Atharv Ketkar

CAC and its fat tail

Snippet from previous article

Back in the day, I was into trading. Mostly derivatives. I used to buy and sell options, but buying was particularly enticing. Mostly because it had high potential for profit, but that potential came from the risks that came because of time decay and delta. To make any meaningful profit, you would have to have both time decay and delta on your side. But when it did, it would multiply whatever you spent.

Time decay is a really uniform thing, not much going on there. But delta is where the fact tail risks reside. Also, its other Greek brothers, such as Gamma. Delta means change in price of the derivative with respect to change in the underlying, and gamma is the rate at which that change will be implemented. A sudden boost in gamma basically multiplies whatever changes are happening in the underlying. And because of these two brothers, Vega also comes into the picture, which basically talks about how much implied volatility will be changed. It is mostly coming from this delta and gamma fluctuations. Basically, you make profit because time didn’t go against you. Delta worked out in your favor. That means Gamma and Vega also contributed.

So basically means that whenever you buy options, if this thing works out in your favor, you have huge potential for profit. It’s like a lot of tiny things add up and it explodes. This is the tail risk for option sellers because time decay and range-bound price movement work in favor of you. Usually you just end up making uniform profits, but all of sudden when the things go against you, when delta, gamma, vega all conspire against you in a really short amount of time, the losses just explode. Whatever premium you could have earned will be multiple times smaller than whatever you lost.

Most of the big institutions you will see are into option selling business. This is why black swan events happen – that all of sudden conditions just manufacture a fat tail risk where delta, gamma, and vega just explode. And that causes their losses to spike all of sudden. This is the tail risk. That things will be under control most of the time but one time it goes haywire, it really, really goes haywire.

Now that we have set context for what Fat tail risk is, I want to explore how it relates to CAC (Customer Acquisition Cost).

Custom acquisition costs are usually stable. You will have a ROAS of a reasonable range. But that ROS is built on top of a lot of things. In a small-tailed world, CAC would behave nicely:
– small fluctuations
– averages would be meaningful
– yesterday would predict tomorrow

But CAC lives in a fat-tailed distribution, which means a few extreme events dominate outcomes.

Most of the time, CAC looks reasonable. Then:
– one channel saturates
– CPMs double overnight
– an algorithm changes silently
– creatives fatigue all at once
– attribution breaks while spend keeps flowing

There are a lot of first-hand stories where people talk about how their CAC just exploded 2-3x of what it usually was. And because of that, their unit economics just don’t make sense.

It’s like small risks keep adding up, adding up, and it just explodes at once. That’s why there are rules of thumb like you should always have the average order value as at least five times your CAC. So that even with these fluctuations, you have a margin of safety. In options also, we see that there are hedging strategies so that you don’t do naked short selling. There are strategies like straddles and strangles and many others.

The point here is that CAC can explode, and you need hedges. In a business cycle, there are 5 steps.

  1. Spending money to find work
  2. Spending money to get that work done
  3. Spending money to deliver that work
  4. Getting paid for the work done
  5. Reinvesting into step 1

Only when you are able to get to step 5 and get it done, you have a recurring cycle, and you will not be able to do the step 5 if CAC is too high for whatever you got paid or whatever capital you have. And thus the business cycle stops, and you don’t have a business anymore.

And if there are fatal risks baked into this step 5, that means how careful you have to be when you are designing the system. You have to make it anti-fragile. You have to make it be able to withstand such pressures. Treating CAC as a stable average is dangerous.

How do you think Nassim Taleb would create a system to handle CAC? His whole thing is anti-fragility and black swan events. I’ll just ask Claude and paste the reply.

Well, I don’t agree with the whole thing, but it’s quite good. Here is the reply by Claude.

