How winning CRM teams turn BFCM shoppers into repeat buyers with AI

The following recap pulls from our recent webinar, How Winning Teams Use AI to Turn BFCM Shoppers into Repeat Buyers, featuring Katharine Toll and Paige Rotar from Simon AI. Both work closely with direct-to-consumer brands, helping marketing teams use agentic AI to launch campaigns faster, uncover new revenue opportunities, and personalize at scale.

You'll acquire 30-40% of your annual customers this Black Friday / Cyber Monday. Most will never buy again.

Not because your product failed them or because they didn't like your brand. But by the time you figure out who they actually are and what they need next, the moment is gone. You spent millions on acquisition, but your tools can't process changing customer intent and real-world context fast enough to convert them into repeat buyers.

After 90 days, first-time buyers who haven't returned rarely become valuable customers. If you convert just a fraction more during this window, you’ll fundamentally change your growth trajectory. The stakes are high. The window is narrow. And most teams can't act fast enough to capture it.

Why most CRM teams can't keep up

By early December, you’re looking at tens of thousands of first-time Black Friday / Cyber Monday buyers grouped into a single “BFCM 2025” segment. Yet, they certainly don’t all belong in the same follow-up journey.

Some shoppers came in for the discount and are unlikely to respond to full-price offers this month. Others were buying gifts and probably won’t re-engage until they start shopping for themselves later in the season. Then there’s a sizable group who spent real time exploring new categories, browsing repeatedly, adding items to their carts, and reading reviews. These are all signals that they may be open to a second purchase soon.

Your job is to separate these patterns so each customer gets the follow-up that actually fits their intent.

The data you need is in your warehouse, but accessing it means submitting a ticket to engineering. By the time the fields get built, it's mid-December. The 90-day window is half gone. Even teams that already have the right data face the same slowdown because segment creation is still manual, rules-based work.

Traditional marketing operates on rules and schedules. But customer intent ebbs and flows, changing quickly. The difference between a team that can process all these signals while they're fresh and one that's always two weeks behind is the difference between profitable growth and expensive churn.

How leading brands are solving this with AI

Working with retail brands through their highest-stakes acquisition moments, I've seen the teams that are winning this race for repeat buyers do something different. They're not trying to move faster manually. They're using AI to process customer signals and real-world context at machine speed so they can respond to intent with value and relevance while it's still fresh.

Three AI plays are fundamentally changing how this work gets done. Together, they deliver personalization that matches how customers actually shop and give CRM marketing teams what they need to turn first-time BFCM buyers into repeat customers.

Each play helps you surface intent earlier, respond while it’s still fresh, and keep more of your newly acquired shoppers engaged during the 90-day window when repeat behavior forms. AI turns data into live attributes, activates campaigns when real-world signals align with intent, and evaluates every customer individually.

‍

Here's what each play does, and how they work together:

Play 1: AI-powered audiences — building and scaling actionable signals.

Most CRM teams work with the same static segments—"Deal-Seeker," "Lapsed," "High-Value"—not because those buckets represent the only signals that matter, but because those are what is wired into their systems. Product reviews, support transcripts, browsing patterns, and sentiment signals can be difficult to access and understand, require engineering tickets, and risk long turnaround times. 

AI-powered audiences transform the marketing workflow from static labels to executing with dynamic, predictive attributes called AI Fields. Instead of just knowing someone was a "Deal-Seeker," you can layer on social-trend affinity, post-peak purchase propensity, or channel preference shift. This represents a shift from manual, slow, rule-based segmentation to automatic, fast, and adaptive relevance for the real-world signals you need to identify who is most likely to buy again soon.

How AI-powered audiences work

AI-powered audiences run through four ongoing steps that update audiences as customer behavior changes:

• Analyze live data - Agentic AI brings together what customers browse, buy, and respond to, along with the signals they leave in reviews or support chats.
  
• Uncover hidden patterns - As behavior shifts, AI picks up emerging themes, changes in sentiment, and early signs that interest or loyalty is moving in a new direction.
  
• Assign actionable attributes - These signals become AI Fields that continuously update to reflect what a customer tends to prefer, how they engage, or which channels they lean toward.
  
• Create dynamic audiences - When similar patterns show up across customers, new audiences form automatically. They stay current because they use live data to update based on what people are doing now, not old segment rules.
  
  

For example, if someone moves from weekly to monthly purchasing, the system notices the shift immediately, AI Fields and audiences are updated, becoming immediately actionable rather than discovered weeks later.

Play 2: Contextual activation — triggering messages when real-world context and intent align

Your winter apparel campaign runs on a schedule. Every Monday at 10 am, everyone who browsed jackets gets the same message, regardless of whether it's 65 or 30 degrees in their city. Your segments are static. They can't keep up with changing conditions like weather shifts, trending styles, and local events. By the time your campaign launches, the moment may have already passed for a customer. 

Contextual activation is powered by AI agents that create dynamic, always-on optimization that incorporates real-world signals, like weather, social trends, and local events. Instead of calendar-driven campaigns, you combine audience intelligence with context to trigger messages when intent and conditions align. The result is hyper-relevant engagement that is based on the moment, not just the audience.

How contextual activation works

The system works in a steady loop:

1. Signal fusion - AI combines what customers are doing with what’s happening in the world around them, weather shifts, local events, seasonal changes, or trends that may influence why they’re shopping.
   
