Platform · Feature flags
Feature flags and gradual rollouts
A feature flag is a release that is not a test. Turn a change on for a share of your traffic, widen it on a plan, and turn it off in one click if something is wrong.
How a flag runs
From a draft to fully on
Raising the share only ever adds people.
Two kinds of flag, a rollout you widen by hand or on a plan, and who gets it before everyone else.
- Code flags your server or your page asks for, each side with an optional JSON payload
- Editor flags: changes made in the visual editor that go live as a release, not a test
- Any share from 0 to 100% by hand, and Turn off as the kill switch
- Rollout plans of up to ten steps that only ever widen
- Sticky: nobody who has the change on loses it as the share grows
- A holdback of up to 50% that always stays off, so a comparison group remains
Two kinds
Flag the code, or flag a change made in the editor
A code flag is whatever your code does when it is on: the server SDK, the decide endpoint or the browser asks for it by key and gets on or off, with a payload if you set one. An editor flag carries changes made in the visual editor, applied on the page like a variant, live straight away and with a kill switch and a schedule.
- Read a code flag on the server, over HTTP or in the browser
- A JSON payload on each side, for a value rather than a switch
- A reminder when an editor flag has been fully on for 30 days, to build it into the site and archive the flag
Rollouts
Widen it on a plan, stop it in one click
Set the share by hand, or lay out a plan of up to ten steps, each with a minimum time or a date it may not start before, and the scheduler moves it on. Pause and resume it whenever you like: a resumed plan never narrows. Schedule a draft to publish, or a live flag to turn off, at a set time.
- A change made by hand pauses a running plan
- Publish and turn off on a schedule
- Notified when the scheduler moves a flag, finishes a plan or turns one off
Who gets it
Decide who gets it before everyone else
A visitor has to be eligible first, by the same URL, device, cookie and activation targeting your experiments use, or by attribute rules for a code flag on the server. Rings come next: up to ten groups with a fixed side, such as staff and a beta list on and key accounts off. Everyone else falls into the rollout.
- Rings match on attribute rules or a list of up to 1,000 IDs, and the first match wins
- Server rules for code flags, matched on the attributes your SDK call passes
- A holdback set before publishing, always off, even at 100%
Kept apart
A release, not an experiment
Flags stay out of everything that reads experiments: the library, the statistics, Learnings and the monitor. They live in their own list under Experiments, from draft to running to ended, and an ended flag serves off for good. An editor flag still counts in the overlap checks, because it changes pages your experiments run on.
- Their own list, with a kind column and the next step
- Every change on the flag page, in its activity log
- Billed in sessions, like experiments, on every plan
Everything in the box
The full feature set
Experimentation
- Visual multi-variant tests
- URL redirect tests
- Server-side tests
- Server SDK
- No-code visual editor
- AI editor
- Chrome extension launcher
- Component library
- Dynamic content
- Data feeds
- Custom CSS and JS
- Replace, insert, append, before and after
- URL targeting on seven conditions
- Device targeting
- Cookie targeting
- Activation targeting
- Traffic weights per variant
- Automatic weight rebalancing
- Concurrent tests without splitting traffic
- Variant screenshots
- Duplicate an experiment
- Test runs before launch
- Sticky variant assignment
- Anti-flicker rendering
- One-click winner deployment
- Implementation hand-off
- Hypothesis tracking
- PXL prioritisation scoring
- Run length plan
- Pipelines of staged experiments
- Library search, filter and results
- Library timeline
Statistics
- Frequentist results
- Bayesian results
- Lift and confidence intervals
- Probability variant is best
- Expected loss
- Sequential testing
- Multiple-comparison correction
- Holm-Bonferroni, Benjamini-Hochberg and Šidák
- Sample-ratio-mismatch alarms
- Primary and secondary goals
- Revenue per variant
- Average order value per variant
- Multi-currency
- Power analysis
- Minimum detectable effect
- Days-remaining projection
- Combination reporting
- Per-variant time series
- Plain-English explainers
- Ask about this experiment
Monitoring
- Health monitor
- Attention queue across accounts
- Program report
- Experiment debrief
- Scheduled starts and stops
- Stop on significance
- Minimum runtime guard rails
- Trust gates on broken data
- Collision detection
- Interaction detection
- Overdue and no-data alarms
- Email alerts
- Slack alerts
- In-app alerts
- Per-event preferences
- Activity log
- Client roster view
- Saved console layouts
- GA4 and Matomo tracking
- Roles and permissions
Voice of customer
- On-site NPS
- CSAT
- CES
- Custom surveys
- Guided survey builder
- Score, choice and text questions
- Custom texts and scale labels
- Page and device targeting
- Share of visitors
- Page-view and delay triggers
- Exit-intent triggers
- Scroll triggers
- Event triggers
- Four widget positions
- Frequency capping
- Per-question results
- Responses over time
- AI sentiment analysis
- Surveys inside experiments
- Results by variant
- Answers tied to variant
- Validated responses
- Free-text themes
- Response export
- Loaded only where a survey runs
Pertento is a conversion-rate-optimisation platform for websites and online stores, built for both in-house teams and CRO agencies running experimentation programmes across a client roster.
Integrations
Works with the stack you already run
One snippet drops into any site, or go fully server-side. Nothing to re-architect.
One-line snippet
Paste one tag in your head and you are live: a 0.9 KB loader, no build step, no dependencies.
Google Tag Manager
Deploy and manage experiments straight through GTM. No developer time required.
Server-side API
Run tests beyond the browser: pricing, search and routing, flicker-free.
- Shopify
- BigCommerce
- Shopware
- Centra
- Geins
- Saleor
- WordPress
- Optimizely CMS
- Storyblok
- Next.js
- Vue
- Astro
- Matomo
- Amplitude
- RudderStack
- Slack
- WooCommerce
- Salesforce
- Wix
- Litium
- commercetools
- Medusa
- Drupal
- Contentful
- Webflow
- Nuxt
- Angular
- Remix
- Piwik PRO
- Mixpanel
- Tag Manager
- Webhooks
- Adobe Commerce
- PrestaShop
- Squarespace
- Norce
- Shopify Hydrogen
- Vendure
- Umbraco
- Sanity
- Framer
- React
- SvelteKit
- GA4
- Adobe Analytics
- Segment
- Klaviyo
Not on the list? If it renders HTML, Pertento can test it: read how each one connects ↗