The Power of Bloomreach’s Unique Data Architecture

Divya Mulanjur
Divya Mulanjur
Bloomreach's architecture and in-memory framework enabling real-time personalization

In ecommerce, the brands that win are the ones that make the right call at the exact moment a customer interacts with them, whether that’s on the product page, in the inbox, in the search bar, or through an ad. 

And, with advances in agentic AI, customer expectations have shifted to keep pace — now, a relevant message that arrives an hour late is no longer relevant. 

This is why real-time marketing matters. That speed is what makes it possible to react to what your customer’s doing in the moment, using everything you know about a customer and what they’re currently doing. That’s what we call intelligence at the edge. 

Loomi, Bloomreach’s AI for customer engagement, acts on this real-time intelligence, helping businesses personalize the way they engage with their customers across email, search, web, and more. The architecture underneath it is what makes that possible.

What Is an In-Memory Framework (IMF)?

An in-memory framework (IMF), which includes an out-of-the-box in-memory database, allows users to have responses and data ingestion within milliseconds. It stores and processes data in a computer’s memory rather than on disk or external storage, so reading and writing it is far faster.

With Bloomreach, customer data lives in the IMF. Everything needed for a decision is available at split-second read/write speeds, with real-time reporting and execution. Data goes from ingestion to activation in 5ms–2s, depending on the tracking type and instance.

Speed of Bloomreach's in-memory framework

That window is what turns speed into real-time intelligence. When a shopper clicks, buys, unsubscribes, or abandons a cart, that event is part of their profile before the next decision about them is made. The next email, ad audience, recommendation, or search result reflects what just happened, not what happened yesterday.

An in-memory framework also eliminates the need to ingest data in a predefined schema, which enables a flexible no-schema data structure or the ability to track specific types of non-standard interactions. In practice, that means you don’t need perfectly clean, fully modeled data before you start. That belief came up in roughly one in four of our sales conversations over the past six months, and it’s the single most common reason we see data projects stall before they begin.

What this ultimately means is that real-time marketing is actually “real time” instead of “in a little bit.”

The speed and customer understanding enabled by IMF is the difference between meeting the expectations of your customers in their moment of need and disappointing them with subpar, irrelevant experiences.

Built To Work With Your Data Stack 

Today’s data teams aren’t choosing between a marketing database and nothing. Many have invested in a warehouse or lakehouse such as Databricks, and they’re weighing warehouse-native and zero-copy tools that activate data where it already lives. That’s the right foundation, but it’s also only half the picture.

Think of it this way: 

  • You have a core set of data (aka your warehouse). It’s the system of record for who the customer is: lifetime value, purchase history, propensity models, margin, etc. It’s deep, governed, and built for analysis at scale.
  • Then, you have the edge — this is where the customer is right now (e.g., the click 10 seconds ago, the cart they just abandoned, etc.). Decisions here have to be made in milliseconds, at the exact moment of interaction. 

Bloomreach’s IMF enables us to act at the edge. It takes what the core knows about a customer and combines it with what the customer just did, then decides what to show, send, or suppress. The warehouse stays the record of truth, but Loomi makes that truth actionable in the moment. 

In this way, we complement your warehouse instead of replacing it. Your data team keeps the stack they’ve built, and your marketers get decisions that use it at the speed customers expect. 

How Bloomreach’s IMF Gives Marketers an Advantage

Some marketing decisions are only right for a few seconds. Make them with data that’s an hour old (or even a few minutes old), and you get them wrong. These are the moments where intelligence at the edge earns its keep.

Stop Retargeting the Moment Someone Purchases

Let’s say a customer abandons their cart. An email gets sent, and two hours later, a paid retargeting campaign kicks in. Meanwhile, the customer opens the email, clicks, and buys. If their profile isn’t updated in real time, then you keep paying to convert someone who already converted. With intelligence at the edge, Loomi knows to remove them from the ad audience immediately after they purchase. 

Respect a Recent Unsubscribe

The last thing you want is to send an SMs to someone who opted out five minutes ago. Consent is a decision that changes in an instant, and the stakes are especially high in regulated industries like gaming and fintech. Because consent data is available in milliseconds, every send checks the customer’s current status, not last night’s.

Change the Segment While the Customer’s Still Browsing

A visitor may land on your site as a casual browser, but three clicks later, it’s clear they’re shopping for a gift. IMF enables the brand to update the segment mid-session, so site search, merchandising, and on-site content shift to match the shopper’s behavior before they leave. A system that uses batch processing, on the other hand, wouldn’t recognize the change until the next day. 

Act While the Shopper’s Still in the Mood To Buy

When a customer adds an item to their cart and keeps browsing, Loomi knows in real time that they’re actively shopping. That lets you:

  • Retarget them automatically with paid ads on other sites
  • Send an abandoned browse email or SMS with personalized recommendations as soon as they leave the site
  • Send an abandoned cart email with a personalized coupon a few days later
  • Personalize the hero image and other content to match their preferences when they return

Each of these is a decision with a short shelf life. Intelligence at the edge makes it while it’s still the right one.

