TL;DR

• CAC is what you spent to get a customer. Most crypto teams report it as cost per signup, and a signup is the cheapest, least valuable thing in the funnel. The number that matters is cost per funded user.
• Almost every crypto CAC benchmark in circulation has no source. We traced the most-quoted figures back and found nothing underneath them.
• Two public companies give you real numbers. eToro's 2025 results imply roughly $573 to $628 per net new funded account. Coinbase's S-1 showed $5.15 per verified user. Both are correct, and they measure completely different things.
• The gap between them is the whole problem. Run eToro's own filing with the wrong denominator and you get $47 instead of $573, from the same two figures.
• Calculate signups, funded users, and first actions separately, then use retention as the tiebreaker. Sources for everything are at the end.

‍

What Is Customer Acquisition Cost?

Customer acquisition cost, or CAC, is one division. You take everything you spent to get customers, and you divide it by how many customers you got.

Spend $50,000 in a month and get 500 customers, your CAC is $100.

That's it. The math never gets harder than that. Everything that goes wrong with CAC goes wrong in the two inputs, not in the division.

The top number is bigger than you think. Most teams put media spend in there and stop. But the KOL you paid, the agency retainer, the referral payouts, the signup bonus, the tokens you handed out: all of it was spent to get those users, so all of it belongs in the numerator.

The bottom number is where crypto breaks. In most businesses, "customer" is obvious. Someone paid you. In crypto, a person can create an account, verify their identity, connect a wallet, and never move a dollar. Are they a customer? Your CAC answer depends entirely on how you answer that, and most teams never decide on purpose. They just count whatever their analytics tool counts by default.

That default is almost always the signup. Which is how you end up with a number that looks great and tells you nothing. And that could cost you dearly in the end.

‍

A Simple Example

You run a campaign. It brings 5,000 signups for $50,000. You report a $10 CAC and everyone's pleased.

Three weeks later, finance asks how many of those people actually funded an account.

The answer is 180.

Your real acquisition cost was $278 per funded user, not $10. You just spent a month calling a bad channel a good one, and you may have already doubled its budget.

Nothing about that campaign changed between the two numbers. The spend was the same. The people were the same. The only thing that changed was what you agreed to count as a customer.

Now imagine you ran a second channel that month. It was expensive per signup, so you paused it. But 40 percent of its signups funded. On signup CAC it looked like your worst channel. On funded CAC it was your best one, and it's now switched off.

That reversal is the most expensive measurement error in crypto marketing, and it's happening in most reports right now.

‍

Your CAC Number Is Probably Measuring the Wrong Thing

So why does everyone keep reporting the cheap number? Not because teams are careless. Because every incentive in the process points that way.

‍Signups are easy to count. The event fires on your own site, it needs no engineering work, and it happens fast enough to report on while the campaign is still running.

Funding is hard to count. It happens days later, often on-chain, sometimes after the user has come back through a different route. Connecting that deposit back to the campaign that caused it needs tracking that somebody has to build, and it's never anyone's top priority.

Nobody wants the smaller number. A $10 CAC is a good slide. A $278 CAC needs an explanation. The incentive at every level, agency and in-house alike, favors the metric that looks better.

The lag hides it. Signup to funded can take a week or more. The campaign report goes out before the real number exists, and nobody goes back to update it.

Verification sits in the middle and nobody owns it. For exchanges, a large share of signups drop out during KYC. That's mostly a product problem, but it lands on marketing's number, so both sides quietly ignore it.

And other people's numbers are worse than yours. When a project or an agency shows you a CAC in a deck, it's usually napkin math: total spend divided by total users, media only, signups counted as customers. Rarely a lie, and always flattering. Then you benchmark yourself against it, or you pick the agency quoting $12 over the one quoting $180, without knowing the second was counting funded users. One question sorts it out. Which stage are you counting, and what's in the numerator? A team that measures properly answers in a sentence.

None of this is fixed by trying harder. It's fixed by counting three things instead of one, and by asking anyone who quotes you a number which of the three they meant.

‍

The Three Stages to Measure Separately

- For an exchange, that's signup, funded account, first trade. 

- For a wallet, install, wallet created, first transaction. 

- For a DeFi protocol, connect, first deposit, second deposit.

The labels change and the logic doesn't.
Report cost at every single stage, plus the conversion rate between them. The conversion rates are where the story is: a channel with a great signup cost and a 3 percent funding rate is not a great channel.

