PostHog measures the product.
Xi runs the marketing experiment.
PostHog is the modern PLG analytics suite: events, session replays, feature flags, experiments built on those flags. Xi captures a different shape of work — the contract behind a marketing experiment, outside the application surface, decided by the rules.
Why teams switch from PostHog Experiments to Xi
PostHog is the right tool for product experiments — feature flags, exposure tracking, analytics in one place. Marketing experiments rarely fit that shape. They span channels PostHog cannot instrument (paid networks, outbound email, pricing pages, onboarding sequences) and the unit is not a flag exposure, it is a verdict against a contract. Teams move to Xi for four reasons.
- The unit of work matches the experiment. A pricing change, a paid campaign, a content cadence — none are flag exposures. Xi captures them as contracts.
- Channels PostHog does not reach. Paid ads, outbound emails, pricing pages on landing services — Xi works on all of them because it does not need to instrument anything.
- The agent runs it. Xi exposes a remote MCP server. Claude or Cursor can commit experiments, log values on a cadence, and confirm verdict against the threshold. PostHog has APIs and a great UI but no native MCP.
- A contract decides, not significance. Marketing experiments often run on small N. Xi commits to a threshold up front and the rules decide the verdict.
If you are running product experiments at PostHog scale, keep PostHog. If you are running marketing experiments, switch.
Where the two tools actually differ.
When each tool is the right call.
- You are running product experiments tied to features in a product team.
- You want analytics, replays, and experiments in one tool with one event stream.
- Your experiments are flag-based: gradual rollouts, exposure-based variants.
- You are already a PostHog customer and want experiments wired to your existing event taxonomy.
- Your experiments span channels PostHog does not instrument.
- You want a contract per experiment, not a flag per feature.
- You want zero SDK footprint and zero PostHog dependency.
- You want the agent to commit and confirm experiments via MCP.
Common questions, short answers.
How is Xi different from product experimentation built on flags?
Different unit of work. Flags expose code variants; Xi captures contracts about what we are trying. Different decision rule too: pre-committed thresholds versus statistical analysis on exposure.
Can I use Xi alongside PostHog?
Yes. Use PostHog for product analytics and feature experiments tied to your event stream. Use Xi for the marketing experiments and the cross-channel contract archive.
Does Xi have analytics?
No. Xi captures hypothesis, metric, threshold, verdict. Analytics belong in PostHog or your existing stack — Xi just reads the metric you trust.
Do I need PostHog to use Xi?
No. Xi works with whatever metric source you already trust — your CRM, an ad platform, a manual log, or a query you run weekly.
Take one idea. Turn it into an experiment.
Free plan, unlimited archive, no card required. See it in Claude / Cursor / Codex in 30 seconds.