CAC, LTV, ROAS and the rest: how to measure online marketing (with real benchmarks)

Every ad platform hands you a dashboard full of acronyms: CPC, CTR, CPA, ROAS. Most of them describe the ad. Very few describe the business. You can have a cheap click, a great click-through rate and a "4x ROAS" and still lose money on every customer.

The two numbers that decide whether marketing works are simple: what it costs to get a customer (CAC) and what that customer is worth (LTV). Everything else is a way of explaining why one of those two moved.

This guide covers what each metric means, how to calculate it without fooling yourself, and what typical CACs look like in specific industries, so you know whether your number is normal, great or a warning sign.

The two numbers that matter

CAC: customer acquisition cost

CAC = everything you spent to win new customers ÷ the number of new customers you won, over the same period.

The trap is the top half of that fraction. Most people only count ad spend. A fully loaded CAC also includes:

• Ad spend on every platform
• Agency, freelancer and contractor fees
• Marketing and sales salaries (or the share of time spent on acquisition)
• Tools: email platform, SEO tools, landing page builder, analytics
• Creative production: video, photos, UGC
• Affiliate and referral commissions, and discounts given to first-time buyers
• Sales commissions, if anyone sells by hand

The bottom half has a trap too: count new paying customers, not sign-ups, leads, installs or trials. Those are steps on the way, and they have their own metrics (below).

Example: in March you spent $6,000 on Meta and Google, $1,500 on a freelancer, $500 on tools and $1,000 of your own salary-equivalent time. You won 60 new paying customers. Paid-only CAC looks like $100. Fully loaded CAC is $9,000 ÷ 60 = $150. Both are useful, but only the second tells you if the business works.

It's also worth splitting CAC two ways:

• Blended CAC divides all spend by all new customers, including the ones who found you through word of mouth or organic search. It flatters paid channels.
• Paid CAC divides paid spend only by customers who came from paid channels. It's the number to use when deciding whether to raise an ad budget.

LTV: lifetime value

LTV is how much gross profit a customer brings in over the whole time they stay. Not revenue: profit after the cost of delivering what you sold.

For a subscription business:

LTV = average monthly revenue per customer × gross margin ÷ monthly churn rate

A customer paying $50 a month, at 80% gross margin, with 5% of customers cancelling each month: $50 × 0.8 ÷ 0.05 = $800.

For an ecommerce store:

LTV = average order value × gross margin × number of orders a customer places over (say) 12 or 24 months

A $70 average order, 50% margin, and 1.8 orders per customer in their first year: $70 × 0.5 × 1.8 = $63 in first-year LTV.

Two warnings. First, use a time limit. "Lifetime" for a brand-new business is a guess, and a 2% churn rate turns into a fantasy 50-month lifetime. Many teams use 12-month or 24-month LTV instead. Second, measure churn on real cohorts (everyone who joined in January, then everyone who joined in February) rather than averaging across the whole customer base, which hides whether newer customers are leaving faster.

LTV:CAC ratio

Divide LTV by CAC. The common rule of thumb is at least 3:1: each customer returns three times what they cost to win, leaving room for overheads and profit.

• Below 1:1 you lose money on every customer. Growth makes it worse.
• Around 1–2:1 the business is fragile. One rise in ad prices or one bad churn month and it tips over.
• 3:1 or higher is healthy.
• Above 5:1 usually means you're under-spending. You could probably buy more customers profitably.

For context, Benchmarkit's B2B SaaS data puts the median LTV:CAC at about 4x, and DTC ecommerce brands are often reported closer to 2.4:1, because repeat purchase is harder to earn than a subscription renewal.

CAC payback period

LTV:CAC tells you if a customer is profitable eventually. Payback tells you how long your cash is tied up before it comes back.

Payback (months) = CAC ÷ (monthly revenue per customer × gross margin)

With the $150 CAC and $50/month at 80% margin from above: $150 ÷ $40 = 3.75 months.

Payback matters most for small businesses, because you have to fund every month of that gap yourself. Rough guides:

The SaaS median comes from the Aleph and Benchmarkit 2026 report, which found payback improved from 18 to 16 months between 2024 and 2025.

Typical CAC by industry

These are averages, and averages hide a lot. Your price, your country, your channel mix and how you count all move CAC more than your industry does. Use these to sanity-check, not as targets.

B2B companies (blended organic + paid)

From First Page Sage's dataset of client accounts from January 2022 to August 2025. "Organic" is mostly SEO and organic social; "paid" is mostly search ads and paid social.

Notice the pattern: paid CAC is higher than organic in almost every industry, often double. Search and content take longer to build but keep paying back. Paid gives you customers this week, at a premium.

SaaS by vertical

From the same source, average CAC for software companies selling into each industry:

Within SaaS, price and sales motion matter more than vertical. Self-serve products where people sign up with a card and no sales call typically land around $100–$500 per customer. SMB products with some sales help land around $200–$700. Enterprise deals with a sales team routinely cost $10,000+ per customer, and are worth it because contracts are large.

Ecommerce and DTC brands

Paid CAC for direct-to-consumer brands is usually quoted somewhere between $20 and $120, depending on the category and on whether creative, agency and discount costs are included. Commonly cited paid-media figures by category:

The ranges are wide because different reports count differently. The thing to compare is not CAC alone but CAC against first-order profit. A $40 CAC on a $35 t-shirt with 50% margin loses money on day one, and only works if that customer comes back. A $90 CAC on a $60/month supplement subscription is fine if they stay four months.

