Approach
A hard system has fixed rules and a measurable output. The eligibility criteria are set and an application either completes or it doesn’t, so from the inside, nothing looks ambiguous. Soft logic is the part you can’t read off the system.
A form works exactly as specified and people still abandon it at step four, and neither the rules nor the analytics tell you why. Finding out means watching somebody fill it in and asking what they thought it wanted. The method is qualitative, which is where the name comes from, and it’s what actually moves completion rates.
Operating rules
What produces an outcome is often not a screen. The training and the handoffs around it decide results the interface cannot reach, so they get designed in the same pass rather than left to whoever inherits them.
Participants get selected for how likely they are to fail, not for how easy they are to reach. A convenience sample is mostly people the system already works for, which is why it so reliably confirms what the team expected.
Most contracts name the deliverables before any research happens, so by the time findings arrive, the scope is already signed and they can only rearrange what's inside it. The statement of work here leaves room for findings to change what gets built.
Strategic workflow
01.
Time on site watching the system in use before asking anyone about it, then sessions that begin unstructured and narrow as specific claims emerge worth disproving. Notes are coded for the moments where somebody hesitates, backs up, or routes around the system rather than through it.
Methods
Contextual inquiry, in-depth interviews, participant observation, thematic coding, competitive and category review
Deliverables
Research plan, coded transcripts, findings report, prioritized friction inventory
02.
Mapping what has to happen behind the interface for the front to work, including who holds which information and at which handoff it stops moving. Most failures a user experiences were created several steps upstream, which is why the blueprint comes before any decision about the interface itself.
Methods
Service blueprinting, journey mapping, stakeholder mapping, front-stage and back-stage analysis, gap analysis
Deliverables
Service blueprint, journey map, failure point register, prioritized opportunity areas
03.
Turning the blueprint into structure. Which decisions sit together, where a sequence cannot be avoided, and what recovery looks like when a step goes wrong. The artifacts here are wireframes and flows, so the review is about whether the logic holds rather than how it looks.
Methods
Information architecture, user flows, wireframing, card sorting, tree testing, content modeling, accessibility audit
Deliverables
Sitemap, annotated wireframes, user flows, content model, accessibility requirements
04.
Voice, type, color, and the rules for applying them, specified against the architecture rather than ahead of it. The system has to survive being used by people who were never on the project, so the rules for extending it are written alongside the rules for applying it.
Methods
Brand strategy, brand architecture, positioning, messaging frameworks, visual identity, design systems, creative direction
Deliverables
Positioning and messaging framework, visual identity, brand guidelines, component library, master files
05.
Building and testing in one loop, with each round returning to the same participants where possible so a response can be compared against one already on record. Failures are documented alongside what shipped, because a pattern that didn’t work is the cheapest one to inherit and the most expensive to rediscover.
Methods
Low and high fidelity prototyping, moderated and unmoderated usability testing, A/B testing, design QA, accessibility testing
Deliverables
Interactive prototypes, test plans, round-by-round findings, final patterns with failures recorded
06.
The first weeks of real use measured against what testing predicted, with revisions where the two disagree. Live behavior produces information no session can, because the people arriving have not agreed to be studied and are not trying to be helpful.
Methods
Analytics review, funnel analysis, session recording, post-launch research, iteration planning, team training
Deliverables
Measurement plan, post-launch findings, revision set, training materials, documentation
Copyright © 2026 HCM Strategy
