← Selected work
GovTech B2B service design 0 → 1 concept Evidence → pilot

A clearer path through Government services

MyBusiness Plan turns a maze of agency transactions into one living plan—so a first-time owner can see what to do, why it matters and what is blocking the next step.

Scope note: This is a hypothesis-led design proposal prepared for the IMDA Senior UX Designer assessment. It is not a shipped Government service, and the targets shown below are pilot measures—not claimed outcomes.

The product idea

One plan across many agencies. Each agency stays the source of truth.

Orchestration layer
built on GoBusiness
01PlanIn progress
02Set up entityACRA · CorpPass
03Check premisesURA / HDB
04PrepareDocuments
05ApplySFA · GoBusiness
06OperateIRAS · CPF
My role
Lead Product Design
Service strategy → prototype
Priority user
First-time micro or small business owner
Service surface
GoBusiness extension, not a new portal
Proposed pilot
Food & beverage business setup

Why this problem

The form is rarely the hardest part. The sequence is.

Singapore has invested heavily in digital Government services. The remaining opportunity sits between them: understanding which transaction comes next, which decision unlocks it and who owns the delay.

A first-time F&B owner does not think in agency boundaries. They think, “Can I open my bakery?” Yet their path can span entity registration, digital authorisation, premises approval, licensing, inspection, tax and employment obligations.

Each service may work well on its own. The business still carries the orchestration work: interpreting rules, sequencing tasks, moving data, chasing status and recovering when something goes wrong. That hidden coordination cost lands hardest on small firms with no dedicated operations team.

“This is not a search problem. It is a confidence problem: am I doing the right thing, in the right order, with the right evidence?”Design framing for the assessment

Business impact

Delay

A missed dependency can push back rent-paying opening dates, hiring and first revenue.

Service impact

Rework

Incomplete or mistimed applications create avoidable contact, clarification and resubmission.

Public value

Trust

People judge Government as one system—even when responsibility sits across several agencies.

Evidence & validation

Strong signals. Clear limits.

The public evidence is enough to justify discovery, not enough to pretend the answer is already known. I separate what is documented from what the team still needs to learn.

14

Government touchpoints were once needed to set up an eatery.

Historical · GovTech GoBusiness case study
845 → 90

Fields after a previous GoBusiness simplification effort.

Historical · proves simplification can move the needle
10–14

Days saved in the same Government redesign.

Historical · not a forecast for this concept
12

Steps currently documented for a Food Shop Licence.

Current · SFA application guidance
Evidence boundary: The first three figures describe an earlier GoBusiness redesign. I use them to show that cross-agency friction has been material and addressable—not as proof that the same gains would repeat.
Focus the evidence:
Documented

People can still need clarification

GovTech reported that, even after launch, business owners asked for clearer regulatory guidance. SFA’s current process also contains rejection, resubmission and response deadlines.

Documented

The ecosystem is already large

GoBusiness brings together more than 120 Government e-services and has supported more than six million transactions. Improving orchestration can scale without launching another destination.

To validate

Sequence drives avoidable delay

I would test whether owners start the wrong task, wait unnecessarily or miss a dependency—and quantify the time and rework attached to each failure mode.

To validate

A shared plan improves confidence

The concept assumes a goal-led plan can reduce uncertainty without oversimplifying policy. Prototype testing should challenge that assumption before integration work begins.

Current-state handoffs

Where the business becomes the integration layer

EntityRegister the businessACRA
AccessAuthorise staff or an agentCorpPass
PremisesConfirm that the proposed use is allowedLicence readiness
LicencePrepare documents, apply and respondSFA · GoBusiness
OperateSet up tax and employment obligationsIRAS · CPF

People & priorities

Design for the owner. Include the people who help them.

I would prioritise the first-time owner because they carry the highest uncertainty and have the least organisational support. Delegates and service officers remain essential: they often keep the journey moving.

Primary

First-time owner

Goal: open on time without becoming a regulatory expert.

Pain: unclear order, duplicated evidence, silent waiting and fear of making an expensive mistake.

Secondary

Employee or agent

Goal: complete assigned work with the right authority.

Pain: blurred ownership, overshared access and no clean hand-back.

Service ecosystem

Agency officer

Goal: receive complete, policy-compliant applications.

Pain: avoidable contact, missing context and inconsistent status language.

PriorityNeedWhy nowDesign response
1Next safe actionKnow what can be done now and what must wait.Wrong sequence creates the costliest downstream rework.Dependency-aware plan with a clear reason for every block.
2Shared progressSee status, ownership and due dates in one place.Reduces chasing without changing agency decision rights.Normalised statuses with links to the source service.
3Safe delegationAsk for help without sharing full account access.Many small firms rely on staff, agents or family.Task-level permission, expiry and audit trail.
4Data reuseAvoid re-entering information Government already holds.Valuable, but higher in privacy and integration complexity.Progressively introduce consented prefill after trust is earned.

End-to-end journey

Following Mei from idea to opening day

Mei is opening a 12-seat bakery. The journey makes dependencies visible before they become surprises and keeps the original agency transaction one step away.

01

Plan

Answer a short set of questions about business type, premises and timing.

Moment: “Is this plan relevant to me?”
02

Entity

Register the business and establish who can act for it.

Dependency: UEN before downstream access
03

Premises

Check land-use conditions before committing to fit-out or licence work.

Moment: confidence before spending
04

Prepare

Collect layout plans, tenancy evidence and food-safety documents.

Risk: unclear evidence creates rework
05

Apply

Submit, respond to requests and prepare for inspection.

