Every DTC founder has felt it: the same channels that used to bring in a customer profitably now feel more expensive. That's real — acquisition costs have gotten harder to manage across most DTC categories since 2023, driven by auction saturation on Meta, more brands competing for the same inventory, and platform measurement that's gotten less reliable since iOS tracking changes. But the magnitude varies dramatically by category, geography, price point, and business model, and most of the public "CAC benchmark" content treats it as one universal number. It isn't. The more useful question is which levers actually move the number, because most of the obvious ones don't.
Before we go further: what do we actually mean by CAC?
E-commerce teams tend to calculate acquisition efficiency at several different levels, and treating them as interchangeable is where a lot of confusion about "rising CAC" starts. These aren't an official taxonomy — just the practical levels acquisition cost tends to get measured at:
- Platform CPA / reported CPA — the cost per purchase Meta, Google, or another platform reports inside its own dashboard. This isn't necessarily CAC: the purchase it's counting might belong to a returning customer, and each platform decides for itself, using its own attribution logic, which touchpoint gets the credit.
- New Customer CAC (nCAC) — relevant acquisition spend divided specifically by new customers acquired, which filters out the returning-customer noise sitting inside platform CPA.
- Blended CAC — total relevant acquisition spend across every channel, divided by total new customers acquired in the same period, regardless of which platform claims the credit.
- Fully loaded CAC — a broader financial view that also folds in acquisition-relevant costs platforms don't report: creative production, agency or team time, software, and any commissions or incentives spent to close the sale.
None of these is "the" CAC. A number is only meaningful once you know exactly what's sitting in the numerator and the denominator — worth checking before comparing your figure to anyone else's, including a benchmark table.
What's actually driving CAC higher in 2026?
Auction saturation and rising CPCs on Meta are real — more brands are competing for the same inventory as paid social matures as a channel, and platform measurement has gotten less precise since iOS tracking changes reduced visibility into user-level behavior. But part of what looks like "CAC went up" is also a measurement story, not a pure cost story: a brand that used to eyeball platform CPA and now also accounts for creative production, software, and the returning-customer contamination sitting inside that number will see a jump that has nothing to do with media getting more expensive — it's the same spend being measured more completely, at a different one of the four levels above than the one it was compared against last year. Before concluding media costs are the problem, it's worth checking which definition actually moved.
Is your reported CAC even accurate?
Separate what your analytics observes from what it decides to credit. Meta, Google, Klaviyo, and GA4 each apply their own attribution logic to the same customer journey and will routinely assign credit for the same purchase to different channels — none of them is wrong exactly, they're just running different models on the same data. That's a different question from what actually caused the purchase.
The distinction worth holding onto: attribution asks which touchpoint receives credit; incrementality asks whether the purchase would have happened without that marketing activity at all. Most platform dashboards only answer the first question, and it's easy to mistake a confident attribution number for a causal one. Attribution is directional — useful for spotting a large, consistent skew — until incrementality has actually been tested through a holdout, a geo test, or an in-platform lift study.
Comparing platform data against your backend (Shopify or equivalent), a GA4 view, and a simple post-purchase "how did you hear about us" survey is a way to triangulate — to see whether several imperfect sources roughly agree — not a way to arrive at one perfect number. Brands that "fix" CAC by cutting whichever channel one platform's attribution credits least often cut the wrong thing, because they've mistaken a single model's opinion for the causal answer.
How we think about CAC in practice
At SciGrowth, we don't usually try to assign a perfectly precise CAC to every individual touchpoint — the attribution-versus-incrementality problem above doesn't fully resolve no matter how much tooling gets thrown at it. Operationally, we more often monitor cost per purchase alongside the split between new and returning customers, because that split changes what a moving number actually means.
A falling cost per purchase is not automatically evidence that new customer acquisition improved. It can just as easily mean more existing customers are buying — a segment that's typically cheaper to convert than a stranger seeing the brand for the first time. Blend the two together and a genuinely worsening acquisition problem can hide behind an improving headline number. That's why we track new-versus-returning revenue and orders separately rather than trusting one blended trend line.
