The Frame
In an app, design has interaction to lean on: taps, scrolls, progressive disclosure. An email has none of it. Customers open, scan, and close within seconds. Mail clients render the same code differently. Legal boundaries strip out the marketing toolkit before the file even opens.
What's left is the information itself.
When you can't rely on interaction, information architecture is the UI.
The Problem
A platform migration was the trigger. Chase was moving Ultimate Rewards email off a legacy vendor and onto an internal system, the same engine that would eventually power every redemption email across the program. The migration didn't ask for a redesign. It just made one possible for the first time in years.
Then we looked at what we were migrating.
Seven redemption modules. Two brands: Chase and JPMorgan. Millions of emails sent every month, every one of them mandatory and unsubscribable, every one of them carrying transaction details a customer might need to verify, dispute, or trust. And underneath all of it, a content system that had grown one template at a time, one team at a time, for years.
The first signal didn't come from analytics. It came from branch tellers and call center transcripts. JPMorgan customers, whose banking app uses a distinct theme, distinct colors, and distinct visual language, were receiving redemption emails styled in Chase blue. They didn't recognize them. So they did what cautious customers do: they walked into a branch, or called the line, and asked whether the email was real. Some flagged it as spam and deleted it.
We pulled the numbers.
The second problem was structural. On mobile, where most of these emails are opened, critical information was buried below decorative elements that had accumulated over years of one-off additions. Customers had to scroll past a card image, a greeting, and a marketing-flavored header before reaching what they actually came for: did the redemption go through, how much, when. By the time the data arrived, the customer had already decided how the email made them feel. Usually: uncertain.
Trust is the one thing a transactional email has to earn every time. So before we touched the design, we needed to understand exactly where it was being lost.
The Research
Before redesigning anything, we needed to know exactly which moments in the email were producing doubt, and which were producing trust. So we went where the doubt lived.
Four findings shaped every decision that came next.
01
Unlike app browsing, where customers explore, tap, and drill in, email behavior is binary. Subject line. Status. Key data. Out. Most readers spent under fifteen seconds inside the email. The implication was uncomfortable: any information that didn't land in the first half-second was, functionally, not in the email at all.
02
In testing across error and refund scenarios, the colored status tag was the single element customers consistently looked for first. Not the dollar amount. Not the redemption type. The tag mattered because it answered the question they actually had: did this go through, or do I need to do something? Across every module we tested, the pattern held. The status tag wasn't a decoration; it was the entry point to comprehension.
03
Customers told us, directly and indirectly, that getting a precise transaction breakdown made them feel safe. Not "approximately 25,000 points." Exactly 25,317. Visible, scannable, in a format that matched what they remembered from the redemption flow. A data list that felt loose or rounded was treated as suspicious, regardless of whether the underlying numbers were correct.
04
Legal disclosures, FAQ links, and marketing-adjacent footers were not reaching customers in any meaningful way. Engagement metrics confirmed it. Heat-map style scroll behavior confirmed it. The bottom of the email existed for compliance reasons, not customer reasons. This finding wouldn't change what we put down there, because legal requirements set those boundaries, but it sharpened what not to put there. Anything we wanted customers to actually see had to live above the fold.
The research didn't tell us how to design the email. It told us what every customer was already trying to do, and where we were failing them.
Three Decisions
All three were refusals, and each one cleared space so the right signal could land.
01
Every email had to anchor the customer's attention somewhere. The question was where.
There were three candidates. A personalized greeting, warm, human, easy to defend. The data summary, straight to the numbers so the customer gets everything in one block. And a colored status tag, a single visual signal that answered the question before the data did.
In testing across error and refund scenarios, the third won immediately. Customers weren't asking "what's the breakdown?" They were asking "did this go through?" The greeting answered nothing. The data summary answered the wrong question first. The status tag answered the right question instantly.
Click a row ↓
The status tag became the anchor of every email across every module, every status, and every brand. That single decision gave seven modules the same opening pattern. Whatever the redemption type, whatever the outcome, the customer's eye lands on the same element first.
02
The card image was the first thing customers saw, and the first thing we cut.
The legacy template led with a card product image. It was visually warm. It was on-brand. Every team that touched the email, from design to marketing to brand, had a reason to keep it. The argument for removal had to be louder than five years of accumulated comfort.
Testing made it loud. The card image consumed the top quarter of the email on mobile, pushing the status tag below the fold in error scenarios. And the only piece of information the image carried, which card was used, was already present in the transaction details, where customers actually looked for it. The choice wasn't "card image or no card image." It was "card image or status tag in the first half-second." That comparison wasn't close.
Before
After
Removing the card image moved everything up. The status tag claimed the space it needed. The transaction details rose into the first viewport. Mobile customers stopped scrolling past decoration to reach information. The card itself didn't vanish from the email; the card name and last four digits stayed in the transaction details, exactly where customers had been looking for them all along. What disappeared was the decoration. The information stayed.
03
The space at the bottom of the email was the most contested real estate in the project.
There was a proposal on the table to add commercial entry points to the email footer: redemption banners, cross-sell prompts, and links into other UR earning products. The case for it was reasonable. Customers were already opening the email. Engagement was guaranteed. The footer was unused space. Marketing wanted it. The business wanted it.