Taleb’s Core Principles Applied to CAC

1. Barbell Strategy – The Foundation

Taleb would absolutely structure CAC exposure as a barbell:

Left side (90% of budget): Extremely safe, boring, anti-fragile channels

  • Organic content that compounds over time (SEO, YouTube library, email list building)
  • Word-of-mouth systems with negative CAC (referral programs that pay for themselves)
  • Owned media that can’t be turned off by algorithms
  • High-repeat-purchase products where CAC is paid once, LTV accrues forever

Right side (10% of budget): Crazy, high-variance bets

  • Test wild new channels aggressively
  • Experimental creative that might 10x performance or fail completely
  • Guerrilla marketing stunts
  • Influencer lottery tickets

Middle (0% of budget): NOTHING

  • No “moderately optimized Facebook ads”
  • No comfortable reliance on any single paid channel
  • Avoid the middle precisely because it’s where you get fooled by randomness into thinking you have stable CAC

The left side survives the CAC explosion. The right side might be what gets you acquired or breaks out. The middle is where you slowly boil to death.

2. Via Negativa – Subtraction Over Addition

Taleb would focus on removing CAC fragility rather than optimizing CAC:

What NOT to do:

  • Don’t build a team sized for your best CAC month
  • Don’t commit to fixed costs based on ad channel performance
  • Don’t take VC money that requires you to maintain unsustainable growth rates
  • Don’t use attribution platforms as truth
  • Don’t have vendors/tools you can’t cut within 24 hours

The practice: Every month, identify one dependency that would kill you if CAC spiked, and eliminate it.

3. Antifragility – Gain from Disorder

This is where it gets interesting. Taleb wouldn’t just try to survive CAC spikes – he’d build a system that profits from them:

When CAC explodes for everyone:

  • Competitors with thin margins die
  • Customers become available from defunct competitors
  • Talent from failed DTC brands becomes cheap
  • Partnership opportunities emerge (desperate brands will do rev-share deals)

How to position for this:

  • Maintain 12-24 months runway at ZERO revenue
  • Have a list of acquisitions/hires you’d make if CAC chaos creates opportunity
  • Build brand moat during cheap CAC periods so you can survive on organic when CAC spikes
  • Create content/systems that get discovered when people search for your dead competitors

Example: When iOS 14.5 killed Facebook attribution, the antifragile move was to acquire customer lists from dying DTC brands for pennies and email them your offer.

4. Lindy Effect – Time-Tested Over Novel

Taleb would heavily bias toward channels that have survived multiple CAC regime changes:

Lindy channels (prefer these):

  • Word of mouth (worked for 10,000 years)
  • Content that ranks in search (survived every algorithm change)
  • Email (keeps working despite “email is dead” for 20 years)
  • Strategic partnerships/B2B deals
  • PR/earned media
  • Community building

Anti-Lindy channels (use only for the 10% barbell side):

  • Whatever platform launched in the last 2 years
  • Arbitrage opportunities (they close)
  • Algorithmic feeds (they change)

The older the channel, the longer it’s likely to survive. Facebook ads are ~15 years old, so probably have another cycle in them. TikTok ads are ~4 years old – could disappear tomorrow.

5. Optionality – Maximum Reversible Decisions

Every CAC decision should be reversible within 72 hours:

  • Monthly contracts only (never annual commits for “discounts”)
  • Creative that can be produced in-house on 48hr notice
  • No specialized team members who only know one channel
  • Tech stack that doesn’t lock you into platforms
  • Financial runway that lets you go dark on ads for 6 months

Taleb’s test: “If this channel disappeared tomorrow, would we be out of business?” If yes, you’re fragile. Fix it.

6. The Ludic Fallacy – Don’t Trust Your CAC Data

Taleb would say your spreadsheet showing “$32 CAC” is manufactured certainty in an uncertain world.

His approach:

  • Assume your attribution is 40-60% wrong
  • Assume CAC will 3x with no warning
  • Assume your “best” channel will die
  • Build scenarios, not forecasts

Instead of: “Our CAC is $32, so we can spend $X”

Taleb would say: “We don’t know our real CAC. It could be $20 or $60. We’ll spend as if it’s $60, and if we’re wrong, we’ll be pleasantly surprised.”

