Umair Yahya
All work

I rebuilt Sendoso’s navigation around two contexts: finding things, and working on one.

Sendoso, 2022. The study began as my own and reached the roadmap after I presented it to leadership. Testing and feedback sent my first structure back; working with the team, I rebuilt it around two contexts before a beta reached 15 customers.

Narrow the list, open a campaign, then open the menu at each width.

Role
Senior Product Designer on a three-month project in 2022, from research to handoff. I was the hands-on designer throughout; the direction was set with the team.
I owned
The information architecture and navigation structure, the responsive design system from 1536px to 320px with every state drawn, and the clickable prototypes the studies depended on.
Worked with
Once the study reached the roadmap: the Design Manager, the VP Design, the UX Research Team Lead, the Head of Product, a UX Researcher and a Product Manager. Engineering, through a design-review-to-code-approved path.
Shipped
A feature-flag beta to 15 customers on June 15, 2022. Verified.

I created the mandate before I redesigned anything.

This did not start as an assigned redesign. Sendoso is a B2B sending and gifting platform, and by its Series C it reported more than 800 customers and 20,000 users. One left rail had to carry more modules, more roles and more frequent tasks. Users described four problems:

  • The wrong optionsSome were unnecessary, while options people needed were missing.
  • The wrong orderThe arrangement did not reflect how often things were used.
  • Too much spaceThe navigation took up a large amount of usable screen.
  • Not responsiveIt did not work on smaller screens or on mobile.

Those pointed at information architecture, task context and responsive behavior at once, so a cosmetic fix would have kept the original assumptions. I broke the existing architecture and the roadmap into discrete actions, studied how people understood them, and presented the findings to leadership. The work was accepted onto the roadmap.

The plan then ran in three phases: research and analysis; requirements, user stories and use cases; then wireframes, the design system and interface design across five sprints. That order kept the visual work behind the architecture.

The product I started from: every module hung off one left rail.
The structure underneath the complaints: browsing, details and actions all arrived in the same place.

Three decisions that set the structure.

After the study reached the roadmap, each was made with the team named above. The piece beside each one works.

A card sort with 69 people, read as evidence rather than a menu

Chose
A card sort across three kinds of participant, so no one audience stood in for the others: 22 design and product experts, 12 power users and 35 new users.
Instead of
Designing the menu from the roadmap alone. The study did not produce a menu; it produced evidence about which actions people saw as related, pieces the team still had to assemble.
Decided with
The roadmap team. A review of Reachdesk, Alyce, Postal.io and Snappy separated category conventions from Sendoso’s own decisions; it was an input, not a vote.

The real sample and the nine groupings. Each view shows who took part, not how that group sorted.

From the synthesis board: the behaviors and dashboard surfaces the new framework had to serve.

Two contexts instead of one route through the product

Chose
Two contexts. In browse, people find and narrow: global navigation across the top, filters down the left, the set in the middle. In focused, they understand and act on one campaign or object: an overview, then its details.
Instead of
Defending a detailed model. The first direction moved people into a browse path or a build path; I took it from low-fidelity to detailed wireframes, with default and minimized states, and testing and feedback sent it back to the drawing board.
Decided with
The roadmap team, after testing and feedback. I stayed the hands-on designer and produced the final wireframe.

Both structures come from the working file. Open a screen to read it full size.

The first model’s minimized state, with campaign creation as a full-screen step beside a preview.

A responsive system that made the structure buildable

Chose
A system specified from 1536px down to 320px, with the wordmark reducing to the S mark as space tightened, and every state drawn: idle, hover, focus, active, disabled and error. Cards had defined aspect ratios and truncation rules.
Instead of
Leaving small screens and edge states for engineering to improvise. Not working on mobile was one of the four reported problems, so the small end had to be designed, not inferred.
Decided with
The roadmap team, and engineering, who held every component to the same path: design review, design approved, coded, code approved.

A model of the principle; the line limit is illustrative. The navigation at each width is at the top of the page.

Specified at both ends of the range, with the review path every component passed.
Three of the six states drawn for a filter pill, each with its token names.

A beta reached 15 customers before I left.

The designs were approved and handed to engineering, and a feature-flag beta launched to 15 customers on June 15, 2022.

In the campaign list that went to the beta, budget against spend, sends, team, health and time left sit in one row, so a campaign running low shows up as status, not as a surprise later.

What went to the beta: low inventory and low budget show up as status in the campaign list.

What the work shows, and where my evidence stops.

  • VerifiedA study I started on my own was accepted onto the roadmap after I presented it to leadership.
  • VerifiedA card sort with 69 participants, synthesized into nine groupings.
  • VerifiedA feature-flag beta to 15 customers on June 15, 2022, with the responsive system specified from 1536px to 320px.

What I can claim

The mandate, the research, the structure and its rethink, the responsive system, the approved handoff and the beta. Not adoption, retention or satisfaction: the full rollout was planned for fall 2022, after I left. The findings that triggered the rethink are not in my records, so this page shows the documented sequence.

What I would do differently

Test the structure before detailing it. A tree test of the new groupings needs nothing more than a list of labels, so a rethink would cost days instead of a detailed model. And file test findings with the design files.

What it taught me

Detail is not proof. Work can move backward in visual completeness when the structure needs to move forward in clarity, and a mandate can be earned rather than waited for.

Verified on the record: a live product, a public page or a project document. Internal test measured by me or the team, outside production. Reported someone told me; I have not seen the data.

Building something harder than it looks?

Open to Principal and Lead Product Design roles, in Dubai or remote.

umairyahya48@gmail.com

LinkedInResume