Small Batches Sizes

Batch is a way to group some actions/things in a group

Prefer to work in small batches.

Prefer to release to production a small batch of issues, and small batch of code to avoid creating a lot of regressions or problems.

Small batch are easier to review.

Prefer work in small batch when writing, write a bit and get early feedback to make sure you are going in the right direction

Machine learning algorithm use a small rate of learning (batch), to make sure it converge to the right solution, if the rate is too big it will go in the wrong direction

Small batches generate a fast Feedback Loop .

How to Take Smart Notes

WIP Written

Principles of Product Development Flow

---

The benefits of small batch sizes

Reducing your batch size you reduce the cycle time.

Cycle time is the time you take from an issue/task/feature ideation to go to production/live.

The small the batch size the small is your cycle time, and fast you improve your product/service.

The faster you ship new features

The faster you fix bugs.

The faster you add improvement

Reducing your batch size reduces variability of flow.

If one of your sprint has a batch size of 10 issues and the next one has a batch size 60 issues, which one will you deliver faster? If you keep your batch size small you always keep the same flow of delivering.

Reducing batch size accelerates feedbacks

If you have a small batch size you deliver fast and get feedback faster. This will make sure you are building the right thing or find bugs or errors faster. Your errors/failure will also be smaller.

Reducing batch size reduces risk

It is much safe to deliver a few new items/changes then many of them.

Instead of doing one off release of a new feature, try to release it incrementally.

Incrementally you can adjust to changes in requirements, you can adjust to errors during the releases, and you can adjust to feedback.

Reducing batch size reduces overhead

Large batches sizes increases the average number of bugs. If you have 300 open bugs, when you find bug number 301, we need to recheck the other 300 for duplications. For small batch sizes it would have many less number of test to retest if a new bug was found.

Large batches reduces efficiency

You can think of large chunk of work by an individual is more efficiency, but this is only local as reduce the feedback loops and overall efficiency. The individual could put some wrong assumptions and have to redo most of the work done.

Large batches lower motivation and urgency

Large batches dilutes responsibility, as if I’m responsible to deliver 1 modules from 100 modules, I feel much less urgency. It also make more probable of someone delaying one of those modules. It takes to much time to see your work in production and get the feedback from user.

Large batch size increases cost and schedule growth

It is hard to estimate schedule and cost for large batch sizes.

The batch size is limited by its worst element

the batch size will take as long as its worst element, 1 item will delay and make all other cost more.

Good infrastructure enables small batches

If you have CI/CD with tests and deploys all automated, it makes it easier to handle small batches and do many releases on a daily basis

Feature flags can reduce batch size

Feature flags let you deliver new feature faster for a few customer, reducing risks, accelerating feedback loops.

To reduce batch size reduce scope.

Decide to deliver less to deliver faster and in time.

To do wait 100% of the requirement to start designing and coding. these requirement will always change a long the way.

Converging fast to Solutions

Reducing your Batch Sizes