PostHog Marketing Analytics

PostHog Marketing Analytics

When analyzing marketing data:

• Use posthog_list_events and posthog_list_persons (tool names may have a suffix if multiple PostHog servers exist) to trace user journeys and conversion funnels.
• Prefer posthog_list_insights / posthog_get_insight over raw events when answering funnel or retention questions.
• When comparing campaigns, use distinctId and event filters in posthog_list_events for attribution.
• Always specify projectId; call posthog_list_projects first if unknown.
• For conversion funnels, list events with relevant filters (event names, time range) then summarize steps and drop-off.
• When asked about retention, use persons and events to identify returning users and cohort behavior.

Step-by-step instructions

1. If projectId is unknown, call posthog_list_projects and pick the relevant one.
2. For funnel questions: identify the event sequence (e.g. signup → trial_start → paid), call posthog_list_events with projectId, event filter, and after/before; summarize steps and drop-off rates.
3. For attribution or campaigns: use posthog_list_events with distinctId and event filters to relate events to campaigns; summarize by campaign or source.
4. For retention: use posthog_list_persons and posthog_list_events to count returning users and cohort behavior over time.
5. Prefer posthog_list_insights / posthog_get_insight when available; otherwise cite filters and time range used.

Examples of inputs and outputs

• Input: “What’s our signup-to-paid funnel look like this month?”
  Output: Short summary: event names per step, counts or rates, main drop-off step; mention projectId and date range used.

• Input: “Which campaign drove the most signups?”
  Output: Ranked list or table (campaign/source, signup count or share) with attribution from event properties; cite query filters.

Common edge cases

• No projectId: Call posthog_list_projects first and ask the user which one, or use the default if only one.
• No events in range: Report “No events in this period” and suggest a wider range or different events.
• Ambiguous event names: List available events or ask the user to confirm event names before building the funnel.
• Rate limits or errors: Tell the user the request hit a limit or failed and suggest narrowing the range or retrying later.

Tool usage for specific purposes

• posthog_list_events: Use to query events with projectId, event name, after/before, distinctId, personId for funnels, counts, and attribution. Always include a sensible time range.
• posthog_list_persons / posthog_get_person: Use for retention and cohort-style questions; relate to events via distinct_id.
• posthog_list_insights / posthog_get_insight: Use when they support funnel or retention answers.