Happy Sunday, Everyone!

Q1 is in the rear view mirror. 2026 is officially 25% over, conference season is in full swing, and most clients are wildly optimistic about growth despite the macro uncertainty of the last few weeks. That optimism is fine — it's useful, even — as long as it doesn't lead to the most common strategic mistake I see teams make during this particular season of reflection: mistaking diminishing returns for failure.

It happens everywhere, every Q2, like clockwork. A team looks at a campaign or a channel that previously performed at a higher level, concludes it's broken, and blows it up. They make massive pivots in strategy, rooted in their perceptions of past performance rather than the reality of it. They make dramatic structural changes that undo months of learning in search of a fix for something that wasn't actually broken.

So let's define terms. Failure is when something - a strategy, a channel, an offer - is no longer viable for its intended purpose. The inputs aren't producing outputs at any meaningful scale or efficiency. It's a structural problem that cannot be solved with patience or step-wise optimization. Something is fundamentally broken.

Most of what I see in actual ad accounts isn't that. It's diminishing returns. And those are something else entirely. Diminishing returns are what happens when a strategy works, so you do more of it. Your adoption of the tactic increases, so the surplus value (the outsized numbers on your dashboard/reports) decreases. This is classic S-curve behavior: as the slope flattens, your next dollar of investment produces comparatively lower output than the previous dollar. The underlying mechanism isn't broken. You've reached the natural ceiling of what the current inputs can produce.

Diminishing returns are not failures. Understanding the difference between the two is one of the highest-leverage cognitive upgrades a performance marketer can make.

A 200-Year-Old Idea Your Ad Account Is Living Right Now

In 1798, Thomas Robert Malthus published An Essay on the Principle of Population. His core argument: food production grows linearly, but population grows exponentially. At some point, output can't keep pace with input. The more you add, the less each addition yields. Malthus wasn't describing a flaw in the agricultural system - he was describing a law of nature.

Economists formalized his intuition into the law of diminishing marginal returns: the principle that every additional unit of input, holding all else equal, eventually produces less output than the unit before it. It showed up in 18th-century farming. It shows up in compute used to train AI models. It shows up in your Meta account every time you scale a campaign past its efficiency threshold. It’s all the same law at work.

To illustrate this in seasonally-appropriate terms: it's Easter weekend (for most - Orthodox Easter is next weekend). My daughters just finished collecting the first round of eggs (I'm writing this on Saturday) and are deep into the sugar high, thanks to chocolate eggs, chocolate bunnies, jelly beans, peanut butter cups - the works. The very first piece they ate was met with pure joy. The second was almost as exciting. But by Cadbury Egg #5, something had changed. The chocolate hadn't gone bad, but they had. Each additional piece yielded less delight than the one before it. That's diminishing returns in its most innocent form.

The same phenomenon happens in your ad account, but the shape of the curve varies in ways most teams never bother to examine. On Meta, at modest budgets, your ads reach the people most likely to convert — the algorithm's best guess at your ideal customer. As budget scales, you exhaust that pool and begin buying lower-probability impressions. Frequency climbs. Relevance drops. Effective CPM rises because you're now competing for inventory that used to go to advertisers whose audiences were a better fit. Same creative. Same offer. A fundamentally different audience composition that your blended dashboard will never show you.

But here's what matters and almost nobody measures: where the bend happens is not random. It's a function of creative concept depth, audience definition and offer-market fit. An account running 3 variations of the same concept (different thumbnail, same angle) will hit the inflection point faster than an account running three distinct concepts against the same audience. The former is refreshing the ad. The latter is refreshing the audience's relationship with the brand. Those aren't variations of the same strategy; they're different strategies with different diminishing returns curves. If you're not decomposing your diminishing returns by concept (not just creative), you're likely falling victim to Simpson’s paradox.

On Google, this is easiest to see in search: your high-intent keywords convert at a strong rate until you've captured essentially everyone actively searching for them. At that point, Smart Bidding is forced to explore other, lower-intent variations in order to use your budget. The result? Your CVR falls while CPA rises. In this case, Google isn't failing. To the contrary, it is doing precisely what it's designed to do: move down the curve until the aggregate value falls below your prescribed target. It’s “averaging in” suboptimal outcomes (the higher CPA ones) to your already-captured super-optimal outcomes (the customers acquired below your target CPA) in order to reach your designated scale target (defined by your budget). This is diminishing returns at work.

The Math That Actually Matters: Marginal vs. Average

This is where most marketers are working with the wrong numbers, and it's costing them (or, more likely, their clients).

Average performance is the number on your Meta or Google Ads dashboard: your blended CPA, your account-level ROAS, the aggregate view of how everything is running. It's a useful snapshot, but a terrible decision tool.

