“The peak is now a season, and many build plans still assume a weekend.”

Cyber Week is five days. The season around it is much longer. Adobe forecasts $275.1 billion in U.S. online sales from November 1 to December 31, 2026, up 6.7% year over year. It separately forecasts $95.8 billion in October alone, up 8%, and says the October Prime Day event has changed the shape of the holiday season by adding a pre-season spike in shopping interest.

A build plan organized around one peak weekend fits that calendar badly. The work of being relevant now has to be spread across weeks, and our view is that sending more will not solve it. The brands that do well shorten the distance between a relevant idea and a live journey that has already been through QA. BrazeAI Operator can help close that distance, and this article looks at where it helps, where people still make the call, and what has to be in place first.

Holiday personalization gets harder just as the stakes rise

The old holiday model made sense. Build capacity is finite, and a calendar of a few large segments and scheduled promotions is easier to QA and safer to send. For the anchor sends of the season, that logic still holds.

What has changed is that the offer keeps moving. Adobe forecasts discounts of up to 14% in early November, up to 21% just before Thanksgiving, and 10 to 15% through December. Each shift is another offer state that creative, exclusions, and rules have to reflect. Layer on audience differences and channels, and the variant count grows faster than a team does.

Concurrent campaigns add their own load in frequency, suppression, and overlap decisions, often across teams. And the offer is not the only thing that shifts during the season. The shopper does too.

Why is holiday relevance a build capacity problem?

Adobe says many shoppers are using discounts to stock up on essentials, and forecasts Cyber Week sales of personal hygiene products up 150% and clothing basics up 210% against September averages. One shopper is buying a gift while another is restocking their own home, and the same promotion lands differently on each.

In Braze’s 2026 BFCM consumer research, only 9% of surveyed consumers said brands exceeded their expectations during the 2025 BFCM season. It covers the U.S. and four European markets, so it reads best as a directional view alongside Adobe’s U.S. numbers. Even so, it suggests very few shoppers feel brands are getting the season right.

At this point, relevance is a capacity problem. Every extra audience path needs logic, creative, fallbacks, and QA. A team that can only build two versions will only ship two. Braze positions Operator for teams with more ideas than build bandwidth, which makes it worth being clear about what Operator actually takes off a team’s plate.

What BrazeAI Operator builds, and what still needs human approval

So the useful question is which parts of holiday build work AI should take on.

“AI operator” gets used for many things, including general-purpose agents. Here it means one product: BrazeAI Operator, an assistant inside the Braze dashboard that became generally available on April 23, 2026. From a plain-language brief, it drafts campaigns, Canvases, segments, Liquid, copy, and images. By default, each proposed change arrives as an action card that a person reviews and approves. Operator works within the permissions of whoever is using it, and it cannot launch anything that affects customers. A person does that.

Decisions made while a customer is inside a journey sit elsewhere. Agent Console agents run in a Canvas Agent step when a customer reaches it, handling tasks such as content generation, classification, and routing. One of Braze’s own examples returns a single route key from a list of eligible options, with DEFAULT as the fallback when data is missing or ambiguous. Operator can help configure such agents, but the agent does the in-flight work.

Braze already supports behavior-triggered, cross-channel journeys, and Operator changes the human effort needed to build and vary them. Braze reports that the American Diabetes Association used it to move an ecommerce journey from a single message to several steps.

In our view, Operator is a build and execution assistant rather than a learning system. Our comparison of Operator, Agent Console, and Decisioning Studio covers where each one fits. With that line drawn, here is where Operator can help most during the holidays.

Five holiday journeys BrazeAI Operator can help build

Across five holiday journeys, AI takes on the build, and a person keeps the decision.

1. Personalized holiday messaging. Operator can write Liquid grounded in the attributes, events, and catalogs already in the workspace, with fallbacks for missing values. Which signals should change the message is still a human call.
2. Cart and browse abandonment. You can describe a moment such as an abandoned cart, and Operator drafts the Canvas with audience, content, entry criteria, and steps. People set timing and decide how offers escalate.
3. Product and offer personalization. Operator can build segments with catalog conditions and propose AI item recommendation predictions. Offer rules, including depth, margin, and exclusions, remain a business decision. Individual-level optimization belongs to Decisioning Studio.
4. Journey routing. An Agent step can read recent behavior and return one route from an eligible list, with DEFAULT as the safe fallback. People define the routes. We have built client-specific agents that classify users by intent level, purchase frequency, and engagement depth (our Operator guide).
5. Personalized copy at scale. Operator drafts channel copy and applies brand guidelines added as context. Agents can write per-customer copy at send time within character limits, and Braze’s examples return an explanation field for QA to check. People review what ships.

Across all five, the gain comes from routing and content choices rather than volume, and total sends may well be lower. Operator shortens the build, an agent handles routing, and a person sets the offer rules. That split only holds up if a few conditions are in place first.

What has to be true for Operator to work in holiday campaigns?

None of this works on its own. Operator’s output is grounded in what exists in the workspace: attributes, events, and catalogs. If browse events or purchase history are not tracked cleanly, there is little to personalize with, and that is where we would start any holiday plan.

Review is the second condition. Approval is Operator’s default, but defaults can be changed: Braze offers an auto-approve setting that lets suggested actions run without manual review, and its guidance on limiting data exposure includes leaving that off. Launching stays with a person either way, but we would settle who can switch auto-approval on before peak season, when review is most tempting to skip. There is also a company-wide daily usage limit, with Canvas edits counting as heavier use, so plan builds early.

Production discipline is the other half, and we see it in our work for Sur La Table. Since October 2022, Mavlers has worked with Sur La Table, including a dedicated campaign manager for daily production: a target of four to five email campaigns a day, multiple SMS campaigns a week, QA before every send, exclusion lists and segmentation, and a pre-production workflow for next-day requests, with zero missed deadlines. The work started in Emarsys and carried through Sur La Table’s move to Braze, which Mavlers also supported. It predates Operator, but it shows what high-volume production demands.

AI multiplies how many variants a team can create, and QA and exclusion discipline decide whether those variants are safe to send. As a certified Braze partner, Mavlers works in that operating layer: journey architecture, data readiness, and QA. With those conditions in place, the next question is where to begin.

Where ecommerce teams should start before the holidays

Start with a short list of journeys rather than the whole calendar.

Pick two or three holiday journeys where the gap between generic and relevant is widest, and volume is highest. Cart abandonment is usually the obvious first candidate. Build and vary them with Operator, define the routes and offer rules yourself, QA everything, and measure against a holdout. If a change freeze is close, refine journeys that already exist rather than rebuilding them.

If you want a second opinion on where to start, our Braze team is glad to look at it with you.

A holiday season is won on the number of distinct decisions a team can afford to get right, and that number has very little to do with how many messages go out.