Unit Economics for SaaS: Calculate LTV, CAC, and Payback with AI

Unit economics is the set of metrics that determine whether a business is profitable at the level of a single customer. For subscription products, three core metrics — LTV, CAC, and Payback Period — answer the essential questions: how much does it cost to acquire a customer, how much will they generate over their lifetime, and when does the investment pay off. AI tools like Claude let you calculate all of this in minutes instead of hours of manual spreadsheet work.

This article covers formulas, benchmarks, ready-to-use prompts for Claude, a worked example with real numbers, and sensitivity analysis — everything you need to evaluate SaaS product economics.

Three Core SaaS Unit Economics Formulas

LTV (Lifetime Value) — Total Revenue per Customer

Total revenue from a single customer over their entire time using the product.

The basic formula for subscription models:

LTV = ARPU / Churn Rate

ARPU (Average Revenue Per User): average monthly revenue per user. Churn Rate: the share of users who leave each month.

Example: ARPU $29/month, monthly churn 5%.

LTV = $29 / 0.05 = $580

Average customer lifespan at 5% churn: 20 months (1 / 0.05).

The formula is simple, but the details matter. Which ARPU do you use — mean across all plans or median? Gross churn or net churn (accounting for upgrades)? Revenue churn or logo churn? Each choice gives a different number.

A more accurate formula that accounts for gross margin:

LTV = (ARPU × Gross Margin) / Churn Rate

With 80% gross margin (typical for SaaS): LTV = ($29 × 0.8) / 0.05 = $464. Lower, but it reflects actual profit.

CAC (Customer Acquisition Cost) — Cost to Acquire One Customer

The cost of acquiring one paying customer.

CAC = (Marketing Spend + Sales Spend) / New Customers

Example: $10,000 on marketing and $5,000 on sales in a month, 50 new customers:

CAC = $15,000 / 50 = $300

A common mistake: counting only the ad budget. CAC includes marketer salaries, tools, content production, and sales team compensation.

An important distinction: blended CAC vs paid CAC. Blended includes organic traffic (all costs divided by all customers). Paid: only paid channels divided by customers from paid channels. If blended looks good but paid doesn’t, organic growth is masking a scaling problem.

Payback Period — Time to Recoup Acquisition Cost

How long it takes to recoup the CAC.

Payback Period = CAC / (ARPU × Gross Margin)

With CAC $300, ARPU $29, and margin 80%:

Payback Period = $300 / ($29 × 0.8) = 12.9 months

Nearly 13 months before a single customer breaks even. Until that point, every acquired user generates a loss.

SaaS Unit Economics Benchmarks

Reference points for SaaS products:

Metric	Poor	OK	Good	Great
LTV:CAC	< 1:1	2:1	3:1	> 5:1
Payback Period	> 24 mo	12–18 mo	6–12 mo	< 6 mo
Monthly Churn	> 8%	5–8%	3–5%	< 3%
Net Revenue Retention	< 90%	90–100%	100–120%	> 120%

LTV:CAC of 3:1 is the classic target. Below 1:1, every customer is a net loss. Above 5:1: either you’re underinvesting in growth, or the data is off.

Payback Period under 12 months for an early-stage product is a strong signal. Over 18 months raises the question of whether you have enough cash to survive until payback.

Net Revenue Retention above 100% means upgrades outweigh churn. Older cohorts bring in more money over time. Growth is possible even without acquiring new customers.

Why AI Beats Spreadsheets for Unit Economics

A static Google Sheets table gives you one data point. Decisions need a range.

“LTV:CAC = 3.2” — and then what? What happens if churn goes up 2%? How does the economics change when you scale CAC? What’s the effect of a $10 price increase?

Each of those questions means a new formula, a new cell, another hour in a spreadsheet. Or one prompt.

The same principle applies across AI infrastructure. For example, monitoring LLM calls with Langfuse solves an analogous problem: turning raw data into actionable metrics, but for AI pipelines instead of business economics.