Take a realistic journey: someone discovers a brand through a Meta ad, signs up for email a week later, sees an organic Instagram post from a friend, opens a Klaviyo campaign, searches the brand name on Google a few days after that, and finally buys on a direct visit. We can observe pieces of that journey. We can't know with certainty which single interaction "caused" the eventual purchase — different systems will assign credit differently to the same sequence of events, and things like brand memory, organic content, and word of mouth often aren't measurable with any precision at all.
We treat attribution as directional rather than absolute, and focus on having enough measurement to make better allocation decisions — not on creating the illusion of perfect attribution. In practice that means following one chain: cost per purchase → new vs. returning customer mix → acquisition efficiency → repeat purchase → contribution margin. The objective is not to know exactly which channel deserves credit for every order. The objective is to have enough measurement to make better decisions about where to invest the next dollar.
What's the highest-leverage creative change?
UGC and creator-style content frequently outperforms polished, brand-produced creative in cold-audience acquisition placements — it reads as a recommendation in-feed rather than an ad, which lowers the psychological cost of the click. How much it outperforms varies widely by category, offer, and execution quality, so treating "make UGC" as a guaranteed percentage improvement is the wrong takeaway; that number should come from testing your own account, not a published benchmark. The more durable version of this lever isn't a single creative format — it's building a testing system: brief a batch of variants (mixing UGC-style and produced), run them at low spend across a range of hooks and angles, and let cost-per-acquisition data pick the winners rather than internal opinion. Brands that treat creative as a pipeline rather than a one-off project are the ones who keep finding new winners as old ones fatigue.
What lowers CAC without touching ad spend at all?
Referral programs, done properly — but "near-zero cost" oversells it. A referral program has real costs: the reward paid to the referrer, the incentive given to their friend, the software or platform fee, the margin given up on the discount, and some amount of cannibalization from customers who would have bought anyway. What referral programs reliably do is lower blended CAC relative to paid acquisition, and tend to produce a higher-quality customer — referred customers are frequently more loyal and more likely to refer someone else themselves, a compounding effect paid channels don't replicate, though the exact magnitude is worth measuring on your own list rather than assuming from a published figure. The common failure mode is launching with a generic discount and a footer link and never promoting it with the same attention a paid campaign would get — a referral program is a funnel, not a feature toggle.
Which channels can reduce blended CAC?
There's no universally "cheap" acquisition channel, and the benchmark tables that rank email against Meta against Google Shopping on a single dollar scale are comparing different things by different math. Google Shopping can be efficient because it captures existing search demand — higher intent, shorter path to purchase. Meta plays a different role: it's often more expensive at the first-purchase level, but it generates demand and scale that search can't — it can create the awareness that later shows up as a "cheap" branded search or email conversion somewhere else in the funnel. Email and SMS look extremely inexpensive on a last-touch basis, but frequently only because the cost of acquiring that subscriber in the first place sits somewhere else, uncounted.
Concretely: if a brand pays $3 in paid media to acquire an email subscriber, and 10% of subscribers eventually convert to a first purchase, the media component of that "cheap" email-driven CAC is already roughly $30 before software, creative, or discount costs get added. The useful question isn't "which channel has the lowest benchmark CAC" — it's which mix of channels produces the lowest blended CAC while sustaining enough volume and acceptable contribution margin. That's a channel-mix optimization, not a channel ranking.
What's the fastest technical fix for wasted spend?
Upload your current customer list to exclude from prospecting campaigns, and check it monthly as the list grows. Brands running cold acquisition without an up-to-date suppression list routinely waste spend re-targeting people who already bought — this is one of the few fixes on this list with essentially no downside and no real debate about whether it works, though the exact amount wasted varies by account and is worth checking rather than assuming. LTV-based lookalikes (built from your highest-value segment rather than all past purchasers) are worth testing on top of this, but treat any specific improvement number with caution — Meta's targeting has shifted heavily toward broad and Advantage+ delivery, where the platform's own signal-matching does more of the work than a hand-built lookalike audience used to. What used to be a reliable, quantifiable lever is now something to test on your own account rather than assume.
How do you know if a CAC number is actually a problem?
Not by comparing it to a generic benchmark — by checking whether your own unit economics can support it. LTV:CAC of at least 3:1 is a reasonable sanity-check heuristic at best, not a target to optimize toward: two brands can both report 3:1 and sit in completely different financial positions depending on when that value actually arrives and how much of it is margin versus revenue.