One warning about cost per wallet.
It's the most quoted metric in crypto acquisition and the least comparable. A connected wallet is not a customer. Whenever someone quotes you a CAC, the first question is always which stage they counted.

‍

How to Calculate It

Three stages, three formulas.

Cost per signup = total acquisition cost divided by new signups

Cost per funded user = total acquisition cost divided by users who funded‍

Cost per first trade = total acquisition cost divided by users who completed a first trade

The numerator is the same in all three. Total acquisition cost means everything you spent to get those users:

• Media spend
• KOL and creator fees
• Agency fees attributable to the channel
• Referral and affiliate payouts
• Signup bonuses and deposit matches
• Token incentives, valued at what they were worth on the day granted
• Quest, bounty and points program budgets
• Compliance review time on campaign creative, which for regulated products is a per-campaign cost rather than overhead
• Fraud and incentive-abuse write-offs

The last three are the ones almost nobody counts, and in incentive-heavy crypto campaigns they can be bigger than the media line.

Two rules keep the numbers honest.

Use the same window every time. Pick 30 days from signup and stick to it. A funded user who took 45 days doesn't count in this month's number, and that's fine, as long as the rule never moves.

Count the user in the channel that acquired them, and decide your attribution rule once. Last touch is the common default. It's imperfect, and it's far better than changing the rule between reports.

‍

A Worked Example

The numbers below are invented for illustration. They are not benchmarks and shouldn't be used for planning.

A hypothetical exchange runs three channels for one month.

Now the three costs:

Read the top row and paid social is your best channel by a factor of four. Read the bottom row and it's your worst by a factor of nearly three.

The newsletter looks terrible on signup cost and turns out to be a strong channel. It brings few people, and the people it brings are already qualified.

The KOL campaign wins on the metric that matters, and it would have been ranked second on the metric most teams report.

Same spend, same month, opposite decisions.

‍

Where CAC Benchmarks Actually Come From

Once you're measuring properly, the next question is what to compare yourself against. Before you do that, it's worth knowing what you'd be comparing against.

We went looking for the origin of the numbers that circulate in this category. The result isn't encouraging.

We read the pages currently ranking for crypto acquisition cost and checked every figure on them for a source. One publishes roughly forty separate cost figures across product types and channels, and states no methodology and names no data source for almost any of them. Another reports DeFi at $85 per user, exchanges at $150, and airdrop-acquired users at $500 to $1,000, none of it attributed. The precise-looking numbers that circulate most widely, things like $3.12 per wallet or $1.86 per stablecoin checkout, trace back to pages like those and then stop. There's nothing underneath them.

We used some of those numbers in an earlier draft of this article. We've taken them out.

This isn't an accusation of bad faith. It's what happens in a category where almost nobody publishes real acquisition data, so estimates get repeated until they acquire the texture of fact. The practical consequence for you is simple: if a benchmark has no source, treat it as a rumor with a decimal point.

There are exceptions, and they're worth more than everything else combined.

‍

Two Real Numbers, From Public Filings

Public companies have to disclose. Two of them let you calculate crypto acquisition cost from primary documents rather than from a blog.

eToro: Roughly $573 to $628 per Net New Funded Account

In its fourth quarter and full year 2025 results, also filed with the SEC, eToro reported:

• 3.85 million funded accounts at year end, up 9% year on year
• $868 million net contribution
• Marketing at roughly 21% of net contribution, with guidance to scale toward 25% in 2026

Work it through. A 9% increase from 3.85 million implies about 3.53 million at the end of 2024, so net new funded accounts for the year were roughly 318,000. Marketing at 21% of $868 million is about $182 million.

$182 million divided by 318,000 is $573 per net new funded account. Use a slightly higher marketing figure and you land near $628.

That's a real, checkable acquisition cost for a regulated trading platform with a large crypto business. Not an estimate. Arithmetic on a filing you can open yourself.

‍

Now Watch the Denominator Break It

Take the same two numbers. Divide $182 million by the 3.85 million total funded accounts instead of the 318,000 new ones.

You get $47.

Same filing. Same year. Same company. One number is twelve times the other, and both are arithmetically correct. Only one answers the question "what did it cost to acquire a customer."

We flag this because we watched it happen. A summary of eToro's own results circulating online performs exactly this division and reports roughly $47 to $52 as the cost per funded account. It isn't a typo. It's the denominator problem, live, on a public company's audited numbers.