Marginal performance is the question that actually drives smart scaling: what does the next dollar buy me? Not the average dollar across my whole account or this campaign, but the next dollar I'm considering investing. That number, in a world of diminishing returns, is almost always worse than your average (though with learning phases, there are cases where it’s higher than your average at low volumes/low spends). And as you scale, it deteriorates faster than your blended metrics will show.

Think about it through the lens of poker: a good poker player doesn't manage their chip stack in aggregate. They think at the margin. The question is never "how am I doing overall?" The question is "should I call this specific bet, right now, with these cards, against this opponent, with this much left in my stack?" Those are completely different questions. Conflating them is how you go broke. A bad player feels fine because they're up overall. A good player is reviewing the broader trends, the state of the table and the expected value of the bet to assess if this bet makes sense or if they’re better off holding resources for when the squeeze hits.

And when that squeeze hits - when hand quality reverts to the mean and the losses start piling up - tilt enters the equation. I've written about tilt before (Issue #147 and Issue #159) in the context of Bayesian thinking and emotional decision-making: the moment a bad beat short-circuits your logic and you start playing feelings instead of odds. In poker, tilt turns a choppy, up-and-down session into a disaster. In paid media, it turns a performance plateau into an account rebuild that sets the account back months.

The marginal vs. average mistake is the analytical version of tilt. You're not reading the moment. You're reacting to the session average. And diminishing returns are one of the most reliable tilt triggers in the business. Performance deteriorates mildly, someone panics, an execute demands action be taken - so the agency/freelancer/marketing person blows up the account/tactic/strategy (and with it weeks or months of learning) and implements a "new strategy" ….that starts from square #0 and leads toward the same diminishing returns curve that caused the panic in the first place.

Let me share a real example from an account I reviewed earlier this year.

This brand was spending about $10,000/month on PPC and acquiring customers at a $50 CPA. Results each month were pretty consistent at 200 customers (plus or minus a few). The brand wanted to grow in advance of a fundraise. They established a new target of 300 customers/month. To accomplish that, they budgeted for $15,000 in total PPC spend, expecting the same $50 cost/customer. But those next 100 customers were not available at $50; they were closer to $95/each. Why? Because this brand was already capturing the proverbial lowest-hanging fruit available with their current $10,000/mo spend. The incremental customers they wanted required more resources to find + capture.

The result? Their account-level CPA increased ~30% to ~$65 while their number of new customers hit 300. PPC was still "profitable" in aggregate. But those marginal customers - the ones added by scaling - cost $95 each. For this business, the unit economics only supports a CAC up to ~$100 - which means they’re getting pretty close to value-destructive growth.

That large increase in CAC was the reason I was asked to review the account. They were considering abandoning Google because it “wasn’t working like it used to” - my response was that if they put their budget back to where it was, it would very likely perform as it used to. Their CAC was increasing because they hit diminishing returns, not because Google suddenly hated them or was broken.

Now here's the uncomfortable question: do you actually know your marginal CAC right now? Not your blended number. Your marginal number. Most brands have no answer to that question, because their reporting infrastructure doesn't support it. They know their blended CPA and have no idea what the last 20% of spend actually bought them. That's the gap. And until you close it, every scaling decision is a crapshoot.

This matters even more in 2026 than it did two years ago, for one simple reason: the point at which diminishing returns kick in is arriving earlier, at lower spend levels, across more accounts than I've seen in the last decade.

CAC is rising across most categories. Consumers are strained - inflation has been stubborn, interest rates are still high, housing costs are elevated and the ongoing conflict in Iran has kept energy (and gas) prices well above their usual levels. That cost pressure ripples through household budgets in ways that compress discretionary spending. When the consumer has (or even feels as if they have) less financial leeway, the bar for every sale you're trying to drive goes up. That doesn't change the shape of the diminishing returns curve; it shifts the entire curve upward. Your efficient zone gets smaller. The inflection point arrives sooner. The marginal customer who used to cost $95 at the $20k spend level now costs $115 - because you're not just competing for inventory, you're now competing for a consumer whose willingness to convert has been degraded by forces completely outside your ad account.

Meanwhile, platform consolidation and more advertisers flooding auction-based systems have pushed baseline CPMs higher. The diminishing returns threshold that used to kick in at $500,000/month is now hitting some accounts at $100,000.

The practical implication: you need a profitability floor, not just a CPA target. Know the maximum CAC your unit/service economics can support. Know what LTV looks like by cohort. Then work backwards: how much can you spend before marginal CAC crosses that floor? That number is your real budget ceiling.

Diagnosing diminishing returns correctly is half the battle. The other half is knowing which lever actually moves the curve and why it has that ability.

