The interface was doing the damage
A model, process, or policy problem was often amplified by a surface that hid the state or consequence.
THIRTY STUDIES · NINE SECTORS · ANONYMISED
Real products. Real constraints. One evidence standard.
Each study connects the product failure to the interaction that changed it. You can see the role, timeline, team, baseline, outcome, interface details, design decisions, research artefacts, and client-evidence boundary.
THE LIBRARY30
studies
543334422
NINE CONTEXTS
The product problem changes the interface grammar.
AI trust, sales behaviour, attribution, consumer choice, enterprise operations, credit, care, degraded systems, and procurement do not deserve the same template.
THE FULL LIBRARY
Thirty arguments. Thirty craft records.
Filter by context or read in sequence. Every card opens the complete study.
AI TOOLS · US
An AI agent that shows its work before it acts, not after.
78%AGENT RETAINED @ WK 4from 11%
94%ACTIONS REVIEWED BEFORE SENDby design
3.4xPERMISSION SCOPE GRANTEDvoluntary escalation
CONFIDENCE 82%
Action preview ready
AI TOOLS · NETHERLANDS
Where the model is wrong, and how the interface says so.
31CLAUSES REVIEWED / HOURfrom 18
6%FLAGS DISMISSED AS WRONGfrom 34%
CONFIDENCE 82%
Action preview ready
AI TOOLS · INDIA
Retrieval you can audit, for a regulator who will ask.
3 weeksTIME TO COMPLIANCE APPROVALfrom indefinite
100%ANSWERS WITH FULL TRACEby construction
CONFIDENCE 82%
Action preview ready
AI TOOLS · MUNICH
The prompt is not the interface. The workflow is.
340ASSETS SHIPPED / WEEKfrom 84
1.4REVISION ROUNDS PER ASSETfrom 4.1
CONFIDENCE 82%
Action preview ready
AI TOOLS · US
Onboarding an AI product is teaching a mental model, not a feature tour.
31%ACTIVATION @ SESSION 1from 9%
44%DAY-30 RETENTIONfrom 12%
BUYING GROUP
SALES AUTOMATION · US
A sequence builder that shows the human on the other end.
5 stepsMEDIAN SEQUENCE LENGTHfrom 11
4.8%REPLY RATEfrom 1.9%
BUYING GROUP
SALES AUTOMATION · INDIA
Data enrichment where the source is always one click away.
8%RECORDS DISCARDED ON ONE BAD FIELDfrom 61%
2,400FIELD CORRECTIONS / WEEKfrom 40
BUYING GROUP
SALES AUTOMATION · MUNICH
A CRM view built around the buying group, not the contact.
41%SINGLE-THREADED DEALS IDENTIFIEDof open pipeline
74%MULTI-THREADED @ CLOSEfrom 29%
BUYING GROUP
SALES AUTOMATION · US
Call intelligence that produces a decision, not a transcript.
81%MANAGER WAUfrom 14%
2.4 / callACTIONS TAKEN FROM CARDfrom 0.1
MODEL CONFIDENCE
Paid searchHIGH
EventsLOW
UnknownDECLARED
MARTECH · US
Segment building for people who do not write SQL.
84%SEGMENTS BUILT SELF-SERVEfrom 6%
4 minTIME TO FIRST SEGMENTfrom 11 days
MARTECH · NETHERLANDS
Attribution that states what it cannot know.
QuarterlyUSED IN BUDGET DECISIONSfrom never
22%UNATTRIBUTABLE, STATEDpreviously distributed
4.1 / 5CMO-REPORTED CONFIDENCEfrom 1.4
MODEL CONFIDENCE
Paid searchHIGH
EventsLOW
UnknownDECLARED
MARTECH · INDIA
Campaign QA that catches the mistake before the send.
0.3BROKEN SENDS / MONTHfrom 14
1,840UNRENDERED VARIABLES CAUGHTpre-send, 6 mo
YOUR FITM / REGULAR
Confidence 86%
D2C & COMMERCE · INDIA
Size selection as a conversation, not a chart.
-38%FIT-RELATED RETURNSvs baseline
74%FIT-FINDER COMPLETIONvs 6% size chart
YOUR FITM / REGULAR
Confidence 86%
D2C & COMMERCE · US
Subscription management that does not punish pausing.
4.1%NET MONTHLY CHURNfrom 9.0%
58%PAUSE CHOSEN OVER CANCELof intent-to-leave
YOUR FITM / REGULAR
Confidence 86%
D2C & COMMERCE · MUNICH
A checkout for a market that does not trust checkout.
51%CHECKOUT ABANDONMENTfrom 79%
44%INVOICE PAYMENT SHAREof completed orders
SHIFT / SATURDAY
08:00Coverage 92%
16:002 open
00:00Covered
ENTERPRISE INTERNAL · US
Shift scheduling that respects the person being scheduled.
