“I have a number. What does it mean?”
Without comparison or a shared reference, performance became an abstract number. Feedback arrived late or without enough context.
← Back to projects
Numbers
Performance ClarityCase study
Numbers
with meaning.
I turn scattered records into one verifiable result. Employees immediately understand how they are doing; managers see why, and the conversation can begin with a concrete next step.
JavaScriptData automationData validationData UXResponsive UI
P Performance Clarity
DSNS
REFERENCE PERFORMANCE412 items/h
SECOND METRIC38 units/h
# / PROFILETimeItems/hUnits/hResult
01 Profile 047.42448 108.7%42 110.5%109.6%
02 Profile 116.88431 104.6%39 102.6%103.6%
03 Profile 075.95407 98.8%37 97.4%98.1%
04 Profile 026.21382 92.7%34 89.5%91.1%
Shared viewSame result. Same reason.
✓Traceable logicFrom source to decision
14 daysrelevant context, rather than a blind average
2 rolesemployees and managers sharing the same facts
4 stepsfrom a raw record to a decision
1 answerwhat the result means and what to do next
Why this project existsThe numbers existed.
The numbers existed.
Confidence did not.
“How am I doing today?” was a simple question without a simple answer.
Employees saw individual values without context. They did not know what a good result was, where they stood during a shift or what to focus on. Managers could access data but lacked a clear way to show and explain it.
“I know where the data is. I cannot show it quickly.”
Every explanation started with searching, filtering and combining values. Preparing the data consumed time intended for supporting people.
Data → decisionFour steps
Four steps
from a question
to an answer.
01Ask the right questionClarify what the person needs to know to act
02Validate and merge dataCombine identity, time and metrics into one valid record
03Explain the resultCompare with relevant history and show the reason for a deviation
04Recommend the next stepTurn findings into a concrete, understandable decision
01 / SCATTERED VALUES02 / VALIDATION & MATCHING03 / SHARED CONTEXT04 / DECISION-READY VIEW
LOADED0 / 10 valuesbut no answer to “How am I doing?”
IDENTITYWho and where
TIMEWhen and for how long
METRICSWhat can be compared
Profile 04
448 items/h
Process flow A
7.42 h
42 units/h
Day window
12. 08. 2026
Source B
109.6%
Reference range
Identity and work windowmerged
Incomplete or zero valuesisolated
Reference group and weightscalculated
01Profile 04109.6%
02Profile 11103.6%
03Profile 0798.1%
Behind the interfaceMore than another report.
More than another report.
It was a better conversation.
01 / Product discovery
The real problem was meaning getting lost between the system and the person.
The initial request could have led to another spreadsheet. Instead, I analyzed the whole feedback moment: who needs an answer, how they assemble it, where uncertainty arises and what they must understand in a few seconds.
01 / IDENTITYReliably link the same person across exports with different structures.02 / COMPARABILITYCreate a fair reference despite differences in time worked.03 / EXPLAINABILITYShow the result and the specific reason it changed.
A calculation that can be repeated, checked and explained.
I split the logic into separate layers. Input errors cannot silently become a colored score, and every final value has a traceable path back to its source.
01INGESTStandardize schemas and identifiers
02VALIDATEChecks for time, completeness and outliers
03COMPUTEWeighted reference and final score
04EXPLAINContext attached to every result
I designed the system for times when the data is imperfect.
An empty day, zero time, a late record or an outlier must not silently change the result. Every unusual state has its own rule and a clear interface response.
#N/A exclude0.00 h do not calculate23:48 assign to the correct windowOUTLIER flag and explain
Before & afterSame data.
Same data.
An entirely different way of working.
report_final_v7.xlsxA B C D E04 382 6.7 ??? #N/A11 417 7.1 39 08/1207 344 — 34 night
Open sourceshistory_14days.csvshift_today.csvnotes_v2old_report?
New message
“Is 382 a good result?”
awaiting an answer · 12 min3 sources · 3 different structures
which value is valid?
3 conflicting values
- 5 places to search
- Calculation without explanation
- The conversation starts with waiting
ProcessProcess 1
Reference412 / 38
Work windowDay
01Profile 04109.6%
02Profile 11103.6%
03Profile 0798.1%
- Everyone understands their result
- The manager shows the same context
- The conversation starts with the next step
← CHAOSCLEAR OVERVIEW →
↔
My roleWhat this project says
What this project says
about me as a creator.
I can find the real problem, even when it is hidden behind “make us a report”.
I first mapped the entire process: employees needed to understand their results quickly, while managers had to find the right period, check record quality and build an explanation. I removed unnecessary steps and designed a shared view that shows both sides the same result, context and reason.
QUESTION→VERIFIED DATA→CLEAR CONTEXT→DECISION
Every iteration had to remove a specific uncertainty.
Each prototype tested a different risk: the data flow, understanding the result and usability during real operations.
The result is transparent.
The interface can show a value’s source, transformation rule and reason for inclusion in a comparison. Trust comes from a traceable path from record to decision.
SOURCE14 valid work windows2 incomplete records isolated
→RULEWeighted by actual timethe same logic for every profile
→OUTPUT96% with traceable contextdetails one step away
The best dashboard is the one that leaves no need to ask: “What does this mean for me?”
The interface first shows the answer needed at that moment: the current state, its meaning and a possible next step. Calculation details, historical values and supporting data follow for deeper review.
This progressive hierarchy reduces data overload without hiding anything important. Employees can quickly understand their results, while managers can expand and verify them as a basis for a conversation.
Project impactFrom searching for values
From searching for values
to a faster decision.
Faster understanding
Relevant performance values are prepared in one ordered view, without browsing an export.
Consistent comparison
Every result uses the same historical reference and comparison method.
Feedback without guesswork
Employees and managers see the same result, reason and supporting data. The conversation starts with what to do next.
Problems visible earlier
Deviations are no longer hidden in a daily summary. The overview shows which part of the result caused them and whether they represent a one-off fluctuation or a recurring pattern.
Interactive demo
See how the data behaves in practice.
The demo uses realistic anonymized data and preserves the real calculation logic, processes, shifts and status indicators.