Skip to content
railsengines.com

Chili vs Split

Chili and Split both sit in experiments, but they solve different parts of it.

Thin comparison — they do not replace the same product. Kept readable, left out of search.

Split

Better known

Rack-based A/B testing with a mountable dashboard

No ratings yet
# Gemfile
gem "split"

# config/routes.rb
mount Split::Dashboard, at: "/split"

What Chili does

The spicy feature toggle framework for Rails Have you ever wanted to test out a new feature on only a subset of users? Did that implementation end up being lots of if/else statements embedded in the main code? If so, Chili can help. 214 stars on GitHub · MIT licensed · maintained by balvig. Source: <https://github.com/balvig/chili

Read the full Chili listing →

What Split does

The long-running Rack A/B testing framework, with a dashboard you mount. Experiments with any number of alternatives Multiple goals per experiment Weighted alternatives, and forced variants for QA A dashboard to start, stop and reset experiments and pick a winner Redis-backed Split stores in Redis and has been around longer; Field Test stores in your database and reports Bayesian probabilities. If you want experiment data joinable with the rest of your schema, take Field Test.

Read the full Split listing →

Why you might pick it

  • It also replaces VWO.
  • It has substantially more adoption — 2,700 stars against 214.

Side by side

Rows where the two differ are marked.

Chili vs Split compared across category, licence and coverage
Chili Split
Category Experiments Experiments
Replaces (differs) VWO
License MIT MIT
GitHub stars (differs) 214 2700
Gem (differs) chili split
Mounts at (differs) /split
Member rating

What each one gets off your invoice

Both replace

Nothing unique here.

Only Chili

Nothing unique here.

Only Split

Questions

What is the difference between Chili and Split?
Chili and Split both sit in experiments, but they solve different parts of it.
Can I mount both Chili and Split in the same Rails app?
Technically yes — engines mount at different paths — but they overlap, so running both usually means two sets of tables doing one job. Pick one.

Other comparisons in Experiments