Optimizely has been reimagining the future of progressive delivery and experimentation, improving every part of the platform to empower technical teams to build, ship, and iterate faster. Learn about the latest enhancements to Optimizely Full Stack and the Optimizely Data Platform, and get a sneak peek at the upcoming roadmap.
5. How can we align progressive delivery
and experimentation?
6.
7. Key Concepts
Flag: a juncture in code where a decision is made
Rule: conditions for what to deliver and why
Ruleset: the full set of Rules for to a Flag
Variation: the payload a Flag delivers
8.
9. Flags always deliver a variation
● Experiments & Targeted Delivery have the same “shape”
○ Simplest variations: On and Off
○ Even the simplest On / Off toggle is an opportunity to experiment
● Allows for seamless transitions between Targeted Delivery and Experiment
Experiment
Rollout
10. Interacting with Flags
● No breaking changes
○ Migrate to Flags without codebase updates or SDK upgrades
○ Standalone Experiments => Flags with Rules
● Coming Soon: Decide
○ Provides a single, unified implementation
decide(flag_key)
Run an A/B test or
MAB
Turn feature on for
specific users
Customize a
feature by
audience
11. Understand the impact of
changes to your deployed
Flags, and use that data in
other places where it will be
useful.
Governance Automation Monitoring
Build workflows and processes
around Flag deploy,
management, and cleanup.
Flags as a
Foundation
Control access and ownership
of Flags, and define oversight
and change management
policies.
13. Dynamic Flexible Extensible
When new information
becomes available, the event
pipeline should handle
asynchronous, in-place updates
to existing datasets.
Actionable insights come in all
shapes and sizes. Events
should not conform to rigid
structures, and neither should
the consumption patterns of
those events.
The data used to run statistical
models should be the same
data used to create
business-level KPIs.
Data as a
Product
14. Enriched Events Event-level dataset containing all records
you send to Optimizely
Enriched with experiment and user
attributes from the server
With a partitioning scheme that serves
more of your use cases
15. Enriched Events Analyze experiment results with any SQL
or data analysis tool
Join Optimizely data with other data
sources in your data warehouse
Monitor experiment impact using your own
dashboards
16. Flag Monitoring Is my flag configured correctly?
Where is my flag evaluated?
How often is my flag evaluated?
17. Stream Service Create filtered data streams using a
generic subscription API
Consume and use Optimizely data in real
time
Make product decisions faster and more
efficiently
18. Event Debugger Inspect and validate your event
instrumentation
Uncover missing events
Ensure the right attributes are being sent
with your events