B2C service businesses and startups

First Page Sage's startup benchmarks split B2B from B2C. A few consumer-facing examples:

Mobile apps

App marketers usually measure in two steps: cost per install (CPI) and cost per paying user.

• CPI: Business of Apps' 2025 data puts average CPI around $4.70 on iOS and $3.40 on Android. Casual games can be well under $2. Finance apps typically run $3.50–$8, and fintech or insurance on iOS can reach $10–$25.
• Cost per paying subscriber: for North American iOS subscription apps, AppsFlyer data puts the median at roughly $45–$140, depending on category.
• How few installs pay: RevenueCat's 2025 report found a median of just 1.5–2.7% of downloads become paying users within 35 days (around 12% for apps with a hard paywall), and a median revenue per install of $0.31 after 60 days.

That last number is the reality check. If an install costs $4 and the average install earns $0.31 in two months, you need retention, annual plans or a much better paywall before paid installs make sense. RevenueCat also found yearly plans keep 44% of subscribers after a year, against 17% for monthly plans, which is why so many apps push annual pricing.

The funnel metrics: what explains your CAC

CAC is the result. These are the levers. If CAC goes up, one of these moved.

The chain is: CAC = CPC ÷ conversion rate from click to paying customer. If clicks cost $2 and 1% of clicks become customers, CAC is $200. Double the conversion rate and CAC halves, without touching the ads. That's why fixing the landing page or the onboarding is often the cheapest way to lower CAC.

Some reference points for Google search ads, from LocaliQ's 2025 benchmarks: an average CPC around $5.26, an average click-to-conversion rate around 7.5%, and an average cost per lead of about $70, ranging from about $29 for auto repair and $32 for pets up to $104 for business services and $132 for legal services.

A note on CPA: it's only as meaningful as the conversion you choose. A CPA on "sign-ups" is a CPL. A CPA on "purchases" is close to CAC, if each purchase is a new customer. Ad platforms happily report a low CPA on a cheap event. Make sure the event you optimize for is one that leads to money.

Revenue metrics: ROAS, MER and friends

ROAS: return on ad spend

ROAS = revenue attributed to ads ÷ ad spend

A 4x ROAS means $4 of revenue for every $1 of ads. Two problems:

1. It's revenue, not profit. At a 25% margin, a 4x ROAS is exactly break-even. Your break-even ROAS is 1 ÷ gross margin: 2x at 50% margin, 4x at 25%, 1.25x at 80%.
2. It's the platform's claim. Meta, Google and TikTok each attribute the same sale to themselves under their own rules. Add up every platform's ROAS and you'll usually "find" more revenue than you made.

MER: marketing efficiency ratio (blended ROAS)

MER = total revenue ÷ total marketing spend

MER ignores attribution entirely. It's your store or billing system's revenue divided by everything you spent. It can't tell you which ad worked, but it can't be double-counted either. A good habit: use platform ROAS to compare ads within a platform, and MER to decide if marketing as a whole is working. Many brands also track new-customer MER (revenue from first orders only ÷ spend) because returning customers would have bought anyway.

The rest of the alphabet

• AOV (average order value): revenue ÷ orders. Raising it with bundles or free-shipping thresholds lowers the CAC you need.
• ARPU / ARPA (average revenue per user or account): the subscription version of AOV.
• MRR / ARR: monthly or annual recurring revenue.
• Churn: the share of customers (or revenue) lost each month. The single biggest input to LTV.
• NRR (net revenue retention): revenue this year from last year's customers ÷ their revenue last year, including upgrades and lost accounts. Above 100% means existing customers grow faster than they leave.
• Contribution margin: revenue minus product costs, shipping, payment fees and ad spend. The truest "did this order make money" number for ecommerce.
• Incremental CAC: the cost of the extra customers a budget increase brought, not the average. Doubling spend rarely doubles customers; the next customer is always more expensive than the average one.

How to measure all this without fooling yourself

1. Get the source of truth right. Revenue comes from your billing system or store (Stripe, Shopify, the App Store), not from the ad platform. Customers come from your database.
2. Track the event that matters. Make sure your ads and analytics see the purchase or the paid subscription, not just the sign-up. If they can only see sign-ups, they'll find you people who sign up.
3. Tag every link. UTM parameters on every ad, email and social post, so your analytics can split customers by source.
4. Measure in cohorts. Group customers by the month they joined and follow each group. Churn, LTV and payback all hide problems when averaged across everyone.
5. Use the same window everywhere. A 7-day click window on one platform and 30-day on another makes comparisons meaningless.
6. Count the full cost. Paid CAC for daily decisions, fully loaded CAC for the monthly "does this business work" check.
7. Check the trend, not the snapshot. One week's CAC is noise. A rising four-week average is a signal.

A one-page scorecard

If you only look at a handful of numbers each month, make it these:

Benchmarks tell you if you're in the right neighbourhood. Your own trend tells you if you're getting better. Measure honestly, compare against your margin, and the right budget decisions mostly make themselves.

Benchmark sources: First Page Sage (B2B, SaaS and startup CAC), LocaliQ (search ad CPC, conversion rate and CPL), RevenueCat (subscription app conversion, retention and revenue per install), the Aleph and Benchmarkit 2026 SaaS report (payback), Business of Apps and AppsFlyer (app install and subscriber costs), and DTC benchmark compilations for ecommerce categories. Published benchmarks use different methods and samples; treat every figure as a range, not a target.