Retail Email Marketing ROI: The CEO Benchmark From 6 Billion Smart Banners

Ask a retail email team to defend its budget and you will hear the same statistic: email returns $36 for every $1 spent. It sounds unbeatable. It is also close to useless as a management tool, because it measures return on ESP subscription cost, the one line item in the program that never grows. To answer the question that ratio dodges, we measured 6.2 billion opens of Smart Banners and Smart Kickers across retail email programs and published the results as a benchmark. What does good look like, per impression and per click?

I will state my bias up front. I run an email personalization company, so I benefit when brands take email measurement seriously. But the numbers below are the ones I would demand if I sat on the other side of the table, funding both the ad team and the CRM team. The timing gives them teeth: average ecommerce ROAS fell to 2.87 in 2025, declining across 13 of 14 industries in Upcounting’s analysis, and customer acquisition costs are up 40 to 60% since 2023, with Shopify pegging the average merchant’s CAC at $318, up from $274 a year earlier. Every paid dollar buys less than it did two years ago. The owned channel is the cheapest growth lever most retailers have, if anyone bothers to measure it like a performance channel.

So here is the scoreboard: the three numbers that define good in retail email marketing, the 2026 Smart Banners benchmark figures to grade against, and the five questions I would ask my email team this quarter.

The $36 ROI story tells your CEO nothing

Credit where due: the $36-per-$1 figure comes from Litmus, and as a statement about channel efficiency relative to software cost, it holds up. The problem is the denominator. ESP fees are small and mostly fixed, so the ratio stays flattering no matter what happens to actual performance. A program could lose a third of its productivity and still clear $30 per dollar.

That is roughly what happened. Our analysis of industry send data puts batch email revenue per thousand sends down about 43% from its 2018 peak (roughly $227 to $130), and real revenue per subscriber down 35% over the same stretch, while send frequency climbed 63%. Brands compensated for depreciating subscriber attention by spending more of it. We unpacked that math in Email Personalization Stats That Actually Matter. Through all of it, the $36 ratio never moved, which is exactly why it should not be your scoreboard.

Finance teams have noticed the gap. Litmus’s own State of Email research found that 21% of marketing leaders do not know their email program’s actual ROI, and 63% now face increased CFO scrutiny of marketing spend, up from 52% the year before. The scrutiny is coming either way. Better to arrive with finance-grade numbers first.

The three numbers that define good in retail email

Paid media answers to revenue per thousand impressions and cost per acquisition. Email deserves the same treatment, and three metrics do the work:

• Normalized RPM. Attributed revenue per 1,000 impressions, normalized to a $100 average order value so results are comparable across categories and against published benchmarks. This is email’s answer to ROAS.
• Click-to-conversion (CTC). The share of clicks that become orders within a 7-day window. It measures whether content creates buying intent or just curiosity. Retail batch email converts about 2.5% of clicks. That is the baseline to beat.
• Annual revenue per subscriber. The long-run health metric. If it shrinks while list size and send volume grow, the program is strip-mining the asset.

Notice what is missing: opens and clicks. Both are inputs, not outcomes. We made the full case for CTC as the metric your CFO cares about in How to Measure ROI Email Performance.

The 2026 Smart Banners benchmark: 13.6% CTC against a 2.5% baseline

Smart Banners are conditional content blocks that bookend broadcast sends and decide what each subscriber sees at the moment of open: cart contents, a loyalty balance, a price drop on a browsed item, a shipping update. Because every variant renders and gets attributed individually, Smart Banners generate something campaign-level reporting structurally cannot: a per-creative read on what converts, at a scale of billions of impressions. (If the format is new to you, start with The Ultimate Guide to Smart Banners.)

Our Q4 2025 Smart Banner Benchmark Report covers 6.2 billion measured opens across more than 100 live variants, all normalized to $100 AOV. Three findings set the 2026 bar for retail:

• Personalized Smart Banner and Smart Kicker content averaged 13.6% click-to-conversion against the 2.5% retail batch baseline. More than five times the typical email.
• Every one of the 100+ variants cleared the roughly 2% CTC floor set by generic banner content. The worst personalized variant still beat the average batch send.
• Abandoned Cart variants averaged 18.7% CTC, the strongest scenario family in the dataset.

None of this should surprise anyone who has read McKinsey’s personalization research: 71% of consumers expect personalized interactions, 76% get frustrated when they do not get them, and faster-growing companies drive 40% more of their revenue from personalization than slower-growing peers. The surprise is how much retail email volume still carries none of it.