Risk: deadlines and status ambiguity
06

Operate

Move from licence approval into tax, hiring and recurring obligations.

Moment: setup becomes ongoing compliance

Service blueprint

The experience is only as good as the operating model behind it

Layer
Plan
Apply
Recover
Owner sees
Relevant pathTasks, order and estimated effort
One progress viewOwner, status, due date
Clear next moveWhat changed and how to resolve it
MyBusiness Plan
Rules engineBuilds a contextual plan
Status adapterNormalises agency updates
Exception routingKeeps context with the case
Agency systems
Policy sourceRules and eligibility remain owned by agencies
Transaction sourceSubmission and decisions stay in existing services
Resolution ownerAgency handles policy-specific exceptions

UX direction

Orchestrate. Do not duplicate.

The concept extends GoBusiness with a goal-led plan. It does not recreate every agency service or hide policy complexity behind false simplicity.

01

Start with the business goal

Translate “open a bakery” into a relevant, editable path—not a catalogue of agencies.

02

Show the next safe action

Make dependencies and reasons visible. Never present a disabled task without explaining what unlocks it.

03

Keep truth close

Show rule owner and last-updated date, then link to the authoritative transaction.

04

Design for shared work

Support narrow, time-bound delegation with explicit ownership and an audit trail.

Competing needs

SpeedandAccuracy

Block irreversible actions when evidence is uncertain; keep planning lightweight.

PersonalisationandPrivacy

Ask only what changes the path; make data reuse consent-specific and reversible.

SimplicityandEdge cases

Progressive disclosure for exceptions, with a clear assisted path.

ConsistencyandAgency ownership

Standardise task and status language, not policy decisions.

Representative screens

The plan answers three questions at a glance

What should I do next? Why now? Who owns it? The screens favour action hierarchy and trust cues over a dense dashboard.

Concept screen 01 · Business overviewDesktop
01 · HierarchyOne dominant action, followed by task context—not four competing CTAs.
02 · TrustRule owner and recency sit beside the advice, where doubt happens.
03 · RecoveryBlocked is paired with the reason and what unlocks the task.

Navigation

Goal-led, not agency-led

Plan, Applications, Documents and People map to the owner’s mental model.

Consistency

One status grammar

Complete, waiting, blocked and upcoming always include owner, reason and next move.

Accessibility

Meaning beyond colour

Status text, visible focus, 44px targets, plain language and reduced-motion support are built in.

Delivery strategy

Earn the right to integrate

I would resist starting with a costly data-sharing programme. First prove that better sequencing solves a valuable problem; then add integration where it removes measured friction.

0

Baseline the problem

Interview 12–15 recent or active F&B founders, shadow five setup journeys, analyse GoBusiness search/support themes and map failure demand with service officers.

Gate: repeatable sequence failures with material time or rework
1

Test a read-only plan

Prototype the six-stage plan with eight target users. Use static rule content and deep links; no systems integration needed.

Gate: ≥80% find the correct next action unaided; no critical comprehension errors
2

Pilot shared status

Run a four-week F&B pilot with a small cohort and selected agency partners. Normalise a limited set of statuses and track handoffs.

Gate: less rework and assisted contact without loss of status accuracy
3

Add consented data reuse

Prefill only high-frequency, low-ambiguity fields. Show source, purpose and a review step before anything is submitted.

Gate: measurable time saved, high correction accuracy and no material privacy concerns

Measurement

Measure confidence in action—not page views

North-star outcome

Businesses complete the right setup tasks, in the right order and on time, without avoidable assisted support.

User outcomes

  • ≥80% find the correct next task unaided in prototype testing
  • Confidence after key handoffs
  • Time to understand and recover from a blocked task

Business outcomes

  • −15% setup delays attributable to missed dependencies
  • Time from first plan to licence-ready submission
  • Owner-reported coordination effort

Service outcomes

  • −25% incomplete or avoidable resubmissions
  • Assisted contact per active plan
  • Exception resolution time across agencies

Guardrails

  • >99.5% task/status accuracy in the pilot
  • Accessibility defects by severity
  • Consent withdrawal, correction and complaint rate

Targets are hypotheses. I would set final thresholds only after the baseline study. A fast journey with the wrong advice is a service failure, so accuracy and comprehension remain release guardrails.

Lead Product Design practice

The hard part is not the dashboard

A coherent cross-agency experience needs shared decisions, content ownership and an escalation path. My design role would include shaping that operating model—not stopping at screens.

Shared outcomeAgree one journey-level measure across product, policy, operations and agency teams.
Decision rightsGoBusiness owns orchestration patterns; agencies own policy, eligibility and decisions.
Content governanceNamed rule owners, review cadence, freshness signals and a rapid correction path.
Design governanceReusable task/status model, accessibility acceptance criteria and cross-service critique.
Risk managementStart read-only, log provenance, audit exceptions and expand integrations progressively.

Open questions

What I would investigate next

Which failures create the most business cost?

Quantify delay and rework by failure mode: wrong sequence, missing evidence, unclear status, expired response window or authority/access issue.

How much status consistency is feasible?

Map agency status taxonomies, latency and exception states before promising a single progress view. Test the normalised language with both business owners and service officers.

Where does personalisation become advice?

Define policy and legal boundaries for goal-led guidance, including uncertainty language, provenance, recency and human escalation.

What should never be delegated?

Co-design task-level permissions with security, CorpPass and business users; identify high-risk actions that require the owner’s explicit approval.

Sources

Public evidence used in this proposal