Step-by-Step: Calculating Unit Economics with Claude

Step 1. Export Data from Your Payment System

From your payment system (Stripe, Paddle, or any other): a CSV with transactions. Minimum fields: date, user_id, amount, type (new / renewal / churn).

From analytics: number of new users per month and marketing spend.

Step 2. Calculate Base Metrics with a Single Prompt

Prompt for Claude:

Here's a CSV with Stripe transactions from the last 12 months [attach file].

Calculate:
1. ARPU by month
2. Monthly churn rate (logo and revenue)
3. LTV using the formula ARPU × Gross Margin / Churn Rate (margin 82%)
4. CAC — here are the marketing costs by month: [paste]
5. LTV:CAC ratio
6. Payback period in months

Show the result as a table. Below the table: three main takeaways.

Beyond the raw numbers, the model will spot trends that are easy to miss in manual analysis: churn climbing for three months straight, a 2x gap between paid and blended CAC, anomalies in specific cohorts.

Step 3. Sensitivity Analysis — the Key AI Advantage

This is the key advantage of the AI approach: you can run dozens of scenarios instantly.

Take the current metrics from the previous calculation.

Build a sensitivity analysis:
- Churn: from -3% to +3% vs current (1% step)
- ARPU: from -$10 to +$10 ($5 step)
- CAC: from -20% to +20% (10% step)

For each combination, show LTV:CAC and Payback Period.
Highlight in red any combination where LTV:CAC < 2:1.
Which parameter has the strongest impact on unit economics?

The result: a table with 30+ scenarios. Doing this manually takes at least half an hour. With AI — under a minute.

A typical insight from sensitivity analysis: for most SaaS products, churn affects LTV more than price does. Reducing churn by 1% gives a bigger LTV boost than raising the price by $5. Yet effort is often allocated the other way around: weeks experimenting with pricing, minimal work on retention.

Sensitivity analysis makes this visible. Not in theory — on your actual data.

Step 4. Cohort Analysis to Uncover Hidden Problems

Averages hide reality. Average LTV can mask the fact that one cohort churns in 3 months while another stays for a year.

Here's a CSV with transactions. Split users into cohorts
by the month of their first payment.

For each cohort, show:
- Retention by month (% remaining from the initial size)
- Revenue retention (% of revenue from the initial amount)
- LTV to date
- Projected full LTV (extrapolating the retention curve)

Which cohorts are abnormally good or bad? What might have changed
in the product between them?

The model will build retention curves and find patterns: a retention jump after a certain month might point to an onboarding change, a product update, or a shift in acquisition channel.

Step 5. Forecasting Unit Economics for the Next 6 Months

Based on current metrics and 12-month trends,
forecast unit economics for the next 6 months.

Three scenarios:
1. Base: current trends continue
2. Optimistic: churn drops by 1% per month,
   ARPU grows 3% (planned price increase)
3. Pessimistic: churn grows 0.5% per month,
   CAC rises 15% (seasonal ad cost increase)

For each scenario: MRR, customer count, LTV:CAC,
Payback Period, cash flow from new customers.

Three scenarios by hand means three separate models in Google Sheets with cross-references. One prompt replaces all of that.

Worked Example: B2B SaaS at $11K MRR

A numerical example for illustration. B2B SaaS, two plans ($19/mo and $49/mo), 14 months on the market.

Input data:

• 420 paying customers (310 on the $19 plan, 110 on the $49 plan)
• MRR: $5,890 + $5,390 = $11,280
• Monthly churn: 6.2% (logo), 4.8% (revenue — upgrades partially offset losses)
• Marketing spend: $4,200/mo (Google Ads $2,800, content $1,400)
• New customers: ~35/mo
• Gross margin: 78%

Calculation:

ARPU = $11,280 / 420 = $26.86
LTV = ($26.86 × 0.78) / 0.048 = $436
CAC = $4,200 / 35 = $120
LTV:CAC = $436 / $120 = 3.6:1
Payback Period = $120 / ($26.86 × 0.78) = 5.7 months

At first glance, solid numbers. LTV:CAC above 3:1, payback under six months. But a deeper look reveals three problems:

1. Logo churn at 6.2% is above the benchmark. Average customer lifespan is 16 months. For B2B SaaS, that’s short. If churn rises to 8%, LTV drops to $340 and LTV:CAC falls to 2.8:1.