Two things made us push back. The first was a regulation: mandatory transactional emails in this category cannot carry commercial CTAs or marketing campaigns. Only content tied directly to the transaction and applicable terms. That was the floor, and it was non-negotiable.
But the regulation alone wasn't the argument. A commercial banner in a transactional email, even a small one and even one we could legally justify, would do something the rest of the design was working against: it would make the email feel like marketing. The whole redesign was built on the premise that customers needed to trust this email on sight. The moment a CTA appeared, the email became a sales surface. Trust dropped. The very signal we were trying to amplify, this is a real, official, transactional message, would be diluted by the thing meant to drive engagement off it.
Excluded
The footer stayed transactional. Legal disclosures, applicable terms, customer service contact, and nothing else. The decision cost the business an engagement channel. It bought back something more valuable: an email that customers could open without scanning for what was being sold to them. Every other decision in the redesign was about removing a competing signal. This one was about refusing to add a competing motive.
The system that emerged was the smallest version of the email that still did its job. Everything else had been refused into existence.
THE SYSTEM
Three refusals, three redesigned moments, but the deliverable wasn't a template. It was a system: one configurable email engine that every redemption module could route through, every brand could re-skin, every status scenario could plug into. Built once. Configured every time a new module came online.
Every redemption email, whether successful, failed, refunded, in-progress, or cross-module, passes through the same six zones, in the same order. What changes between emails is what each zone is configured to carry: which status color, which transaction structure, which contextual explanation, which legal terms. The structure stays. The content reconfigures.
The proof is in the variations. Below: the same template, configured for three redemption journeys: a Gift Card success, a Points Transfer that failed to proceed, and a Cash Back confirmation. Different statuses. Different content blocks. Different module-specific data. Same underlying structure. A customer who has used Chase Ultimate Rewards before will recognize the shape of the email even when they've never received this particular kind before. Recognition compounds across time.
The system also configures across brand and viewing mode. JPMorgan customers see a JPMorgan-themed email with the same structure and status patterns, but with JPM's bronze accent and brand voice instead of Chase blue. Both brands ship in light and dark mode without per-mode redesign. The finding that the email looked like a scam closed at the system level: when a JPM customer opens this email, it looks like JPM. When they open it at night, it looks like night.
The hardest part of this project wasn't designing the email. It was making the email designable by people who weren't on this team. Each redemption module has its own product team, its own content owner, its own roadmap. If the system required a designer in the loop every time a Gift Card promotion launched or a new Transfer Points partner came online, it would have failed within a quarter.
So the work expanded. We co-mapped the full content inventory with our content strategy partner, covering every module, every status, and every legal-required line, into a single source of truth that module owners could pull from directly. We aligned with engineering on which zones would be component-level configurable vs. content-level configurable. We brought JPM brand team in early to validate brand parity before any module-specific reskinning began. By the time the system shipped, Gift Card and Transfer Points teams were configuring their own variants in the new template without filing a design ticket. That was the proof. Not the visuals. The fact that the visuals could be configured without us.
The redesign didn't make the email better. It made every future email easier to make right.
IMPACT
This case study isn't a retrospective. The system is in flight: two modules, Gift Card and Disney Redemption for Disney card customers, are running in pilot at 40% of their respective audiences, with the remaining five modules scheduled to ship to all customers by the end of June 2026. What's already true, what's being measured, and what the system is built to make possible are the three things worth reporting now.
Two pieces of feedback came in before any quantitative data could. The first was from the Gift Card module team, the people who would have to live inside the new template every day. The second was from JPMorgan's digital branding lead, the person whose customers had been mistaking our emails for spam.
Both are noticing the same shift, in two places: the team that uses the template, and the customer who trusts the email.
Three signals are under monitoring during the pilot phase. The framework, which covers baseline, early signal, and expected direction, is the deliverable here, not the numbers themselves.
If the system is working, all three move in the same direction. If only one moves, the design solved a narrower problem than we thought. If none of them move, either the research was wrong or the rollout was incomplete. Each outcome is informative.
The most durable outcome of this project isn't anything the customer will ever see. It's that two upcoming redemption modules, Invest Your Points and Apple Pay integration, can ship through this template without a designer-led redesign cycle. The Apple Pay team can configure their email the way the Gift Card team is configuring theirs now. The brand and status patterns hold. The legal boundaries hold. The system absorbs the new module instead of requiring a new design pass per module.
That's the version of "shipped" that matters at this scale. Not one good email. A template that makes every next email easier to make right.
Pilot or production, the work is what the next team can do without us.
What's Next
Two open questions. The first is concrete. The second is the one I'm watching longer.
The system configures across brand, status, and mode, but not yet across language. Spanish UR emails currently fall back to English layouts, and Spanish phrases run 20–30% longer than their English counterparts. The hierarchy that holds in English starts breaking before the fold. Language affects information hierarchy itself, which makes it a harder system problem than brand or mode.
Spanish layout current fallback
The system works today because ownership is clear: one design team, one content team, two engineering teams. As modules and configurations stack up, the question shifts from design to governance: who owns the patterns when the original designers move on. That's the part of this work I'm watching most closely.
Every project I work on tries to do one less thing than it could have. This one was no different. The redesign that mattered wasn't the email. It was everything we agreed not to put in it.
