How visiting a factory helped me BUILD BETTER SOFTWARE
Woody Rousseau
CTO & Co-Founder @ Sipios
30
100
40
20
5
Web
Mobile
Data
Fintech
Summary
- The software development open space : a factory
- Standardizing Work
- Kaizen and training
- Results
- Next Steps
The software development open space
A Factory
From Ideas
To Reality
- Always building the same parts
- Always building the same amount
- 640 different parts
- Takt : From 20 min to 2 months
- Client request increased from 4 to 40 pieces per day for a production line
A Production line
Production Step
Quality Check
Machine Assisted
Quality Check
Assembly
Final Check
Quality insurance
A Software line
Production Step
Testing
Code Review
Merging and Deploying
Integration Test
Matching
Production Step
Production Step
Bring Value
=
Matching
Testing
Don't Bring Value but...
standard due to human limitation
=
Machine Assisted
Quality Check
Matching
Bring value
=
Assembly
Merging and Deploying
Matching
Waste : quality issues passed to another developer, rework
=
?
Code Review
Digging
Less than one day
Production Step
≠
Production Step
Less than 20 minutes
We're ages away from what is done in a factory!
The Scrum sea
Speed : 7
Inspection : Daily
"Customer demand" changes once a week
The dream sea
Inspection : Every 20 min or less
"Customer demand" should be able to change every day
15 min
6 min
4 min
Standardized WORK
Breaking stuff up
User Story :
As a customer, I search canal boats by price
Breaking stuff up
User Story :
As a customer, I search canal boats by price
Creating a
model
Creating a
ressource
Testing the ressource
Creating the service
Custom price filtering logic
Daily planning : splitting the user story in “tasks”
Gathering data
Creating a
model
Creating a
ressource
Testing the ressource
Creating the service
Custom price filtering logic
Gathering time and commentaries for high performing developer
12:04 min
//
05:34 min
//
24:12 min
Realised it was smarter to test the ressource mocking the service and then the service itself
> 20 min
1h24 min
Was interrupted by business developer asking an obvious question
Outlier
45 min
Took some time figuring out the steps
Custom
Gathering data
Gathering time and commentaries for high performing developer
Defining standards
Filtering outliers and too long tasks to define standards after several measurements
| Average time excluding outliers | Count | |
|---|---|---|
| Calling an external service | 00:09:29 | 4 |
| Creating a client | 00:13:00 | 1 |
| Creating a domain | 00:03:23 | 8 |
| Creating a json mock | 00:04:00 | 1 |
Defining standards
Filtering outliers and too long tasks to define standards after several measurements
| Average time excluding outliers | Count | |
|---|---|---|
| Calling an external service | 00:09:29 | 4 |
| Creating a client | 00:13:00 | 1 |
| Creating a domain | 00:03:23 | 8 |
| Creating a json mock | 00:04:00 | 1 |
Kaizen & TRAINING
Kaizening
| Causes | Count |
|---|---|
| Forgotten task during planning | 13 |
| Interruptions | 13 |
| Lack of technical training | 10 |
| Code naming issue | 6 |
Using commentaries and go & see to find root causes. For us:
Kaizen opportunities !