Product Recommendation Emails: The Retail Revenue Playbook for Your Highest-Earning Email Block

If you’re in retail email, here’s a fact you can’t afford to ignore: product recommendation emails generate more attributed revenue than any other personalized block in your campaigns. Not banners. Not kickers. Not even hero units. The product grid, the unassuming block of recommended items tucked below your editorial content, is quietly outperforming everything else. And yet, most retailers ship this revenue engine in the lowest possible form: send-time HTML cards with system fonts, raw product images, and frozen logic. That’s like racing a Ferrari in first gear.

Meanwhile, the performance marketing world is tightening. Average ecommerce ROAS fell to 2.87 in 2025 across 13 of 14 industries, with the median hovering around 2.04, meaning half of all ecommerce brands are barely breaking even on paid ads (Upcounting / Triple Whale). Customer acquisition costs have jumped 40-60% since 2023. Meta CPMs are up 20% year over year, Google CPCs up 12.88%. And thanks to iOS ATT, ad platforms can only see 40-60% of conversions. In this environment, every dollar must earn its keep. That’s why product recommendation emails aren’t just an email tactic, they’re a performance marketing lever waiting to be measured and optimized like one.

The real question isn’t whether product recommendations work. It’s why the highest-earning block in your email program is running at its lowest possible quality setting.

The Revenue Case for Product Recommendation Emails

Let’s start with the data. Across Zembula’s platform, which measures billions of email opens and personalized content blocks, the product recommendation grid consistently generates the largest share of attributed email revenue. This isn’t a one-off. It’s not dependent on a single client or vertical. It’s a structural advantage: the post-hero, mid-email position gives recipients time to engage with your content before seeing a curated set of products. And because these grids are personalized, based on browsing, cart behavior, purchase history, or cohort trends, they’re more relevant than generic promotions.

But relevance alone isn’t the full story. The real performance gap opens at the render layer and the decision moment. Most brands use HTML-based product cards. These are built at send time, using system fonts and raw product images pulled directly from PDPs or DAMs. They don’t adapt to brand styling. They’re prone to Gmail’s 102KB clipping limit. And because they’re static, they can’t reflect behavior that happens after the email is sent. The result? A missed opportunity on two fronts: brand fidelity and real-time relevance.

When you render the same recommendation logic as a composed, on-brand image, what we call a Product Grid or Product Highlights block, the conversion lift is 5-7%. That’s not a rounding error. That’s pure margin. And it’s not from better algorithms. It’s from better presentation and timing. For a brand doing $10M in annual email revenue, that’s $500K to $700K in upside, just from switching from HTML to image-based rendering.

For a deeper dive into why this happens, see our breakdown of why your product recommendation email doesn’t look like your brand (and how to fix it).

Five Product Recommendation Email Strategies That Work

Not all recommendations are created equal. The best-performing product recommendation emails use intent signals to serve up the right products at the right time. Here are five proven strategies:

• Bought Together: If someone buys a dress, what else did similar customers add to cart? This is social proof baked into logic. It works especially well for accessories and add-ons.
• Best Sellers & Trending: For broadcast sends with no behavioral data, this is your fallback. But don’t just pull top sellers, use velocity to surface what’s spiking. A product that jumped 300% in sales this week carries more urgency than a perennial bestseller.
• Browsed Together: If a user viewed a laptop, what else did they look at? This captures cross-category intent and works well for consideration-phase shoppers.
• Cart Favorites: For users who abandoned a cart, show what others bought with the abandoned item. It’s a subtle nudge that reduces decision fatigue.
• Last-Order Personalization: For broadcast sends, personalize the grid based on the recipient’s most recent purchase. If they bought running shoes, recommend socks, insoles, or a fitness tracker. This turns one-time buyers into repeat customers.

Each of these can be automated using Zembula’s Product Recommendations engine, which supports all five logic types out of the box. And because the logic is decoupled from the render, you can test different strategies without touching your design system.

For a full breakdown of how these strategies perform, see why the product grid is your highest-revenue email block.

From HTML to Image: The 5-7% Lift

Let’s be clear: the recommendation algorithm matters. But our platform data shows the biggest gains aren’t from tweaking model weights, they’re from upgrading the delivery format.