If it can happen to a filing, it's certainly happening in your campaign reports.
‍

Coinbase: $5.15, and Why That Number Misleads Everyone

Coinbase's S-1, filed ahead of its 2021 listing, is the source of the most-quoted acquisition figure in crypto history. The company reported sales and marketing expense of $56.8 million for 2020 and added 11.0 million verified users, taking it from 32 million to 43 million. That's a blended $5.15 per verified user, up from $4.02 in 2019.

Five dollars. In a category where you've just seen a comparable business spend $573.

Both are true, and the difference is entirely definitional. A verified user, in Coinbase's disclosure, is somebody who created an account and verified their identity. Not somebody who deposited. Not somebody who traded. The funnel hadn't started yet.

This is the single most important thing to understand about crypto CAC benchmarks. The famous low numbers are signup-level numbers. When somebody quotes you a $5 or $20 CAC, they're almost always counting registrations, and when somebody quotes $200 or $600, they're counting funded customers. These aren't competing estimates of the same thing. They're different stages of the same funnel.

Coinbase's later disclosures make the contrast sharper. The company has guided to marketing spend in the hundreds of millions per quarter during market peaks. The $5 era ended when the market changed and the company started buying users instead of receiving them.

Which leads to the last thing the public numbers tell you: crypto CAC is cyclical. Coinbase's own blended figure moved from $4.02 to $5.15 and then to a multiple of that within two years. Any benchmark you read carries the market conditions of the month it was measured.

‍

What You Can Actually Benchmark Against

Here's an honest table. Where a figure comes from a named, checkable source we say so. Where it doesn't, we say that too, because a benchmark you can't trace isn't a benchmark.

Two honest conclusions from that table.

There's no reliable public benchmark for crypto CAC by channel. Anyone presenting one is estimating. That includes us, which is why we've stopped.

Your own funnel is better data than anyone's benchmark. A 90-day internal baseline across the three stages is more useful for your decisions than any industry average, because it uses your product, your market, and your definitions.

The Airdrop Problem

Incentive programs deserve their own section, because they're the most expensive mistake in crypto acquisition and they're invisible on a signup-CAC report.

Airdrops, points programs, and quest campaigns work exactly as designed. They produce enormous volume at an apparently trivial cost per user. Then most of those users leave, because the reward was the reason they came.

Three things make the true cost hard to see.

The tokens feel free. They're not. Value them at what they were worth on the day they were distributed, not at zero and not at today's price.

The denominator is wrong. Dividing by everyone who claimed produces a flattering number. Divide by the users still active after 90 days and it usually multiplies by ten or more. This is the eToro error again, in a different costume.

The damage isn't only cost. Incentive-acquired users sell, which pressures the token, which affects the next campaign. A cohort acquired by reward behaves differently from a cohort acquired by interest, and mixing them in one CAC number hides both.

The practical fix is to report incentive-acquired users as a separate cohort. Same three stages, own line. If they retain as well as your organic cohort, excellent. Usually they don't, and you want to know that before funding the next round.

Retention Is the Real Tiebreaker

Cost per funded user is the right primary metric. Cost per retained user is what actually decides whether a channel was worth it, and crypto retention is worse than most teams assume.

CoinGecko's Q1 2026 retention study tracked wallets that completed at least five transactions in Q1 2025 and checked whether those same wallets were still transacting a year later. These aren't casual users. They'd already shown real activity.

Ethereum retains the highest share and BNB Chain retains the most people. Solana lost more than 16 million active wallets across the year while still retaining more users in absolute terms than Ethereum, because it started with far more.

Apply that to your acquisition math. If you're building on a chain with roughly 8% twelve-month retention, a $150 cost per funded user is really closer to $1,900 per user who is still there a year later. On a chain retaining 26%, the same $150 becomes about $575.

That's not an argument about which chain is better. It's an argument that cost per funded user is incomplete without a retention figure next to it, and that the retention figure varies by an order of magnitude depending on where you built.

Two things follow. Report a 90-day and a 12-month retained cohort alongside your CAC. And when comparing channels with similar funded CAC, the one with better retention is simply the cheaper channel, often by several times.

Where the Drop-Off Actually Happens

You now know what your real number is. The fastest way to improve it usually isn't buying more traffic. It's keeping the traffic you already paid for.