2. Paid CAC is significantly higher than blended. Of 35 new customers per month, 12 come from organic (SEO, word of mouth). Paid CAC = $4,200 / 23 = $183. LTV:CAC for the paid channel = 2.4:1 — already borderline.

3. ARPU depends on the plan mix. The share of customers on the $19 plan is growing (74% vs 68% three months ago). If the trend continues, ARPU will drop within six months, and paid-channel LTV:CAC could slip below 2:1.

None of these problems show up in a static table with three formulas. All three become obvious with a full data analysis.

Ready-to-Use Prompts for Common Unit Economics Tasks

Comparing acquisition channels:

Here are the costs and conversions by channel for the quarter:
- Google Ads: $8,400, 69 customers
- Content/SEO: $4,200, 36 customers (estimated)
- Referral: $0, 22 customers
- Product Hunt launch: $500, 45 customers

Calculate CAC by channel. Note that retention differs:
Google Ads customers have 8% churn, SEO 4%, Referral 3%,
Product Hunt 12%.

Calculate LTV:CAC by channel. Where should the next $5,000 go?

Modeling a pricing experiment:

Current price: $29/mo, 400 customers, churn 5%.
Planning to raise to $39/mo.

Survey data: 15% of current customers are "likely to leave"
if the price goes up. New customer conversion may drop 10–20%.

Model three options:
1. Raise for everyone at once
2. Grandfather existing customers, $39 for new ones
3. New $39 tier with extra features, keep the old plan

For each: projected MRR, LTV, CAC impact over 12 months.

Preparing an investor-ready unit economics summary:

Here are the product metrics: [data]

Prepare:
1. A one-pager with unit economics (3–5 key metrics)
2. Three most likely questions about the metrics and answers
3. Weak spots in the economics and a plan to address them

Limitations of the AI Approach to Unit Economics

Garbage in, garbage out. If the input data is wrong, the calculations will be formally correct and practically useless. AI won’t verify whether your churn is calculated correctly. It takes the number you provide and builds a model on top of it.

No market context. The benchmark “5% churn is fine” doesn’t account for the fact that in your specific niche, the standard might be 2%. You need to add market context yourself.

Tendency toward smooth trends. AI forecasts tend toward clean curves. Reality is jagged: seasonality, competitor moves, technical outages. It’s worth requesting the pessimistic scenario separately.

Not a replacement for a finance specialist. For early stages, AI calculations are sufficient. For serious financial modeling — the kind that venture funds and auditors require — you need a specialist who knows their methodologies. Similarly to how a multi-provider LLM architecture doesn’t replace thorough testing of each model, an AI calculator doesn’t replace financial due diligence.

Retrospective analysis, not prediction. AI is good at dissecting the past. Forecasts beyond 12 months are high-uncertainty territory. Scenarios beat point predictions.

How to Start Calculating Unit Economics Today

A minimum plan for 30 minutes:

1. Export a CSV from your payment system for the last 6 months
2. Use the prompt from Step 2 to calculate base metrics: LTV, CAC, LTV:CAC, Payback Period
3. Run the sensitivity analysis from Step 3
4. Write down three numbers: LTV:CAC, Payback Period, Monthly Churn

These three metrics are the minimum monitoring set. Updating them once a month gives you trend visibility and the ability to react to changes in time.

The quality of your calculations depends directly on how you structure your prompts. If you want to go deeper, the context engineering guide shows how to structure AI requests for maximum accuracy.

Unit economics isn’t financial wizardry. It’s three divisions and one multiplication. AI removes the only barrier: the manual work with data. What remains is a choice — calculate or keep guessing.