Reading was effortless. Remembering wasn't.
The inbox delivered information. It didn’t preserve intention.
People checked email constantly.
While commuting. Between meetings. Waiting in line. Away from their desk.
Messages were opened almost immediately.
Important information was seen.
Requests were noticed.
Conversations stayed visible.
Yet surprisingly little action happened in those moments.
Emails were marked unread.
Starred for later.
Left sitting in the inbox.
People weren’t ignoring work.
They were postponing it.
Most participants described the behavior differently:
"I'll deal with this when I'm back at my computer."
At first, we believed the challenge was acting on email from mobile devices.
Product
Twake Mail
Company
Linagora
Timeline
5 weeks
Focus
Deferred Action
Decision Making
Attention Management
Follow-up Reliability
Methods
Shadowing
Mobile Diary Study
Workflow Mapping
Behavior Analysis
Prototype Validation
UNDERSTAND THE BREAK
Signals from the field
The project began with a pattern that appeared repeatedly during conversations about mobile email use.
People weren't ignoring emails.
They were reading them.
The problem happened immediately afterward.
A notification would appear while commuting.
An email would arrive between meetings.
Someone would quickly scan a message while waiting in line.
The email was seen.
The action never happened.
Across interviews, observations, and diary studies, participants described a surprisingly similar behavior:
They opened emails to avoid missing them. But the inbox offered no persistent signal of what still needed action, so carrying that intention became entirely their own responsibility.
At first, this seemed like a mobile productivity problem.
The behaviors suggested something different.
The challenge wasn’t reading email. It was returning to it.
Signals
Drag the top card to explore the artifacts behind this story.
What Connected These Signals
Although the situations varied, the behavior was remarkably consistent.
People compensated by:
Leaving emails unread
Flagging messages for themselves
Reopening conversations repeatedly
Searching for emails they remembered reading
Forwarding emails to themselves
Creating personal reminder systems outside the inbox
The inbox successfully delivered information.
What it didn’t support was deferred action.
People could see the email. But after they closed it, nothing in the system helped them stay aware that action was still needed.
As a result, unread status gradually stopped representing reading behavior and started representing intention.
The inbox became a place where unfinished work accumulated.
Synthesis Wall
Recurring patterns clustered into five themes:
Deferred Action
Memory Reliance
Unread as Reminder
Follow‑up Uncertainty
Interrupted Workflows
These themes appeared consistently across interviews, observations, diary studies, and industry research.
Raw observations were clustered into recurring behavior patterns, revealing a common challenge: people could read emails immediately, but struggled to preserve intention afterward.
From Patterns to People
The behavior wasn’t limited to a specific role, industry, or team.
The same challenge appeared across support agents, managers, sales representatives, operations staff, and product teams.
Although their responsibilities differed, they often encountered the same situation:
An email arrived. They read it. They understood it. They couldn’t act immediately.
Rather than grouping participants by job title, I focused on how they managed unfinished work after the moment of reading.
Several recurring patterns emerged:
Some relied on unread status
Some built personal reminder systems
Some repeatedly reopened conversations throughout the day
Others postponed action until they returned to a larger screen
Despite their differences, they were all attempting to solve the same problem:
How do I keep track of work I can’t complete right now?
That observation became the foundation for the behavioral archetypes below.
Behavioral Archetypes
Understanding recurring follow-up behaviors
The same challenge surfaced across support agents, managers, sales representatives, operations staff, and product teams.
Their responsibilities varied. Their moment of hesitation did not.
An email arrived. They read it. They understood it.
They just couldn't act on it right then.
Rather than grouping participants by job title, I focused on how they tried to preserve intention after reading.
Seven recurring strategies emerged — each a different workaround for the same missing product support.
Seven behavioral archetypes, each representing a distinct strategy for keeping unfinished work visible. Despite the variety, all compensated for an inbox that didn't preserve intention after reading.
What Connected These Archetypes
Although the archetypes approached email differently, they all emerged because the inbox offered no persistent awareness of unfinished work.
People were forced to carry that intention themselves.