HTML product cards have three fatal flaws:

1. Brand mismatch: System fonts and raw PDP images clash with your email’s design. This breaks immersion and reduces trust.
2. Clipping risk: Gmail cuts off content after 102KB. Complex HTML grids often exceed this, especially on mobile.
3. Send-time freeze: The recommendations are locked in when the email is sent. If the user adds a new item to cart, or if inventory changes, the email can’t adapt.

Switch to image-based rendering, and you solve all three. The Product Grid block uses precomputed catalog math, but the final selection is decided at open time. The image is composed on the fly with your brand fonts, art-directed imagery, and real-time inventory checks. It’s a single image tag, no clipping, no rendering issues.

And the performance speaks for itself: 5-7% higher conversion rates. This isn’t theoretical. It’s measured across multiple brands using block-level RPM and CTC attribution with a 7-day click window. For more on the technical how, read our full analysis.

Decide at Open, Not at Send

Timing is everything. A recommendation that made sense when you hit “send” may be irrelevant a week later. Prices change. Inventory runs out. Users browse new categories. Yet most email personalization is locked in at send time.

Here’s the reality: more than 10% of email revenue comes from opens that happen more than a week after send. For broadcast campaigns, especially daily or weekly digests, this is a massive blind spot. Send-time recommendations go stale. They can’t reflect a user’s new cart, a price drop on a browsed item, or an out-of-stock alert.

With open-time rendering, the logic is precomputed, but the final product selection happens when the email is opened. This is made possible by Zembula’s tiered refresh system: the catalog math runs ahead of time, so the decision at open is fast and reliable. The result? Recommendations stay relevant. Revenue that would have leaked out is captured.

For a technical deep dive, see Moment-of-Send vs. Moment-of-Open: What Real-Time Email Really Means.

Measure Like a Performance Marketer

Email teams have long relied on CTR and open rates. But if you want to compete for budget with paid ads, you need better metrics. Finance teams don’t care about clicks. They care about revenue and return.

That’s why we measure product recommendation emails using two key KPIs:

• Block-level RPM (Revenue Per Thousand Impressions): How much revenue did this block generate per 1,000 views?
• CTC (Click-to-Conversion rate): Of the people who clicked, how many actually bought?

Together, these tell you not just whether a block got attention, but whether it drove results. And because we support longitudinal holdout and equal-size content control testing at the module level, you can validate lifts with statistical rigor.

For example: in a recent test, a brand ran two versions of their product grid. Version A used HTML cards. Version B used image-based rendering with open-time logic. The result? 6.3% higher RPM and 5.8% higher CTC for the image version. The test used a 10% holdout group and ran for six weeks, with 99.9% significance.

To see how these metrics compare across the industry, download our 2025 email performance benchmark report.

The Performance Marketing Case for Email

Let’s connect the dots. Paid ads are getting more expensive and less measurable. Email is owned, privacy-durable, and built on first-party data. Yet email teams are still using 1990s-era metrics to justify their budget.

The shift is already happening. 71% of publishers now cite first-party data as a key source of positive ad results. And ad platforms themselves, Meta, Google, are built on email-derived audiences. Lookalike modeling, Custom Audiences, Customer Match, all of it starts with your CRM list.

So why not redirect a slice of that ad budget to the channel that fuels the ads? Product recommendation emails are not just a retention tactic. They’re a performance channel. And when you measure them with RPM, CTC, and holdout testing, you speak the same language as the CMO.

For a full breakdown of the economics, see Email Is a Performance Marketing Channel, and the Math Proves It.

Key Takeaways

• The product recommendation email block generates more attributed revenue than any other personalized email component.
• Switching from HTML cards to image-based rendering can lift conversions by 5-7% due to better branding and reliability.
• Open-time rendering captures 10%+ of email revenue from delayed opens, which send-time logic misses.
• Measure performance with block-level RPM and CTC, not just CTR.
• Use longitudinal holdout testing to validate lifts with statistical confidence.
• Email is a performance marketing channel, use the same metrics, rigor, and budget logic as paid ads.