Research
A method that turns noticing into evidence.
Anyone can spot a broken experience. What makes it useful to a business is a repeatable path from that observation to a decision someone is willing to fund.
The nine stages
Each stage has a defined input, a defined output, and a reason it can't be skipped.
Skipping synthesis produces anecdotes. Skipping prioritization produces backlogs nobody sequences. Skipping measurement produces opinions.
- 01
Observation
Unstructured time in the environment. No hypothesis yet — the goal is to see the real sequence of events, including the parts nobody reports.
- 02
Documentation
Field notes written in a fixed structure: what happened, who was involved, what it cost, and the participant's own words.
- 03
Pattern recognition
Comparison across visits. A behavior only becomes a finding once it recurs independently across people, days and departments.
- 04
Affinity mapping
Statements are clustered bottom-up until themes stop shifting. Named themes are the unit of work from here on.
- 05
Journey mapping
Themes are placed onto customer and associate timelines so each one has a moment, an owner and a measurable state.
- 06
Prioritization
Impact against operational effort. The bias is toward interventions testable in one location within a quarter.
- 07
Concept generation
The smallest design that could plausibly move the measure, expressed concretely enough to critique.
- 08
Pilot
A time-boxed test with an explicit hypothesis, a control, and a pre-agreed definition of failure.
- 09
Measurement
Results read against the leading measure first, then the lagging one. Findings are published either way.
Synthesis in practice
What affinity mapping looks like when it works.
Raw statements on the left of the process become named, defensible themes — each one traceable back to the observations that produced it.
Findability & adjacency
Heavy items & carts
Locked merchandise
Inventory truth
Prioritization
Sequencing is a design decision, not an afterthought.
Plotting opportunities forces the honest conversation about what an operation can actually absorb this quarter.
Principles
Four rules I hold the work to.
Verbatim over paraphrase
Interpretation is a later step. Compressing too early destroys the finding.
Recurrence over volume
Forty coded observations beat four hundred uncoded ones.
Structural over personal
If the analysis lands on “someone should try harder,” it isn't finished.
Testable over comprehensive
A narrow pilot that can fail is worth more than a program that can't.