The specific behaviors varied. The underlying goal did not.
Participants consistently created ways to preserve intention until they could return and act:
Leaving emails unread
Flagging messages
Reopening conversations repeatedly
Forwarding emails to themselves
Building personal reminder systems outside the inbox
Different mechanisms. The same purpose.
People could access information immediately.
They struggled to preserve intention afterward.
Across every archetype, these workarounds were compensating for the same gap:
The system helped people read information. It provided little support for sustaining awareness of what they intended to do with it later.
As a result, unfinished work often depended on memory rather than a system.
That recurring pattern raised an important question:
Why did people need personal reminder systems simply to keep track of work they had already seen?
That question became the focus of the next stage of analysis.
Where it breaks
The question raised by the archetypes — Why do people build personal reminder systems just to track work they’ve already seen? — pointed to a specific moment in the workflow.
Not when email arrived.
Not when action was finally taken.
The breakdown happened in the space between.
Participants often encountered emails at moments when they could read, but couldn’t act.
While commuting.
Between meetings.
Waiting in line.
Away from their desk.
The email was read.
The intention was formed.
The action was deferred.
At that moment, people faced a small but important challenge:
How will I know this still needs action when I return?
The inbox provided very little support for answering that question.
As a result, people compensated by:
Leaving emails unread
Flagging messages for themselves
Forwarding emails to another inbox
Reopening conversations repeatedly
Creating reminders outside the product
Carrying the intention in their head
Over time, these workarounds became increasingly difficult to manage.
Unread counts grew.
Flags accumulated.
Important follow‑ups blended together with routine messages.
The inbox remained a record of communication.
It gradually became a personal reminder system — because the system itself offered nothing better.
Follow-up Journey
What the Workflow Revealed
The repeated workarounds were not attempts to understand the email.
They were attempts to preserve intention.
Participants already knew what needed to happen.
The challenge was staying aware of that intention after leaving the conversation.
People searched for ways to answer questions such as:
Will I know what still needs action later?
How can I tell which unread emails are actually unfinished work?
What was I planning to do here?
As a result, unread status gradually stopped representing reading behavior.
It became a proxy for future intention.
That distinction became important.
The challenge wasn’t helping people access information.
It was helping people reconnect with work they had already decided to do.
The inbox successfully supported reading.
It provided very little support for sustaining awareness of what still needed action.
What's actually wrong
The pattern observed across interviews, diary studies, and workflow analysis was consistent.
People saw the email.
They understood what needed to happen.
The breakdown occurred afterward — between reading and acting.
Questions that seemed simple became surprisingly difficult to answer:
Which emails still require action?
Which unread emails are actually unfinished work?
What was I planning to do here?
Which conversations can safely be ignored?
The information remained available.
But after the email was closed, nothing in the system kept the intention visible.
Over time, awareness of unfinished work faded.
Expectation vs Reality
User Expectation | System Reality |
|---|---|
I can read this now and deal with it later. | The system doesn't help preserve that intention. |
I'll know what still needs action when I return. | Actionable emails blend together over time. |
Unread means unread. | Unread becomes a stand-in for unfinished work. |
The tool will remind me. | Memory becomes the primary tracking mechanism. |
When users consistently invent their own reminder systems, the problem is rarely discipline.
It's usually a signal that the product isn't supporting an important part of the workflow.
Reframed Problem
The issue wasn't access and visibility.
The issue was follow‑through.
Instead of asking:
"How do we help people read and act on email from mobile?"
We began asking:
"How do we help people return to unfinished work after reading it?"
That shift fundamentally changed the direction of the project.
Rather than optimizing reading behavior, we began exploring ways to preserve intention across time, devices, and interruptions.
Instead of asking:
"How do people remember unfinished work?"
We began asking:
"Why should they have to remember it at all?"
That question became the starting point for the concepts explored next.
EXPLORE THE DECISION
Two ways to solve it
Once the research made the root problem clear, the conversation shifted.
The challenge was no longer helping people read email.
People were already doing that.
The challenge was helping people return to unfinished work after the moment of reading.
