Coherent · 2022
Insurance tooling for non-engineers
Designed Coherent Spark's authoring surfaces so actuaries could ship product changes without an engineering ticket.
Role
Product Designer
Users
Actuaries · Product owners
Region
APAC, EMEA
Photo block
Insurance tooling for non-engineers — hero visual
2400 × 1350 px · JPG/PNG
TL;DR
- Translated spreadsheet-native workflows into a guided authoring experience.
- Cut product-change cycle from quarters to days for pilot customers.
- Reduced support escalation rate by ~40% during pilot.
02 — Scope
What I owned on Coherent
Research, product design and authoring patterns for the Coherent Spark tooling — respecting the actuary's spreadsheet mental model.
Embedded research with actuarial teams across APAC to learn what could change about the spreadsheet metaphor — and what had to stay.
Contextual inquiry
On-site sessions with three actuarial teams across APAC to map real authoring rituals.
Co-design workshops
Tested which review and approval patterns were welcomed versus actively resisted.
Prototype benchmarking
Quantified error rate against free-form editing to validate guided authoring.
Pilot analysis
Tracked product-change cycle time and support escalations through customer pilots.
03 — Context
The spreadsheet was the product
Actuaries lived in Excel. Coherent's pitch was to keep that intelligence but free it from manual deployment. The design problem was respecting the existing mental model while introducing version control, review and observability.
04 — Research question
What we needed to learn
What can we change about the spreadsheet metaphor without breaking the actuary's confidence — and what must stay identical?
05 — Research design
Methods, phase by phase
Contextual inquiry with three actuarial teams across APAC.
Co-design sessions to test which review and approval rituals were welcomed vs. resisted.
06 — Key findings
What the research surfaced
Cell-level trust is non-negotiable
If a cell changed without an audit trail, actuaries lost trust in the whole tool.
3 contextual inquiries
Review rituals must look familiar
PR-style review felt alien; spreadsheet-style change tracking landed instantly.
Co-design workshops
Guided authoring cut error rate by ~35%
Field-level guidance reduced authoring errors compared to free-form editing.
Prototype benchmark
Cycle time dropped from quarter → days
Pilot customers shipped product changes in a release window measured in days, not quarters.
Customer pilot data
07 — Design decisions
Finding → intervention
08 — Outcomes
Measured impact
09 — Reflection
Looking back
What worked
Respecting the spreadsheet mental model instead of fighting it.
What I'd improve
More embedded sessions with non-pilot customers earlier — pilots self-selected for tolerance.
Key learning
In expert tooling, familiarity is a feature.
Stakeholder skill
Selling 'guardrails' to actuaries who had never accepted them before.
Next case study
Healthcare