.avif)
About the brand. Beam sells wellness routines - Dream for sleep, Glow for hormone balance, Biotic for gut, plus collagen, creatine and a kids' line. Products people take daily for months, which makes Beam a subscription-first business. The team ships new flavors, bundles and seasonal drops constantly, and moves quicker than most brands its size.
Beam went to market for a subscription platform for one reason above all others: the team wanted a subscriber portal they could shape - personalize by cohort, run offers inside, and change without a build cycle.
Beam had never meaningfully touched its customer portal. The team took it to their dev agency and asked what customization would involve. The answer was a portal built from scratch, which wasn't a project Beam wanted to own. So it stayed as it was.
For Joy Huang, the bigger miss was what the portal wasn't doing. Beam has a catalog built for cross-sell - sleep next to gut next to collagen and a logged-in subscriber already in the door. But every subscriber saw the same thing, whether they'd been on Glow for a year or started Dream last week.
The day-to-day added up around it. Routine catalog changes turned into manual projects, and reactivation and payment recovery were each running on separate tools alongside the stack.
“I spend a good chunk of time, and other people on my team too, manually bulk updating things every day.”
— Joy Huang
The rebuild started with one rule: nobody gets pitched something already in their box.
Instead of one product shelf shown to everyone, Beam built personalized upsell/cross-sell profiles in the portal each mapped to what a subscriber already has and what they don't. A Glow subscriber sees what pairs with Glow. A Dream subscriber sees what pairs with Dream. Each cohort gets its own banner creative, so the offer reads as the next step in a routine rather than an ad.
Every profile reports on its own, so Beam can see which cohorts are adding products and which aren't - and adjust the list, the order or the creative without touching code.

The portal profiles run all year. On top of them, Beam runs a steady calendar of limited-time campaigns and one-click quick action links. What makes them work isn't the offer - it's how deliberately each one is set up.
Who sees it: start with the list, not the offer
When and where: give every offer a reason and a route
What the offer looks like: easy to say yes, protected on margin
Every route points at the same place. Whether a subscriber clicks an email, a portal banner or an order reminder, they land on an offer built for the subscription they already have.
Most brands judge a cancel flow by its save rate. Beam's is 21% - a little over one in five subscribers who started to cancel stayed. But a save rate only tells you someone clicked “keep my subscription.” It doesn't tell you whether they paid for another box.
The number that does is how much of the MRR you saved actually came back as processed orders. Across Beam's first 12 months, saved subscriptions went on to generate revenue equal to 92% of the MRR the cancel flow saved, with each saved subscription placing more than one further order on average.
12 MONTHS OF SAVES, generated
~$2.16M
in additional revenue Beam has collected from saved subscriptions since go-live
That gap — or lack of one — is what turns a cancel flow from a retention widget into a revenue line. When saved MRR and additional revenue generated sit this close together, the saves are real. When they drift apart, it shows exactly which offers are buying a click rather than another box.
A year in, Beam asks for things the way operators do - specifically, with the use case attached, usually because a launch is already on the calendar. Several of those asks have gone from a message in a shared Slack channel to something live in the product.

That's the quiet reason the move stuck. Beam moves fast, and the platform has kept pace - less by anticipating what the team would need and more by building it once they asked.

Results reflect Beam’s measured experience; outcomes vary by brand, vertical, product, and implementation.