The more useful way to evaluate a CAC number is against the full chain it sits inside: average order value → gross/contribution margin → new-customer CAC → first-order contribution → CAC payback period → repeat purchase → 6- or 12-month LTV. A $100 CAC can be excellent for a subscription brand with strong margin and a customer who reorders monthly, and unsustainable for a one-time-purchase brand with thin margin and no repeat mechanism — the same number, two entirely different verdicts.
So the useful question isn't "is my CAC above the benchmark?" It's "can my unit economics support this CAC at the scale I actually want to grow to?" A rising CAC isn't automatically bad news if payback period and repeat rate are holding steady or improving alongside it; it only becomes a real problem once the full chain shows it — not just a headline ratio moving in the wrong direction.
None of this reduces to a single number worth screenshotting, and that's the actual point: CAC doesn't get optimized in a vacuum by copying a benchmark table or chasing platform CPA lower. It moves when the underlying system does — knowing which CAC definition you're actually looking at, triangulating measurement instead of trusting one platform's attribution, a real creative testing pipeline, a referral program that's actually promoted, a channel mix evaluated on blended cost, and unit economics tracked as a full chain from AOV to repeat purchase rather than one ratio. If you want a read on where your own acquisition system has the most room to move, a SciGrowth Free Marketing Audit looks at measurement accuracy, creative testing cadence, channel mix, and CAC payback against your actual numbers — not a generic industry average.
- What's the difference between platform CPA, new customer CAC, blended CAC, and fully loaded CAC?
- Platform CPA is the cost per purchase Meta or Google reports in its own dashboard, and it doesn't distinguish new customers from returning ones. New customer CAC narrows that to acquisition spend divided by new customers specifically. Blended CAC adds up spend and new customers across every channel, regardless of which platform claims credit. Fully loaded CAC goes further and folds in creative production, team or agency cost, software, and incentives. These aren't an official standard — they're different levels acquisition cost tends to get measured at, and a number is only meaningful once you know which one you're looking at.
- What's the difference between attribution and incrementality?
- Attribution asks which touchpoint gets credit for a purchase, based on whatever model a given platform uses. Incrementality asks whether that purchase would have happened anyway without the marketing activity in question. A channel can look highly effective by attribution while contributing close to nothing incrementally — retargeting someone who had already decided to buy is the classic example. Most platform dashboards only answer the attribution question; incrementality requires a deliberate test (a holdout group, a geo test, a lift study), not just a better dashboard.
- Is there a reliable industry-average CAC number to benchmark against?
- Be skeptical of any single number presented without stated methodology. Ecommerce performance data is highly skewed — a small number of extreme performers can distort a simple average into a misleading "typical" figure. The more credible benchmark approaches (like Triple Whale's peer-benchmarking, drawn from tens of thousands of accounts) match brands by industry, revenue band, AOV, and channel mix rather than reporting one blended average, and explicitly warn against comparing CPA across channels as if one is universally "better." Treat any generic benchmark as a directional sanity check, not a target.
- Why might my cost per purchase look like it's improving even though new customer acquisition isn't?
- Because a blended cost-per-purchase number doesn't distinguish new customers from returning ones. If a growing share of purchases in a given period comes from existing customers — typically cheaper to convert than someone seeing the brand for the first time — the blended number can fall even while new customer acquisition is flat or worsening. Tracking new and returning customer cost and volume separately is the only reliable way to catch this before it shows up as a bigger problem later.
- Does UGC actually work better than professional brand content for lowering CAC?
- It frequently outperforms in cold-audience acquisition placements specifically, because it reads as organic recommendation rather than advertising. How much it outperforms varies by category, offer, and execution — there's no universal percentage that holds across brands. The more reliable lever is running UGC and produced creative against each other in an ongoing testing pipeline rather than assuming either format wins by default.
- Is a referral program really close to free customer acquisition?
- No — it has real costs (referrer reward, friend incentive, software, discount margin, some cannibalization), and treating it as near-zero cost overstates the case. What it reliably does is lower blended CAC relative to paid acquisition and bring in higher-quality customers, which is a real and durable advantage even once the actual costs are accounted for.
Sources:
Triple Whale — What's a Good CPA? Benchmarks by Industry and Channel (42,000+ brands, trailing 12 months)
Triple Whale — Ads Benchmarks: The Missing Layer in Ecommerce Performance