2. Intent detection -  As those signals overlap, the system notices when real-world conditions line up with a customer’s interests. A cold snap, an upcoming storm, or a regional event can turn past browsing into a meaningful moment.
   
3. AI Moments - When the timing is right, the system sends timely messages tied to that context. If temperatures drop in Denver, customers who have been exploring outerwear receive cold-weather recommendations. If conditions stay warm, the message waits until it’s actually relevant.

Campaigns are triggered by real-world signals to match the when and why a customer is likely to make a purchasing decision. Post-BFCM engagement is driven by propensity and relevance, so your CRM marketing team is sending fewer emails to people who aren't ready, and perfectly timed messages to those who are. Open rates climb. Conversion improves. You're building trust through relevance, not eroding it through spam. Contextual activation helps your team reach customers when their interest is strongest, improving repeat-purchase conversion without increasing send volume.

Play 3: Synthetic consumers — Individual-level prediction and decisioning at scale

Traditional segmentation infers intent from past behavior: Someone bought winter gear six months ago, so route them into the winter gear campaign. AI Fields surface smarter patterns—deal-seeker affinity, channel preference shifts, emerging interests—but you're still working backwards from data points, trying to guess what someone wants based on what they've done.

Synthetic Consumers create AI personas that think like your customers. Instead of inferring intent from historical actions, you build a dynamic model of each individual that simulates how they actually make decisions. Then you can ask it directly: "How interested is this person in this specific product right now?"The model reads the customer’s full pattern of behavior and produces an affinity score that reflects their level of interest. It’s not a spreadsheet of data points. It’s a way of asking a representation of the customer how likely they are to engage.

How synthetic consumers work

Synthetic consumers operate through three connected steps. Together, they create a representation of each customer you can query directly and activate in real time.

1. Dynamic customer models - The system builds a working profile, an AI persona,  that acts like the customer would across different situations. It interprets past activity, context, and trends, and turns them into actionable predictions you can use for highly relevant targeting. 
2. Intelligent interest scoring - You can ask this model about specific products or campaigns. Instead of simple yes/no logic, it returns an affinity score that reflects how likely the customer is to care about what you're offering.
   
3. Context-aware reactivation - Those scores are combined with real-world context. When the timing makes sense, AI agents determine when to re-engage the customer. When it doesn’t, the message waits.

Even customers in the same segment—deal-seekers, loyalists, gift buyers—have very different attributes and interests. Synthetic consumers let you personalize at the individual level while making thousands of decisions automatically. This lets you prioritize high-affinity customers during the 90-day window when repeat behavior forms, instead of relying on broad segments that treat all BFCM buyers the same.

Depending on your data, goals, and readiness, all three of these plays can work together. Here's what that looks like in practice. 

How AI turns a Black Friday browser into a repeat buyer

Taylor browses your site on November 29th. She's 34, lives in Denver, and has abandoned cart three times researching insulated jackets for backcountry skiing. She finally makes a first purchase during your 40% off sale: a mid-weight jacket, first-time customer.

Now, AI agents scan her complete behavioral profile. She viewed technical gear, not fashion outerwear. Her browsing history shows skiing and hiking categories. Cart abandons suggest research, not price-hunting. Purchase timing coincided with ski resort announcements. She gets dynamically categorized into "High-Value First-Time Buyers" and "Cold Snap Layering Seekers", live attributes that appeared the day the pattern emerged.

From December 1, AI monitors external signals for contextual activation. Regional temperatures forecasted to drop December 4-6. Local ski resorts are announcing opening day. Her browsing showed interest in base layers she didn't purchase. The system holds the message, waiting for context to align with intent.

On December 4, temperatures hit 28°F in Denver. Ski season opens. The system queries Taylor's synthetic consumer model: "What's her affinity for technical base layers right now?" The model evaluates her complete profile and responds: "I'm gearing up for another ski season in the Rockies. I need reliable gear that can handle backcountry conditions. I'm looking for technical outerwear that performs in extreme cold but won't overheat me on the uphill."

Taylor’s affinity score reaches 89/100 for technical base layers.

Taylor receives a "Comfort starts at the base" email featuring technical base layers, triggered by demonstrated interest, weather alignment, and individual-level affinity scoring. The product recommendation matches what she's planning this weekend. No discount needed because relevance is the value.

At the same time, AI is busy processing similar scenarios for tens of thousands of other BFCM customers, routing gift buyers to March campaigns, suppressing discount-seekers from full-price offers, and matching product enthusiasts to relevant categories.

One journey template becomes thousands of individually timed moments. This level of precision requires clean data foundations and clearly defined campaign goals. But once configured, these decisions happen automatically.

"Getting to that why is important, but being able to systematically do that across your entire customer database is the real challenge." ^{Katharine Toll, Head of Sales, Simon AI}

Explore how AI can improve your BFCM performance

Ready to explore what's possible for your brand? Book a 45-minute AI Use Case Workshop with our team. We'll walk through high-impact AI use cases tailored to your goals, identify fast-start ways to scale personalization, and show you how AI connects customer behavior with real-world context to turn Black Friday / Cyber Monday buyers into repeat customers. You’ll walk away with three use cases to help you activate real-world signals and convert more of your BFCM shoppers into repeat buyers. Request your use case workshop here.