In order of typical impact:

• Verification. For exchanges this is the biggest single loss point. Tiered verification with limited access up front, a visible progress indicator, a realistic time estimate, and a mobile-friendly document upload all move this measurably.
• First deposit friction. Limited funding options, unclear minimums, and network confusion all cost you users who had already decided to join.
• Verified but not funded. These are your highest-value retargeting audience, and most teams ignore them. They've already done the hard part.
• Funded but not traded. A first-trade prompt, a small fee credit, or a guided first transaction converts a meaningful share.

Marketing owns the traffic. Product owns most of the drop-off. Reporting all three stages is what makes that conversation possible.

How to Actually Reduce It

Ranked by what tends to move the number most.

Fix conversion before buying more traffic. Moving signup-to-funded from 3% to 6% halves your cost per funded user without spending another dollar on media. No channel optimization competes with that.
‍

Kill your worst channel and move the money. Most teams discover, once they measure funded CAC properly, that one channel is subsidizing the reported average.

Pay KOLs against outcomes, not reach. A flat fee for a post buys impressions. A structure with a performance component buys funded users. A unique tracked link per creator makes this enforceable.

Separate incentive cohorts. You can't manage what's averaged into everything else.

Retarget the verified-but-not-funded group. The cheapest users you'll ever acquire are the ones who already signed up and stopped.

Reduce regulatory friction where you legally can. Clear geographic eligibility up front, tiered verification, and honest time estimates all recover users who would otherwise abandon.

‍Measure long enough to see retention. A channel with a higher funded CAC and much better 90-day retention is the cheaper channel.
‍

Our crypto marketing playbook covers channel selection, and our Web3 agency explainer covers what should be in place before any of this is measurable.

FAQ

What is a good CAC for a crypto project? There's no single answer, because it depends on what a user is worth to you and on which stage you're counting. As a reference point, eToro's 2025 filings imply roughly $573 to $628 per net new funded account, while Coinbase's S-1 showed $5.15 per verified user. Both are real. They measure different things.

Why do published crypto CAC benchmarks disagree so wildly? Two reasons. Most of them are unsourced estimates repeated between blogs. And where real numbers exist, they count different funnel stages. A signup-level figure and a funded-account figure can differ by more than a hundred times for the same business.

Should I include agency fees and token incentives in CAC? Yes. Media, KOL fees, agency fees, referral payouts, signup bonuses, token incentives at grant-day value, compliance review time and fraud write-offs all count. If you spent it to get the user, it belongs in the numerator.

Why is crypto CAC higher than software CAC? Harder onboarding, a large share of spend going to incentives rather than media, regulatory friction that adds steps between interest and funding, and cyclical attention prices that rise sharply in a bull market.

How do I attribute on-chain activity to a campaign? Unique tracked links per channel into a landing page, then connect wallet connect and deposit events back to that session. Pair it with a referral code as a backup and accept that some share will be unattributed. Consistency matters more than perfection.

Why is airdrop CAC so high if the tokens cost us nothing? Because the tokens do cost something, and because almost all those users leave. Value them at distribution-day price, then divide by users still active after 90 days.

Does retention really change the CAC picture that much? Yes. At the 7.9% twelve-month retention CoinGecko measured for one major chain, a $150 cost per funded user is closer to $1,900 per user still active a year later. At 26.2% it's around $575. Same acquisition cost, very different business.

Sources

Every external figure in this article, with its origin. We've marked which are primary documents and which are not.

Primary and named sources

• eToro Q4 and FY2025 results, 17 February 2026, and the same release filed with the SEC. Funded accounts, net contribution, and marketing as a share of net contribution. The per-account figures in this article are our own arithmetic on those disclosures.
• CoinGecko blockchain user retention analysis, Q1 2026. Twelve-month retention by chain, methodology stated.
• Coinbase Q1 2021 shareholder letter, filed with the SEC. Marketing spend and verified user disclosures.
• Business of Apps, eToro statistics. Independent compilation of eToro's reported user and revenue figures.

On the unsourced benchmark pages

We've deliberately not linked the pages discussed in the provenance section above. They're easy to find, and listing them here would present them as references when the whole point is that they're not.

Want Your Real Numbers?

We've worked with 250+ ecosystems, and the first thing we usually find is that nobody has separated signups from funded users.

Book a Meeting or Let's Talk Strategy. You explain what you're building, and we'll explain how we'd support it.