That led to two competing directions.
Both attempted to reduce reliance on memory.
Both appeared capable of improving follow‑through.
The real question was which approach addressed the underlying problem — and what each required from the user.
Option A — Strengthen Personal Reminder Tools
This direction assumed the fix was helping people manage their own reminders more effectively.
People already used unread emails, flags, stars, and folders as personal reminder systems.
Perhaps the answer was making those tools more powerful.
Early concepts explored smarter flagging, follow‑up scheduling, personal task lists tied to email, and stronger unread management.
What it offered
Familiar behavior. Minimal workflow disruption. Greater personal control.
What it required
Manual maintenance. Active management of another organizational layer.
The user remained responsible for preserving intention — the tool simply provided more places to store reminders.
This direction treated follow‑up as an organizational problem.
Option B — Make Intention Visible
The second direction approached the problem differently.
Instead of helping people create reminders, it focused on preserving intent inside the workflow itself.
Early concepts explored deferred‑action indicators, return‑later states, follow‑up visibility, and context that survives interruptions.
What it offered
Reduced memory dependence. Lower organizational overhead. Better continuity across devices.
What it required
New interaction patterns and states that email interfaces don't traditionally expose.
The system would need to carry some of the awareness that people were currently carrying alone.
This direction treated follow‑up as an awareness problem.
The Real Difference
At first glance, both directions seemed to solve the same problem.
But they exposed a fundamental design tension: memory versus awareness.
Option A assumed: People should manage their own memory more effectively.
Option B assumed: The product should reduce dependence on memory altogether.
One improved personal organization. The other improved situational awareness.
Both sounded useful. Neither was clearly superior on paper.
Two competing directions: strengthen the reminder tools people already use, or reduce the need for those reminders by making intention visible. The trade‑off came down to who carries the burden of remembering — the user or the system.
To move forward, we needed to see how each approach performed against real behavior.
That became the focus of the next round of validation.
EXPLORE THE DECISION
Getting to alignment
The discussion remained unresolved for several weeks.
Both directions appeared reasonable.
Both were supported by real user behavior.
Both seemed capable of reducing forgotten work.
The challenge was deciding which direction addressed the underlying problem rather than the visible symptom.
What We Tested
To move beyond opinion, we created concepts representing both approaches.
Participants were asked to work through realistic email scenarios involving interrupted work, deferred follow‑ups, multiple conversations requiring action, and returning to emails later in the day.
Rather than measuring preference alone, we focused on behavior.
Could people successfully return to unfinished work?
Could they understand what still required action?
Could they recover intent after interruptions?
What We Expected
Initially, several stakeholders believed stronger reminder tools would perform better.
The logic was straightforward: people already use unread status, flags, and stars.
Making those tools more powerful seemed like the lowest‑risk solution.
The expectation was that better reminder controls would naturally lead to better follow‑through.
What Participants Actually Did
The sessions revealed something unexpected.
Participants rarely struggled to create reminders.
They struggled because the reminder preserved only the fact that something was unfinished — not the intention behind it.
The system stored the signal.
The meaning still lived only in their memory.
When returning later, people often asked:
Why did I leave this unread?
What was I planning to do here?
Is this still important?
Does this still need action?
The reminder survived.
The intention often didn't.
Several participants reopened conversations simply to reconstruct the context behind their earlier decision.
The work of remembering had not disappeared. It had merely been postponed.
The Turning Point
A pattern began appearing across multiple sessions.
Participants consistently performed better when the system preserved signals about unfinished work — not because they had more reminders, but because they needed less memory.
The discussion gradually shifted.
Instead of asking:
"How can we help people create better reminders?"
The team began asking:
"How can we help people return to unfinished work without reconstructing intent?"
Decision
The team ultimately aligned around the intention‑preservation approach.
Not because reminder tools were ineffective.
But because they treated the symptom rather than the cause.
The research suggested that forgotten work wasn't created by a lack of reminders.
It was created by the system's inability to preserve context between reading and acting.
Preserving that context became the foundation for the solution that followed.