7. Mediocristan vs. Extremistan

Taleb would classify CAC as Extremistan – where a single event can dominate all others.

In Mediocristan (height, weight), averages matter. In Extremistan (wealth, book sales, CAC), averages are meaningless.

Practical implication:

  • Don’t plan based on average CAC
  • Plan based on maximum survivable CAC
  • One month at 10x CAC can wipe out 12 months of “average” CAC

The Taleb CAC System (Concrete Playbook)

Month 1-6: Build the barbell

  • 90% of effort: Create owned assets (content, email, community, product excellence for WOM)
  • 10% of budget: Wild tests on new channels
  • 0% reliance on any single paid channel

Ongoing: Negative CAC hunting

  • Obsessively seek channels where you get paid to acquire customers
  • Referral programs, affiliate networks, rev-share partnerships
  • B2B deals where someone else pays your CAC

Financial structure:

  • Raise/save enough to survive 18 months at $0 revenue
  • Never spend more than 20% of that runway on paid acquisition in any quarter
  • Size team for minimum CAC scenario, not median

Decision rules (automated):

  • If CAC on any channel exceeds 2x rolling 90-day average for 7 days → cut spend 80%
  • If total CAC exceeds 20% of price → stop all paid acquisition
  • If runway drops below 12 months → immediate austerity mode

Offensive optionality:

  • Maintain acquisition target list (brands, talent, customer lists)
  • When CAC chaos hits market → deploy saved capital to acquire distressed assets
  • Your competitors’ CAC explosion is your M&A opportunity

What Taleb Would NOT Do

  • Optimize click-through rates or conversion rates (local optimization increases fragility)
  • Believe any attribution platform
  • Celebrate low CAC (it’s probably hiding risks)
  • Take VC money that requires consistent growth
  • Hire a performance marketing team larger than 2 people
  • Use debt to fund customer acquisition
  • Trust any consultant who shows you a “formula” for CAC

The Philosophical Core

Taleb would say: “You don’t optimize CAC. You survive it, then exploit others who tried to optimize it and died.”

Your competitive advantage isn’t finding cheaper CAC. It’s being the last brand standing when everyone else’s CAC simultaneously explodes.

Build for that scenario, and everything else is upside.

What’s missing that Taleb would likely add:

  • Skin in the game – Taleb would want whoever is managing CAC to have personal downside. If your performance marketing team gets bonuses for growth but doesn’t lose their jobs when CAC explodes, you have misaligned incentives.
  • Via negativa for product – The CAC discussion assumes you keep selling the same thing. But Taleb might say: what if CAC explosion is the market telling you the product isn’t actually that good? Sometimes the answer isn’t hedging CAC but killing the product.
  • Convexity in positioning – Beyond just surviving CAC spikes, can you position so that some % of your customers actually become more valuable during CAC chaos? (E.g., if you serve other marketers, when their CAC explodes, they need your solution more.)

Practical addition – The “CAC Stress Test”:

Run monthly scenarios:

  • If your top CAC channel disappeared tomorrow, could you hit 60% of revenue targets through remaining channels?
  • If CAC tripled across all paid channels for 6 months, would you survive?
  • If attribution broke completely (iOS 18, anyone?), could you still make acquisition decisions?

If any answer is “no,” you’re fragile.

The deepest insight here:

Your competitive moat isn’t efficiency, it’s robustness. The company that can survive 24 months of terrible CAC beats the company optimized for 6 months of great CAC. Because those 24 terrible months will come – and when they do, you inherit your competitors’ customers.

Damn thats good, not bad at all. Turned me pro seo lol

So playbook can be:

  1. Find some good keywords.
  2. Write really good SEO content. With that, you’ll start getting some organic reach.
  3. Use that to create real useful organic video content and rank on platforms like Reels and YouTube.
  4. You can either already have a product or based on the audience you cultivated, launch a product.

Because if you are a good writer, you can do this ad infinitum. You’re writing it yourself. It’s just basically free for you. You can keep working in your spare time.