) accrues evenly over 36 months. It deliberately excludes acquisition cost from that monthly rate — that cost is what's being paid back, so it can't also be baked into the recovery rate.
＊
"Conversions" is fractional, not a whole headcount. Values with decimals come straight from the underlying data model rather than a simple count — treat it as a very close proxy for new subscribers, not an exact number of people.
＊
LTV:CAC and Net ROAS measure different things. LTV:CAC compares LTV (revenue after refunds and servicing costs) to spend. Net ROAS compares net revenue plus lead-gen revenue — LTV with expert payouts also subtracted — to spend, so it's a more fully-loaded number and usually a bit lower than LTV:CAC. Both are shown because they answer different questions, not because one is wrong.
＊
Scope. This (synthetic) data models US, paid-search-family (PPC) traffic on a single no-trial subscription product — it is a slice of a hypothetical business, not the whole of it.
Is the subscription itself healthy?
Unit economics tell you whether a channel is worth the spend. These five numbers tell you something different: once someone subscribes, how well does the business hold onto that revenue — regardless of channel? A cohort's numbers only fully settle roughly 2.5 years after acquisition, so treat anything more recent than that as a preview, not a final answer.
Each point sums the raw dollars over its trailing 12 cohorts, then divides — so it's not just a smoothed average, and the first 11 months per channel have no point yet (not enough history for a full window).
3-year return rateiRefunds and chargebacks (upfront + subscription) as a share of total forecast sales. Lower is better — it means fewer customers ask for their money back.
Total 3-year returns ÷ total forecast sales (GSS)
Retention proxyiA simple stand-in for retention: the share of sales the business keeps after refunds. It is not a month-by-month churn curve — there's no data here for how long any individual subscriber stays.
1 − 3-year return rate
Dispute rate
Dispute-management fees (paid to firms that help fight chargebacks) ÷ total forecast sales
Customer service ticket rate
Cost of CS agents handling customer tickets ÷ total forecast sales
Contribution margin, % of sales
3-year contribution margin ÷ total forecast sales (GSS) — a different slice than "Margin as % of LTV" on the Unit Economics tab
The business, in plain English
This dashboard models a hypothetical subscription business that connects customers with on-demand experts, for a subscription fee. It looks at one specific slice of that kind of business: US customers who signed up for a "No Trial" subscription (they start paying immediately, no free trial period) after clicking a paid ad.
Search
A text ad shown on Google or Bing when someone actively searches for something, e.g. "talk to an expert online." The customer is already looking for help.
Display
A banner or visual ad shown on other websites and apps while someone browses — not actively searching. It's more like a billboard than an answer to a question.
Performance Max (PMax)
A newer, automated Google ad format. One campaign, and Google's own algorithm decides how to split the budget across Search, Display, YouTube, Gmail and more to get the most subscribers.
Following one customer through the numbers
Every term below uses the same running example: the Search channel's August 2026 cohort of 75,866 new subscribers — the same cohort shown in the revenue bridge at the top of the Unit Economics tab. Dollar figures are shown per-customer (the cohort total ÷ 75,866) so they're easy to picture as "one typical subscriber."
Money coming in
Conversion
One new "No Trial" subscription sold. It's the unit this whole dashboard counts everything per.
Example: 75,866 conversions this cohort
Upfront GSS
The one-time payment collected right at signup — the join fee (plus things like premium add-ons or bonuses, when they apply).
Example: $4.39 per customer
Subscription GSS
The recurring monthly payments a subscriber is expected to make, added up over their first 3 years.
Example: $193.64 per customer, forecast
Total GSS ("gross sales")
Upfront payment plus subscription payments — everything a cohort is expected to pay before anything is subtracted.
Upfront + Subscription = $198.03 per customer
Fact Sub GSS %
How much of the forecasted subscription revenue has actually been collected so far, versus still being projected. The upfront payment is booked immediately — this % is specifically about the ongoing monthly payments, which take years to fully play out.
Example: only 2.0% collected so far (brand-new cohort)
Money that comes back out before we know what's left
Returns (refunds & chargebacks)
Money paid back to customers who asked for a refund or disputed the charge with their bank, on both the upfront and subscription payments.
Example: ~$40 per customer, or 20.3% of total sales
Dispute fees
The business pays outside firms (chargeback-mitigation vendors) to help fight disputes. This is the fee for that service — separate from the disputed amount itself, which is already counted in "Returns."
Example: $2.12 per customer
CS ticket costs
What it costs to pay the customer service agents who handle tickets and requests from these customers.
Example: $1.32 per customer
Other costs
Payment-processing transaction fees, plus VAT (sales tax) in markets where it applies.
Example: $9.12 per customer
Expert payout
What the business pays the Experts who actually answer the customer's questions.
Example: $13.56 per customer
What's left, and what it cost to get here
LTV (Lifetime Value)
Total sales minus returns, dispute fees, CS costs and other costs. What the customer is worth before we pay Experts or count what it cost to acquire them.
Example: $145.30 per customer
Net revenue
LTV minus what the business pays its Experts. Still doesn't include acquisition cost.
Example: $131.74 per customer
Lead-gen revenue
Extra money the business makes by selling some customer leads to other companies — a small revenue bonus on top.
Example: $1.23 per customer
Acquisition cost (CAC)
What the business paid Google, Bing or an affiliate in ads to win this one customer.
Example: $92.20 per customer
Contribution margin
Net revenue, plus lead-gen revenue, minus acquisition cost. The real bottom line: what this customer actually contributed to the business after every cost, including the ad spend to acquire them.
Example: $40.78 per customer
LTV : CAC ratio
For every $1 spent acquiring this customer, how many dollars of LTV come back. Not an official metric from the underlying model — a standard subscription-industry yardstick, added here for comparison. 3× or higher is generally considered healthy.
Example: 1.58× ($145.30 LTV ÷ $92.20 CAC)
Net ROAS
Net revenue plus lead-gen revenue, divided by acquisition cost. The model's own official return-on-spend figure — similar to LTV:CAC, but using the more fully-loaded net revenue instead of LTV.
Example: 1.44× (vs. 1.58× LTV:CAC)
Payback period
How many months it takes net revenue to earn back the acquisition cost. Not an official report metric — modeled here by spreading the 3-year net revenue evenly across 36 months, so it's a rough guide, not a precise cash-flow forecast.
·
Channel Economics