BUILD THE FIX
The system we introduced
Once the direction became clear, the challenge was no longer helping people remember.
The challenge was helping people return.
The solution needed to preserve intention without creating another reminder system to manage.
Rather than introducing tasks, reminders, or additional inbox organization, we focused on making unfinished work visible at the moment people returned.
The goal was simple:
Help people quickly understand what they intended to do — without relying on memory.
To support this, we introduced a lightweight follow‑up system built around conversation state rather than unread status.
Four Intention States
The system offered four user‑applied states, each capturing a distinct relationship with an email:
Save for later – No action needed now, but worth revisiting.
Follow‑up – There’s something specific the user needs to do.
Waiting on – Action is blocked until someone else responds.
Completed – The email has been dealt with and can be mentally cleared.
These states were intentionally manual — the system cannot guess intent.
Applying them takes a long‑press on mobile or a click on desktop, and they sync across devices automatically.
Core Principles
Preserve intent
People should not need to reconstruct why they left an email. The state they chose remains attached to the conversation.
Support interruption
Work frequently pauses and resumes across devices and contexts. The states persist wherever the inbox is opened.
Reduce inbox archaeology
Users should not need to reopen conversations repeatedly to recover context. The state tells them what’s pending at a glance.
What This Changed
Together, these states transformed unread status from a personal reminder system into a clearer representation of actual reading behavior.
Instead of asking users to remember unfinished work,
the system helped unfinished work remain visible.
What changed in the interface
The redesign wasn’t about adding another reminder feature.
It was about making unfinished work visible after the moment of reading.
The original experience assumed people would remember what required attention later.
The redesigned experience preserves that intention directly inside the inbox.
Before: The Inbox Stored Information. The User Stored Intention.
The original inbox primarily communicated message status.
People could see who sent the email, when it arrived, and whether it was read or unread.
What they couldn’t easily see was:
Why they left it unread
Whether it still required action
What they intended to do later
Which conversations represented unfinished work
As a result, many participants used unread status as a personal reminder system.
Once an email was marked as read, it often disappeared from awareness.
The inbox stored information. The user stored intention.
After: Intention Becomes Visible
The redesigned inbox introduces lightweight intention‑preservation states directly into the email list.
Instead of forcing users to remember why they postponed an email, the system makes that intention visible.
These states are not automatically assigned — the system cannot guess intent.
Users apply them with a long‑press on mobile or a click on desktop, and they sync across devices automatically.
The states include:
Save for later – No action needed now, but worth revisiting.
Follow‑up – There’s something the user needs to do.
Waiting on – Blocked until someone else responds.
Completed – Done; no further attention required.
Where intention lives
The intention chip appears next to the subject line in both the inbox list and the email detail view.
This keeps the chip in a predictable location — no matter where the user encounters the email.
How it works on mobile
In the inbox list, Category Chips and Conversation Context remain visible for pre‑open scanning. The intention chip sits beside the subject, clearly distinct.
Once an email is opened, Category Chips and Conversation Context are hidden — their job of providing pre‑open context is done. The intention chip stays next to the subject, and a “Set intent” button (📌) sits in the bottom toolbar alongside Reply and Forward.
Tapping the button opens a bottom sheet with the four intention options, each marked with a coloured icon. After selection, the chip appears immediately next to the subject.
This keeps the inbox list informative and the detail view focused — with the user’s own planned action as the most prominent signal.
Before, the inbox answered “What emails exist?” After, it helps answer “What should I do when I come back?”
What Actually Changed
The most significant change wasn’t visual.
It was informational.
Before, the inbox answered: What emails exist?
After, the inbox begins answering: What should I do when I come back?
The system no longer treats follow‑up as something users must remember on their own.
It helps preserve intention directly within the workflow.
Why This Matters
Participants consistently described reading emails in moments when they couldn’t immediately act — between meetings, during commutes, while multitasking, away from their desk.
The problem wasn’t access to information.
The problem was maintaining awareness after the information had been read.
The redesign shifts the inbox from a place that stores conversations to a place that helps people return to unfinished work with context intact.