Get Clearer Data Analysis and Visualization

Bloomreach’s unique architecture unlocks the ability to conduct exploratory data analysis — an invaluable opportunity for marketers to see important trends and patterns in customer behaviors without having to rely on data scientists for everything.

Using our built-in data visualizations, marketers can:

  • Understand customer behavior. For instance, you can easily answer the question: “How many customers purchased in May 2025 vs. May 2026?”
  • Compare the lifetime value and purchase frequency of a customer whose first product was for an adult versus for a kid, or a customer acquired during Black Friday vs. Cyber Monday or in the fall vs. the spring.
  • Perform a drill-down analysis of male vs. female buyers within a certain age bracket who opened a campaign type, clicked on an ad, and so on.

Keeping Up During Peak Season

The biggest test of “real time” is peak season: Black Friday, Cyber Monday, holiday travel, product launches, etc. With many marketing solutions, the advice is to prebuild segments, reports, and triggers ahead of peak, then change nothing during it to avoid lag or breakage.

But “near real time” that can’t hold up when it matters most isn’t much use. Because the core of Loomi’s campaigns and analytics runs in RAM, it stays split-second fast during peak too, and edge decisions keep working when volumes are highest.

Marketer making real-time campaign adjustments with the power of Bloomreach's in-memory framework

In-Memory Framework in Action

Here’s how Bloomreach customers see the difference in practice.

Benefit Cosmetics Boosts Omnichannel Campaign Success

Benefit Cosmetics successfully pulled off an omnichannel product launch campaign by combining email, lead gen, and onsite banners. Thanks to our in-memory framework and real-time segmentation, the brand was able to drive 40% more revenue.

Email, in particular, played a central role in moving the customer forward in their journey after acquisition in other channels. Benefit used granular segmentation (previous buyers, VIPs, waitlisted folks, etc.), which directly resulted in a 10% CTR for its emails.

Crucially, marketers at Benefit have been able to:

  • Understand the customer journey through Loomi and use those insights to perform the next action
  • Report on the end-to-end success of their launch across channels using Loomi custom reports without the need to rely on complex BI tools or downloading to Excel
  • Attribute leads and orders to the right channel thanks to an all-in-one platform, which helps them make an informed decision on where to invest for their next launch’s success

This level of data analysis is only made possible because of Bloomreach’s unique data architecture.

Arena Turns Real-Time Insights Into ROAS Lift 

Arena needed a more efficient way to target its marketing efforts, but was struggling with out-of-sync channels and one-size-fits-all campaigns. 

With the help of Loomi, Arena was able to combine everything into one channel and connect real-time customer profiles to its ad campaigns. This also allowed the brand to deploy behavior-triggered weblayers on its site to engage high-intent shoppers. 

Thanks to this unified, real-time approach, Arena improved its return on ad spend (ROAS) by 71% and drove a 2x higher click-through and a 3.5x higher purchase rate through its weblayers strategy. 

Yves Rocher Upgrades Personalization With Real-Time Product Recommendations

Yves Rocher wanted to go beyond generic recommendations, even with first-time visitors to its site. To accomplish this, the brand turned to Loomi’s personalized product recommendations.

The cosmetics and beauty brand was able to deliver on-site product recommendations within 0.1s of customer action, leading to a 17.5x increase in clicks on a recommended item and an 11x increase in purchase rate.

Yves Rocher can now deliver product recommendations within 0.1 seconds, fast enough that the recommendation reflects what the customer did moments ago — even for anonymous, first-time visitors.

Real-Time Decisions You Can Verify

With Bloomreach’s unique in-memory framework architecture, you get simplicity, efficiency, and real-time execution. When evaluating vendors, it’s not enough for them to say they’re “real time” — they need to have the technology and the numbers to back that up. 

Your warehouse stays the core (aka the record of who your customers are). Loomi brings that knowledge to where your customers are currently interacting, meeting live behavior with the right email, ad, search result, or recommendation. Plus, marketers can build segments, calculated attributes, and personalized experiences without having to submit urgent requests to IT and BI teams.

Ready to take advantage of our in-memory framework and deliver intelligence at the edge? Schedule a personalized demo today to get started.

Tags
Photo

Divya Mulanjur

VP of Product Marketing at Bloomreach

Divya Mulanjur is VP of Product Marketing at Bloomreach. Over her career she has run marketing teams at brands and marketed software at vendors, so she knows how marketing technology gets bought, sold and judged, and what it takes to make it work once it\’s in place. That background shapes how she writes about ecommerce, customer engagement, and AI for marketers deciding where to put their time and budget.

Table of Contents

Share with Your Community

Copied!

Subscribe to our newsletter

Recent Posts

Maintain an Edge With These New Posts

bloomreach-avatar-menu-1
bloomreach-avatar-menu-3
bloomreach-avatar-menu-2

Join 15,000+ recipients getting the latest insights on AI and ecommerce delivered straight to their inboxes.