Every lever I'm about to describe works the same way: it changes one of the fixed inputs that define your current diminishing returns curve, which creates a new curve with a different ceiling.

This is the same thing that resulted in criticisms of Malthus: he didn't account for innovation. He didn’t think the curve could change. But then the Industrial Revolution happened, and suddenly productivity could expand in ways it couldn’t before - at least, for a while. That's exactly what you're doing when you pull these levers. You're not fixing a broken system. You're expanding what the system can produce.

#1: Introduce Genuinely New, Diverse Creative Concepts

This is the most underutilized lever in paid media.

When performance declines, most brands/agencies “refresh” the ad - a new image and/or new copy with the same underlying idea. That's not enough. Creative fatigue and market saturation look identical on a report/dashboard. The only way to tell them apart is to test fundamentally different concepts: a different problem framing, a different emotional entry point, UGC versus produced, long-form versus short, a new format entirely. A refreshed version of the same concept might buy you a week or two before you're back on the same curve. A genuinely new concept can reset the audience's relationship with your brand and/or unlock an entirely new audience, thereby creating a new curve entirely.

#2: Expand your Audience Stepwise

Once you've saturated your core audience, the move is to go adjacent: lookalikes built from your highest-LTV customers (not just purchasers - the ones who came back), interest and behavioral expansions that share the underlying need, adjacent demographics who haven't been exposed to your brand yet. The key is adhering to this non-negotiable principle: you're looking for audiences whose marginal CAC still clears your profitability floor.

#3: Change the Conversion Economics / Modify The Value Exchange

If your marginal customer costs more to acquire, the question worth asking is: can you change the initial ask? A lower-friction entry offer - a sample, a trial, a smaller SKU, a lead magnet, a free consult - may shift the conversion economics enough to bring marginal CAC back into range while still growing your pipeline/funnel. You're not abandoning your core offer. You're building an on-ramp for customers who are a little harder to reach at the current price of admission. This lever is underrated because it feels like a concession, but it's not. It’s just a structural shift that moves the denominator.

#4: Access New Auction Environments.

If you've hit the ceiling on Meta and Google, there's an entire tier of platforms that most brands underinvest in. But the reason this works as a diminishing-returns lever isn't just "more places to spend." It's that each platform has its own auction, its own audience pool, its own inventory supply. Spend on X or YouTube or CTV isn't competing with your existing Meta campaigns; you're accessing demand that wasn't available to you before, which means you're expanding your efficient frontier rather than pushing further up your existing curve. That's not "diversifying spend” - it’s accessing a different supply.

#5: Build Your Top of Funnel

Upper-funnel channels - CTV, YouTube, display - have long been known to have a “halo” effect that warms audiences before they enter your conversion campaigns. That pre-warming increases the pool of high-intent prospects available to your lower-funnel campaigns, which means your diminishing returns curve in the lower funnel starts later and inflects slower. More of the inventory your conversion campaigns are bidding on is actually good inventory, because the consumer has been primed. I'm not making a branding argument; I'm making a math argument about how you expand the efficient frontier of your acquisition curve by investing upstream.

#6: Make Each Acquired Customer Worth More.

If marginal CAC is rising and you can't bring it down, change the other side of the equation. Email, SMS, loyalty, subscription models, post-purchase sequences - retention investment changes the unit economics of the entire system. A customer with a 30% higher LTV may be able to support a ~20% higher CAC (accounting for risk and volatility). That doesn't fix the acquisition cost, but it can change whether that cost is actually a financial issue. In an environment where consumers are stretched by inflation and elevated energy costs, the brands that win aren't always the ones who acquire the most efficiently. They're often the ones who keep customers the longest.

The Takeaway

Q1 is in the books. Some of it worked. Some of it worked less well than it used to. Some campaigns that were phenomenal performers in Q4 are producing less impressive results today. In the fact of that, it’s tempting to pull the plug or pivot - but before you do, ask the question that matters: is this failure, or is this diminishing returns?

Is the system broken, or has it hit a natural ceiling at the current level of inputs? Those are two completely different problems with two completely different solutions. And conflating them - treating a ceiling as a collapse - is one of the most expensive mistakes you can make.

Malthus made the mistake of thinking the ceiling was fixed. He was wrong. The question heading into Q2 isn't whether you've hit a ceiling. It's whether you have a plan to move it.

Cheers,

Sam

P.S. Whenever you're ready, here's how I can help: if your brand needs a strategic partner that blends performance marketing, analytics, and brand into one integrated team — not five siloed agencies — just reply to this email.

Loving The Digital Download?
Share this Newsletter with a friend by visiting my public feed.

Follow Me on my Socials

​​​

1700 South Road, Baltimore, MD 21209 | 410-367-2700
​Unsubscribe | Manage Preferences​