That shift became the foundation for the behavioral changes observed during validation.
What made this hard
At first, the solution seemed obvious.
If people lost track of unfinished work, simply make unfinished work more visible.
In practice, the challenge was significantly more nuanced.
The inbox already contained a large number of signals competing for attention.
Adding more indicators risked creating another problem: too much visibility can become noise.
Several tensions emerged throughout the design process.
1. Unread Already Meant Something
Many participants used unread status as a personal reminder system.
Introducing dedicated intention states created an immediate question:
If unread no longer means unfinished work, what does it mean now?
The redesign needed to preserve familiar behavior while gradually encouraging a more explicit workflow.
2. Avoid Becoming a Task Manager
Participants wanted help staying aware of unfinished work.
They did not want email to become another project management tool.
Early concepts explored task lists, reminders, action queues, and follow‑up dashboards.
While useful in theory, many participants felt these approaches introduced additional management overhead.
The goal became preserving intention without creating another system to maintain.
3. Naming the States for Instant Clarity
The four states needed names that users could distinguish at a glance, without a learning curve.
Early labels like “Return later” and “Follow‑up” caused confusion — both implied deferral, but the type of deferral was different.
Through iteration, the team landed on:
Save for later – No action; just revisit.
Follow‑up – Action needed from you.
Waiting on – Blocked on someone else.
Completed – Done, no further attention.
The names were tested to ensure they felt distinct and required no explanation.
But one question kept surfacing: Why have a “Completed” state at all? If an email is done, why not just leave it alone?
The answer came from observing real behavior.
Without an explicit “Completed” marker, emails that were finished in the user’s mind would either sit in another state (like “Follow‑up”) indefinitely, or disappear into the read‑mail abyss, making it hard to distinguish them from emails that were never acted on.
“Completed” serves as a deliberate closure signal.
When a user replies to an email, the state can transition to “Completed” automatically.
When they handle something outside the inbox — a quick call, a chat message — they can mark it “Completed” themselves, clearing it from their active mental list.
It also supports transitions from other states.
A “Follow‑up” email, once acted on, moves to “Completed.”
A “Waiting on” email, once the reply arrives and the user is satisfied, moves to “Completed.”
This gives the inbox a clean lifecycle: intention captured → action taken → intention resolved.
Without “Completed,” the system would have no way to signal finality.
Users would be left wondering whether an email still needed something, or whether they had already dealt with it — exactly the kind of uncertainty the project set out to eliminate.
4. Visibility vs. Simplicity
Every new state added clarity.
Every new state also increased complexity.
Questions repeatedly surfaced:
How many states are too many? What deserves inbox visibility? What should remain inside the conversation?
The challenge wasn’t making everything visible.
It was making the right things visible.
5. Mobile and Desktop Behaviors Were Different
Participants often read email on mobile but completed work on desktop.
The solution needed to support continuity across devices without requiring users to recreate context each time they switched environments.
This became one of the most important design constraints.
6. Intent Cannot Be Guessed
The system cannot reliably determine what a user intends to do.
An email that looks like it needs a reply might be informational. One that appears complete might require follow‑up later.
Automatically assigning states would introduce errors, erode trust, and create yet another thing users would need to monitor and correct.
The only reliable path was to let users declare their intent with a single, low‑effort action — and then ensure that declaration survived across time and devices.
What We Learned
The challenge wasn’t helping people keep track of more things.
It was reducing the number of things they needed to carry in their heads.
The final direction succeeded not because it added more organizational tools.
It succeeded because it reduced dependence on personal memory while remaining lightweight enough to fit naturally within email workflows.
IMPACT & THINKING
What changed in behavior
The most important change wasn’t in the interface.
It was in how people managed unfinished work.
Before, participants relied heavily on personal workarounds to bridge the gap between reading and acting.
They read emails while commuting, between meetings, away from their desk.
The message was seen. The action was postponed.
The challenge was staying aware that action was still needed.
The work itself wasn’t always difficult.
Keeping track of it was.
