Turning an ambiguous brief into a north star vision, a phased roadmap, and organisation-wide alignment.
Sometimes the most important design work isn't designing anything at all.
When Sedex asked me to look at search across their global B2B SaaS platform, the brief was deceptively simple: "What can we do to make search better." But an audit quickly revealed that search hadn't just degraded, it had fragmented. Every team had built their own version, resulting in inconsistent user experiences across patterns, terminology and behaviours, duplicated engineering effort, increasing complexity to maintain and extend functionality, siloed data alignment across Product, Engineering and Design domain pods, and no shared ownership or long-term strategy.
There was no single owner, no shared strategy, and no agreed direction. There was no single owner for search, so the work relied entirely on influence and alignment across product, engineering, data, and support teams, without formal authority over any of the surfaces I was working across.
What was being asked for was a UX fix. What was actually needed was a strategic platform decision. Those are very different problems. The goal was not to design a new search interface. It was to make the problem clear and decision-ready.
Outcome
• Established a single, shared search strategy across previously siloed teams
• Reduced duplicated effort by creating shared principles and foundations
• Created clarity on where and how to invest in search over time through a phased roadmap
• Gave leadership a decision-ready path for investing in search
• Shifted senior leadership mindsets from viewing search as a minor feature, to recognising it as a core platform capability
What I did
• Operated as the sole design expertise, leading early-stage discovery to redefine search from a fragmented feature into a platform capability
• Synthesised user research, platform constraints, governance gaps, and market trends into a clear product vision
• Made explicit trade-offs between incremental improvement and long-term scalability
• Defined a phased roadmap aligned to technical feasibility and future AI readiness
• Defined search experience principles, models, and patterns to guide consistent implementation
• Aligned Product, Data, Support, and Engineering teams around a shared direction. Without formal ownership of any of them
Here you are seeing the presentation commnicated across the company.
I led a structured discovery phase across four dimensions:
Reviewed qualitative research and usage patterns across different search implementations. Identified where users abandoned or reformulated searches. Mapped user intent behind common search behaviours.
Interviewed PMs, engineers, support staff, Ethical Trade Coordinators, and operations teams. Audited existing search solutions, constraints, and dependencies. Uncovered duplicated work and conflicting priorities across teams.
Partnered with engineering to understand indexing, performance, and scalability limits. Assessed the feasibility of shared infrastructure versus continued local optimisation.
Analysed how modern platforms were evolving search toward semantic and task-oriented discovery, drawing on examples from Google, Notion AI, GitHub Copilot, and Jira. Identified emerging user expectations around AI-assisted search and what that would mean for Sedex's platform readiness.
The most important finding wasn't about search patterns or UI consistency. It was about intent:
The platform's main users (Buyers, Suppliers, and Customer Support Teams) were not searching for content.
They were searching to complete tasks.
Search was being treated as a passive input/output function. But users expected it to be context-aware, proactive, and integrated into their workflows.
That single reframe, from search as a feature, to search as a task-completion layer, unlocked a fundamentally different direction and created the alignment that had been missing.
I defined a vision for search as a Search Intelligence Layer that would:
• Understand user intent and context
• Surface relevant actions, not just results
• Work consistently across all product domains
• Scale with future AI and semantic capabilities
The goal was not to immediately rebuild everything. It was to establish a clear north star that could guide incremental investment. Giving teams direction without forcing a risky, all-at-once transformation.
Three key trade-offs shaped the direction:
• Incremental unification vs. full rebuild
Chose incremental unification to reduce risk (cost, tech debt, and time), and allow teams to continue delivering value while moving toward a shared foundation.
• Local optimisation vs. shared capability
Prioritised shared foundations over local autonomy, even where some teams initially resisted the loss of independent control. The long-term coherence of the platform depended on it.
• Advanced AI features vs. data readiness
Deliberately deferred complex AI use cases until data quality and governance had improved, and until costs could be properly allocated for the platform storage increases required. Advanced capabilities were positioned as longer-term options, not near-term promises.
These decisions balanced long-term platform value with near-term feasibility, and made the roadmap credible to both engineering and leadership.
Working with Product and Engineering I defined a phased roadmap from the current state to the end vision over two to three years.
Milestones were defined around learning and risk reduction, not just shipping. Each had a clear purpose: deliver some value, learn something important, or make the next decision easier.
My thinking was early milestones focused on alignment and quick win improvements in the most visible problem areas. Mid-stage milestones introduced shared components and datasets once there was confidence they would not slow teams down. Later milestones remained flexible, allowing more advanced ideas to be explored only when the foundations were first in place.
This thinking shaped the roadmap introduced as a "Now-Next-Near-Future-Later" :
Phase 1: Establish shared principles and patterns. Address the highest-impact inconsistencies across the platform.
Phase 2: Introduce shared indexing and search components. Reduce duplicated logic across teams.
Phase 3: Layer in contextual and semantic search capabilities as data quality improves.
Phase 4: Enable proactive, task-oriented search experiences. The full realisation of the Search Intelligence Layer vision.
Search touched every team but was owned by none of them. Progress depended entirely on alignment rather than authority.
I brought product, engineering, data, and support together to agree on the core problems search needed to solve and the constraints we were all working within. By creating shared principles and documenting decisions clearly, teams could move forward consistently without waiting for central approval. This reduced fragmentation and gave leadership the confidence to make investment decisions.
After the direction was agreed, teams did not pause delivery or rebuild search from scratch. The strategy was used to guide ongoing work instead.
Teams began aligning new search changes to the shared direction, which immediately reduced duplication and conflicting approaches. Quick improvements were made where the problems were most visible leading to an improved "Member search" functionality with behaviour and consistent search bar patterns across the platform. While engineering began exploring shared foundations that could be adopted gradually.
Most importantly, search stopped being treated as an ad hoc feature and became a recognised platform capability. That shift made it easier for teams and leaders to decide where to invest, where to hold back, and why.
Success here wasn't about shipping a new search UI. It was about making search easier to build, easier to improve, and more useful for users.
We measured this through estimated reduction in duplicated effort, faster task completion (less click throughhs) for users, and clearer ownership and decision-making around search investment. These measures made it possible to judge whether continued investment was worthwhile, and gave leadership a basis for ongoing prioritisation.
This project reinforced three things I now hold as principles for any cross-cutting platform work.
First, platform capabilities fragment easily when ownership is unclear. Without a shared strategy, teams optimise locally and create long-term complexity, even with entirely good intentions. The solution is making ownership and direction explicit before anyone writes a line of code.
Second, framing problems in terms of user intent rather than features unlocks alignment faster than anything else. Reframing search around tasks, not results, gave every team a shared language and a shared test for whether a decision was the right one.
Third, in cross-cutting initiatives, progress depends less on having the right solution and more on creating clarity, trust, and decision-ready options. The organisation doesn't need certainty. It needs enough confidence to take the next step.
