Most UX research ends as a forgotten deck. Here's how to turn research into living design intelligence that actually shapes product decisions in SEA markets.
Somewhere in your organisation, there is a Confluence page titled something like “User Research — Q3 2024 Insights.” It has 47 slides. Nobody has opened it since the readout meeting.
This is not a storage problem. It is an architecture problem — and it costs brands real product quality.
The Deck is a Dead End, Not a Deliverable
Sen Lin’s piece in UX Collective puts a sharp point on something most design teams feel but rarely name: research that terminates in a presentation has no downstream value. The moment the meeting ends, the insights are archaeologically frozen. A new product question emerges six weeks later, and instead of querying existing knowledge, the team commissions another round of research — duplicating effort, burning budget, and almost certainly re-discovering the same user pain points with slightly different slide aesthetics.
The analogy from data engineering is exact: a deck is a one-time extract, not a pipeline. It answers one question, at one point in time, for one audience. Compare that to a well-structured data warehouse, where the same underlying facts can be queried, recombined, and surfaced against new questions indefinitely. Research should work the same way. The raw observations, tagged user quotes, and behavioural patterns should persist in a structured, queryable form — not trapped inside a linear narrative that made sense in October but is opaque by February.
For Southeast Asian product teams, the stakes are amplified. User behaviour across markets like Indonesia, Thailand, and Vietnam shifts fast — driven by platform dynamics on Shopee and Grab, seasonal commerce spikes, and genuinely different trust signals between demographics. Stale research doesn’t just slow you down; it actively misdirects you.
What a Living Research Artifact Actually Looks Like
Lin’s approach — using Claude Code to interrogate research transcripts and turn them into structured, reusable artifacts — points toward a practical architecture that design teams can implement without a data engineering budget.
The core principle: separate observations from interpretations from recommendations, and store each layer independently. Observations are raw (a user hesitated on the checkout screen for 11 seconds before abandoning). Interpretations connect observations to patterns (hesitation correlates with unclear shipping fee disclosure). Recommendations are contextual and time-bound (add a shipping estimate earlier in the cart flow for Q4). Conflating these three layers is exactly why decks age poorly — the recommendation dominates, and the underlying evidence that could validate or invalidate it later gets buried or lost.
Implementation starting point: build a simple tagging taxonomy in Notion, Airtable, or a Dovetail-equivalent before your next research sprint. Tag every observation by user segment, journey stage, platform (app vs mobile web vs desktop), and market. This takes roughly 90 minutes of setup and transforms three months of accumulated research from a narrative into a database. When a stakeholder asks “what do we know about cart abandonment on Android in Tier 2 cities,” you have an answer in minutes rather than a gap where an answer should be.
The Creative Friction Argument for Imperfect Research Processes
Robert Beatty’s interview with It’s Nice That — ostensibly about designing album covers for Boards of Canada — contains a genuinely useful design principle for research practitioners: the value of deliberate friction in creative process. Beatty’s aesthetic emerges precisely from working against the grain of clean digital tools, embracing artefacts and degradation as signal rather than noise.
Applied to UX research: the instinct to smooth research into a polished deck is itself a form of signal loss. The hedged comment a participant made before backtracking, the moment of confusion that the facilitator noted but didn’t probe — these imperfections are often where the real insight lives. When research is forced into a clean narrative for executive consumption, friction gets edited out. The living artifact model preserves it.
This matters especially when designing for the genuine diversity of Southeast Asian users. A research process that flattens a Filipino urban professional and a rural Javanese first-time e-commerce user into a single persona is doing active harm to product decisions. The friction of holding contradictory data points in tension — rather than resolving them into a tidy slide — is what produces honest, market-specific design intelligence.
For teams using AI tooling to synthesise research (as Lin describes), this is a concrete implementation caution: prompt your model to surface contradictions and outliers, not just consensus themes. The consensus is usually what the team already assumed. The outliers are where the product opportunities hide.
Connecting Research Architecture to Design System ROI
There is a direct line from living research artifacts to design system maturity — and ultimately to conversion rates and development velocity. When research is structured and queryable, pattern recognition compounds over time. A component library built on six months of accumulated behavioural observation is categorically more defensible than one built on a single research sprint and a designer’s intuition.
Shopee’s product teams, for instance, operate across eight Southeast Asian markets with meaningfully different UI conventions and user mental models. The design decisions that work for a Singaporean power user expecting information density actively alienate a Vietnamese first-timer who needs progressive disclosure and trust signals at every step. Managing that complexity requires a research infrastructure that can be queried by market, segment, and context — not a deck that averaged everything into one regional persona.
The business case for investing in research architecture is straightforward: Forrester estimates that every dollar invested in UX returns between $2 and $100, depending on the application. The variance in that range is largely explained by whether insights actually reach product decisions at the moment they’re relevant. A living research artifact is the infrastructure that closes that gap.
Timeline and resource consideration: transitioning from deck-first research to artifact-first research takes two to three sprint cycles to feel natural. The upfront cost is taxonomy design and team alignment on tagging conventions. The compounding return begins around month four, when the second wave of research starts building on the first rather than duplicating it.
Key Takeaways
- Treat each research sprint as an additive layer to a structured knowledge base — not a standalone deliverable — so insights remain queryable when new product questions emerge.
- Separate observations, interpretations, and recommendations at the tagging level; conflating them is what makes research decks expire on contact.
- For Southeast Asian product teams specifically, preserve market and segment granularity in your research taxonomy — averaging across markets produces personas that describe no actual user accurately.
The deeper question this raises: if research is genuinely designed as infrastructure rather than communication, who owns it? Design, product, data, or some function that doesn’t quite exist yet? The organisations figuring out that answer are the ones whose product decisions will compound most cleanly over the next three years.
At grzzly, we work with digital teams across Southeast Asia to build the strategic and operational infrastructure behind great products — including the research systems, design frameworks, and data pipelines that turn good intentions into compounding competitive advantage. If your team is producing quality research that isn’t consistently shaping product decisions, that’s a solvable architecture problem, not a culture problem. Let’s talk
Sources
Written by
Chunky GrizzlyDesigning the foundational plumbing — data warehouses, lakehouse models, and ETL pipelines — that separates organisations with genuine intelligence from those drowning in dashboards.