Before: Follow‑Up Depended on Personal Workarounds
The original workflow provided a way to read email.
It provided very little support for returning to it later.
As a result, participants developed their own reminder systems:
Leaving emails unread
Flagging conversations
Starring messages
Reopening emails repeatedly
Carrying unfinished intentions in their head
Questions such as What was I planning to do here? Why did I leave this unread? Which emails still require action? Did I already deal with this? often required reopening conversations just to answer.
Reading was easy.
Returning was uncertain.
After: Intention Remains Visible
Once follow‑up intent became visible, participants approached deferred work differently.
They no longer needed to rely on unread status as a personal reminder system.
Instead, unfinished work remained visible after the moment of reading.
Participants could leave a conversation knowing that the reason for returning later — whether it was Save for later, Follow‑up, Waiting on, or Completed — would still be available when they came back.
Several participants described feeling less dependent on their own recall throughout the day.
Not because they were reading fewer emails.
Because the system was now carrying more of that awareness for them.
"I don't need to leave everything unread anymore."
— Support team member,
post‑validation interview
"I know why I saved this for later."
— Operations specialist,
post‑validation interview
Neither comment focused on the interface itself.
Both focused on reduced dependence on memory.
What Actually Changed
The inbox stopped functioning as a personal reminder system.
Unread status no longer carried the burden of representing unfinished work.
Participants still postponed emails when they needed to.
But they no longer had to reconstruct their original intention from scratch each time they returned.
The cost of interruption became lower.
The effort required to resume work became smaller.
And because those interruptions happen dozens of times each day, the impact accumulated quickly.
Takeaway
People weren’t struggling to read email.
They were struggling to preserve intention after reading it.
The redesign didn’t eliminate interruptions.
It reduced the cognitive effort required to recover from them — and that small shift changed how people moved between reading, postponing, and completing work throughout the day.
What this problem revealed
At the beginning of the project, the problem appeared to be mobile productivity.
People repeatedly described situations where important emails were read on mobile but addressed much later.
The assumption seemed reasonable: mobile devices make acting on email difficult.
The research revealed something different.
Most participants could already read, understand, and prioritize incoming messages.
The breakdown happened afterward.
Not during reading.
During the gap between reading and acting, where the system left no trace of intention.
The Real Challenge Wasn't Access
Modern communication tools are extremely effective at delivering information.
People rarely struggle to access email.
They struggle to maintain awareness of unfinished work after that information has been consumed.
The email existed.
The intention slowly disappeared.
Over time, participants compensated by creating personal reminder systems:
unread emails, flags, stars, notes, self‑forwards, mental checklists.
None of these behaviors were part of the original purpose of email.
They emerged because people needed a way to preserve intent — a job the inbox itself refused to do.
Information and Intention Are Different Things
One of the most important insights from the project was that reading and acting are separate activities.
Most email products are designed to support information access.
Far fewer are designed to support intention preservation.
As a result, the moment of reading often becomes disconnected from the moment of action.
People understand what needs to happen.
They simply lose awareness of it later — because the tool provides no persistent signal.
Memory Shouldn't Carry Workflow
Throughout the research, a consistent pattern appeared:
whenever users repeatedly invent personal systems to keep track of something, memory is compensating for a product gap.
The problem wasn't discipline.
The problem wasn't attention.
The problem was that the workflow depended entirely on the user to carry intention across interruptions, notifications, meetings, and conversations — and that load becomes unsustainable as the day accumulates.
A Broader Product Principle
The project ultimately changed how I think about productivity systems.
Good products don't simply help people access information.
They help people maintain awareness across time.
Because work rarely happens in a single uninterrupted moment.
It happens across interruptions. Across devices. Across contexts. Across hours. Sometimes across days.
The most effective systems aren't the ones that help users remember more.
They're the ones that reduce how much users need to carry in their heads in the first place.
Closing Reflection
This project began with a question about mobile email.
It ended with a question about human memory.
The challenge wasn't helping people read messages.
The challenge was helping unfinished intentions survive long enough to become completed work.
