Umair Yahya
All work

I designed Sendoso’s address validation so every failed check had a next step.

Sendoso, 2021. Address Confirmation, launched in 2020 before my work, asks a gift’s recipient to confirm or correct their own address. Address Validation, which I designed in 2021, checks an address before a gift ships and decides what happens when the check fails. I connected the two into one system.

SendosoT-shirt for Ben

    What the happy path doesn’t show

    • Recipient choosesLetting Ben choose the options turns address confirmation on.
    • Address confirmationBen confirms his own address before it ships.
    • Couldn’t validateEvery failed check gets a next step, not a red error.
    • Send anywayGuarded, with the rejection and restocking fee spelled out.
    • Bulk sendsOne bad row never stops the campaign.

    Every branch from sender to delivery, mapped before any screen.

    Send the gift, then answer as Ben. Switch the check off for the other path.

    Role
    Senior Product Designer and the design owner, in a design and research sprint from mid-May to early June 2021.
    I owned
    The sender and recipient journey for the Address Confirmation expansion, and the complete original Address Validation design, including the clickable prototype the study ran on.
    Worked with
    Kelly Hoover and Marijan Baranjec, Product Managers; Yume Sato, Product Designer; Grace Tran, UX Researcher, who led the usability studies; Matt Samson, UX Research Team Lead; and engineering.
    Shipped
    The patterns are in production, per Sendoso’s public documentation. Verified.

    The problem was what happened when an address was wrong.

    Address Confirmation was fast-tracked in March 2020, when offices emptied and gifts were still going to obsolete office addresses. A year later the problem was broader: validate data before a send, explain errors without blocking legitimate addresses, and recover safely when the system could not reach a reliable answer.

    I mapped sender controls, validation rules, recipient recovery and fulfillment as one journey before committing to any screen. The map settled three stages, each with a way forward:

    • The sender provides dataOne address, a bulk CSV, or a choice the recipient makes.
    • The platform evaluatesAccept a valid address, suggest a correction, or flag a record it cannot resolve.
    • The journey recoversEdit or deliberately send anyway, trigger Address Confirmation, or let the recipient correct it.
    The working flow board: every branch from sender input to fulfillment, mapped before any screen.
    The email sets expectations before asking for anything.
    The landing page carries the same message into the address form.

    Six decisions about where the complexity should live.

    Each says what I chose, what I turned down and who decided it with me, and each has a working piece beside it.

    Surface the complexity instead of hiding where it moved

    Chose
    When a sender lets the recipient choose gift options, such as a size, Address Confirmation turns on automatically. I chose to explain that at the moment of choice, so the sender knows what the recipient will be asked before the gift goes out.
    Instead of
    Hiding the rule and surprising the recipient later. Explaining it cost the sender one more line to read.
    Decided with
    Marijan Baranjec, who owned the product decision; I represented the experience and the system logic.

    A model of the sender’s setting and what the recipient sees.

    The shipped design: the option says it enables address confirmation, and the settings show it on.

    Validation as a routing system, not a red error message

    Chose
    Every result has a route. A reliable suggestion is shown beside what was entered. With no reliable suggestion, the sender keeps agency: edit, request Address Confirmation, or a guarded Send anyway. If confirmation is already on, the system does not block twice.
    Instead of
    A single error that stops the send and leaves the fix to support or manual outreach.
    Decided with
    Kelly Hoover, Product Manager for Address Validation.
    Address Confirmation already on

    A model of the routes in the original Address Validation design.

    One bad row never stops the campaign

    Chose
    For bulk sends, process valid recipients and hold the ones that need help. Managers can correct records inline, re-upload a CSV, or send confirmation requests to the recipients themselves.
    Instead of
    Rejecting the whole file until it is clean, which punishes every valid recipient for one typo.
    Decided with
    Kelly Hoover.

    Sample data, not real recipients.

    Let the service’s speed decide the pattern

    Chose
    Design both patterns and let the service’s real speed choose. A fast response supports inline correction. A slower one needs pending states, work that can be resumed, and a clear split between valid and unresolved recipients.
    Instead of
    Committing to one moment, on upload, on Send or in the background, before knowing the speed, with an interface that implied a certainty the service did not have.
    Decided with
    Kelly Hoover.

    A model of the two patterns. Measured response times were not kept with the files, so none are shown.

    A position written down before every test

    Chose
    There was no usability-testing practice at Sendoso yet; with the research team I helped build one. I co-wrote hypotheses for 13 sender questions and 8 recipient questions, each recording a position before testing, and built the fully clickable prototype the study depended on. Testing ran on Userbrain.
    Instead of
    Judging the results after the fact, when it is easy to remember having expected them.
    Decided with
    Grace Tran, who led the studies, and Yume Sato, Product Designer, who ranked the riskiest questions with me.

    Summarized from my notes, as written before testing.

    The screen behind the first hypothesis: a message for each gift.

    Research reduced three disclosure ideas to one production rule

    Chose
    One rule for production: show the product name and photo, or hide both. Company branding was essential for trust when asking for an address, and removing every product visual weakened anticipation.
    Instead of
    Keeping all three options configurable. The cut cost the generic surprise illustration and the separate landing-page message: less flexibility, and easier to configure and maintain.
    Decided with
    The team, on the evidence from the study; Marijan Baranjec owned the product decision.

    Simplified from the three tested emails. The real ones are below.

    The three tested layouts. They changed several things at once, which confounded the comparison.

    The system logic held up better than parts of the prototype.

    What held up

    • The sender and recipient journeys were modeled together. Product first treated them as separate tickets; the journey map and the study showed that sender choices changed recipient trust.
    • Validation failures were designed as routes, not dead ends.
    • The study produced concrete scope cuts, not only preferences.

    What did not

    • The long sender page weakened scannability, and the preview link had too little emphasis.
    • The three email variants changed disclosure, layout and action at once, so the comparison was confounded.
    • The prototype asked for a National ID on a US shipping form: a data-minimization failure on a trust-critical form.
    The complete sender page: five sections on one page. Its length made hierarchy the central problem.

    The patterns became part of Sendoso’s address framework.

    • VerifiedSendoso’s documentation shows individual invalid-address prompts with edit, request confirmation and Send anyway.
    • VerifiedInvalid or unverified CSV recipients receive Address Confirmation forms instead of stopping the campaign.
    • VerifiedChoosing a product variant automatically requires Address Confirmation.
    • VerifiedThe first feature usability test at Sendoso came from this project.
    “Umair not only helped me brainstorm for, but also mount and launch the company’s first usability test for its features.”
    Grace Tran UX Researcher at Sendoso

    What I can claim

    The 2021 reframe, the sender and recipient journey, the complete original Address Validation design, prototyping, research planning and handoff. Not the 2020 launch, later releases, or customer results as outcomes of my design.

    What I would do differently

    Vary one factor at a time in the email test, counterbalance the order, and measure comprehension before preference. Ask only for the data a shipment needs. And keep the measured response times with the design files.

    What it taught me

    Automation moves the work; it does not remove it. Every rule the system applied on a sender’s behalf had to be explained to someone, at the moment it mattered to them.

    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