-22%VOLUNTARY TURNOVERfirst two quarters
-41%UNFILLED SHIFTSvs baseline
SHIFT / SATURDAY
08:00Coverage 92%
16:002 open
00:00Covered
ENTERPRISE INTERNAL · INDIA
Performance review software that produces a conversation.
91%REVIEWS CITING EVIDENCEfrom 12%
2.8 hrsCALIBRATION MEETING LENGTHfrom 6.2
SHIFT / SATURDAY
08:00Coverage 92%
16:002 open
00:00Covered
ENTERPRISE INTERNAL · NETHERLANDS
Access requests that expire by default.
-68%STANDING PRIVILEGED GRANTSvs baseline
4%RECERTIFICATION BULK-APPROVEfrom 94%
DECISION TRACEAPPROVE
5 factors · 2 overrides
NBFC & LENDING · INDIA
A credit decision the underwriter can explain to the borrower.
11%OVERRIDE RATEfrom 38%
0.81OVERRIDE CONSISTENCY (INTER-RATER)from 0.34
DECISION TRACEAPPROVE
5 factors · 2 overrides
NBFC & LENDING · INDIA
A collections interface designed against harassment.
-78%REGULATORY COMPLAINTSvs baseline
+18%RECOVERY RATE · 90+ DPDvs baseline
NBFC & LENDING · INDIA
A loan application built for the agent, on the borrower's doorstep.
14%FILES REJECTED AT PROCESSINGfrom 34%
3.1 daysTIME TO DISBURSALfrom 8.4
96%FIELDS COMPLETED ON FIRST VISITfrom 61%
DECISION TRACEAPPROVE
5 factors · 2 overrides
NBFC & LENDING · US
Loan terms a small business owner can actually compare.
71%CHOSE LOWEST TRUE-COST OFFERfrom 24%
-31%EARLY DEFAULT (FIRST 90D)vs baseline
ALARM PRIORITYNeeds action
Room 408 · 02:14
HEALTHCARE · US
An alarm system that earns the attention it demands.
46ALARMS PER PATIENT / DAYfrom 180
-34%RESPONSE TIME · TRUE POSITIVEvs baseline
ALARM PRIORITYNeeds action
Room 408 · 02:14
HEALTHCARE · NETHERLANDS
Referral triage that tells the GP what happens next.
-64%DUPLICATE REFERRALSvs baseline
82%REJECTIONS CORRECTED ON FIRST RETRYfrom 21%
ALARM PRIORITYNeeds action
Room 408 · 02:14
HEALTHCARE · US
Medication adherence without shame mechanics.
+41%ADHERENCE · LOWEST QUARTILEvs baseline
84%RETENTION AFTER FIRST MISSED DOSEfrom 38%
ALARM PRIORITYNeeds action
Room 408 · 02:14
HEALTHCARE · MUNICH
Informed consent that leaves the patient actually informed.
88%RANDOMISATION COMPREHENSIONfrom 34%
91%WITHDRAWAL-RIGHT RECALLfrom 41%
DEGRADED MODE
3 sources delayed
DEFENCE & GOVTECH · US
A common operating picture that degrades gracefully.
0STALE-POSITION MISREADSin evaluation
2.1 secTIME TO ASSESS PICTURE CONFIDENCEfrom 47 sec
DEGRADED MODE
3 sources delayed
DEFENCE & GOVTECH · NETHERLANDS
A permit application a citizen can finish without a consultant.
79%APPLICATION COMPLETIONfrom 32%
18%APPLICATIONS NEEDING REWORKfrom 61%
EVALUATION / OPEN
Price
Risk
Delivery
PROCUREMENT · MUNICH
Supplier evaluation where the scoring is visible to everyone scoring.
0AWARD CHALLENGES UPHELDin 14 months
automaticJUSTIFICATION PACK ASSEMBLYfrom 3 weeks
EVALUATION / OPEN
Price
Risk
Delivery
PROCUREMENT · US
The request form that decides which process you are in.
6%MISROUTED REQUESTSfrom 44%
11 daysMEDIAN CYCLE TIMEfrom 27
No studies match this filter.
WHAT HELD ACROSS THE LIBRARY
The surface changed. Six principles kept returning.
These are positions we would defend in a product review, regardless of sector or aesthetic.
Show the uncertainty
Confidence bands, provenance, disagreement, and boundary increased trust more than false certainty.
Move the decision to the context
The best next action lived beside the evidence, exception, or person it affected.
Delete the punitive mechanic
Recovery, pause, decline, correction, and exit were designed as product features.
Design the worst condition first
Weak signal, peak volume, partial data, permission mismatch, and human stress exposed the real system.
Agency is information
People need to know what happened, what changes next, who owns it, and how to reverse or escalate it.
STUDIES
Bring us the product problem that deserves its own argument.
We will start with the evidence, not a visual reference board.
Start a Product