Inside the Smart Banners variant matrix: one trigger, a 15x spread

Here is the finding that should change how you read your team’s email reports. “Abandoned cart” is not one program. In the benchmark it is a family of 22 Smart Banner variants, and the spread between best and worst is roughly 15x on revenue per thousand impressions:

• Abandoned Cart alone: $135.30 normalized RPM.
• Cart + Loyalty + Price Drop: $469.65, a 3.5x premium for stacking two more signals on the same trigger.
• Cart + Coupon + Low in Stock: $300.99, at only 14.4% CTC. Urgency pulls more clicks, but less qualified ones.
• Cart + BNPL: $31.42 at 6.1% CTC. Financing messaging attracts price-sensitive, low-intent clickers.

Two lessons here. First, RPM and CTC diverge, and the divergence is diagnostic: the highest-click variants and the highest-revenue variants are rarely the same, which is why one blended number misleads. Second, a retailer reporting a single abandoned-cart line item cannot explain why its own revenue moves quarter to quarter. The signal lives in the variant table. The full breakdown is in Email Variant Testing: 22 Abandoned Cart Combinations.

Benchmark email against paid ads, not against last year’s email

Most email teams grade themselves against their own history. Wrong comparison. The budget email competes with is paid media, so grade it against paid media. In 2025 the average ecommerce ROAS was 2.87 and falling, with roughly half of ecommerce businesses operating below a 2:1 return, per Upcounting. Meta CPMs rose about 20% year over year, Google CPCs about 13%, and since Apple’s App Tracking Transparency changes, ad platforms can see only a fraction of the conversions they drive. Costs up, measurement down.

Now the caveat that makes the comparison lopsided: a dollar of email RPM is worth more than a dollar of paid-media RPM, because email carries no incremental media cost. Reaching a subscriber you already own costs fractions of a cent, so when Smart Banners put $135 to $470 per thousand impressions on cart traffic, that revenue lands at margins bought traffic cannot touch. A 3x topline advantage often works out to 5 to 10x on contribution margin. The full unit-economics argument is in Email Is a Performance Marketing Channel, and the Math Proves It.

One more irony worth sitting with: the ad industry already runs on your email list. Meta Custom Audiences, Google Customer Match, and every lookalike model are seeded from first-party email identity. Your ad team rents access to the audience your email team owns outright.

The proof standard finance signs off on, and five questions to ask this quarter

Vendor lift claims deserve skepticism, ours included. The best-known study in this category, Movable Ink’s commissioned Forrester Total Economic Impact analysis, reports a 422% ROI for a composite organization. Composites are directionally interesting, but they are not proof about your P&L. Per-email lift claims against historical baselines have a deeper flaw: under last-touch attribution, revenue shifts freely between triggered and batch sends, so impressive per-email lift is fully consistent with zero channel-level incrementality.

The standard finance will sign off on is a longitudinal channel-level holdout: randomize subscribers into exposed and control groups, measure UTM-scoped email revenue per subscriber, run about four weeks to reach 95%+ statistical significance, and re-run it whenever anyone doubts the number. That methodology is why we can put a 10x ROAS guarantee in writing (our customer average is 15x), and it is the proof you should demand from every vendor, us included.

The audit itself takes one meeting. Ask your email team:

1. What is our normalized RPM, and how has it trended over the past four quarters?
2. What share of our email clicks become orders within seven days, by content variant?
3. What is the RPM spread between our best and worst variant on the same trigger? (If the answer is “we run one variant,” that is the finding.)
4. When did we last run a randomized channel-level holdout, and what was the incremental revenue per subscriber?
5. What is our annual revenue per subscriber, and is it growing or shrinking?

Grade the answers against our Q4 2025 email performance benchmark report. And if the program needs a rebuild rather than a report card, the brand-first retail email playbook covers how teams roll Smart Banners across every send without adding production headcount.

Key takeaways

• The $36:$1 email ROI stat measures return on ESP software cost. It stayed flattering while real revenue per subscriber fell about 35% from its 2018 peak.
• Grade retail email on three numbers: normalized RPM, click-to-conversion, and annual revenue per subscriber.
• Across 6.2 billion opens, personalized Smart Banners and Smart Kickers averaged 13.6% CTC against a 2.5% retail batch baseline, and every variant cleared the 2% generic floor.
• One trigger is not one program: Abandoned Cart variants span $31.42 to $469.65 in normalized RPM. Without variant-level attribution, that 15x spread is invisible.
• The paid comparison line: average ecommerce ROAS is 2.87 and falling, and email revenue carries no incremental media cost, so topline parity understates email’s margin advantage.
• Accept one proof standard: a randomized, re-runnable, channel-level holdout. Per-email lift claims and composite ROI studies are not it.