ProductAnalyst.ai

CSV Import

Import historical event data from any analytics platform and get insights immediately

Overview

CSV import lets you bring historical event data into Product Analyst AI from any tool that can export to CSV. Upload a file, map your columns, and the agent can start answering questions about your data immediately — no live integration required.

This is the fastest way to evaluate Product Analyst AI with real data. Export a few months of events from your existing analytics tool, upload the CSV, and start asking questions.

Compatible tools

Any tool that exports event-level data to CSV works. Here are the most common sources:

PlatformHow to export
PostHogEvents → select date range → Export CSV. Columns distinct_id, event, timestamp are auto-detected.
AmplitudeMicroscope or Chart → Export CSV. Map User Id, Event Type, Event Time during upload.
MixpanelInsights or Users → Export → CSV. Map Distinct ID, Event Name, Time during upload.
Google Analytics 4Explorations → Free Form → export CSV. Or use BigQuery export for raw events. Map user_pseudo_id, event_name, event_timestamp.
HeapDefine chart → Export data (CSV). Map User ID, Event, Time during upload.
PendoData Explorer → Export CSV. Map Visitor ID, Event Name, Timestamp.
LookerRun an Explore on your events table → Download → CSV. Column names depend on your model — map them during upload.
SegmentWarehouse query or Personas export → CSV. Map user_id, event, timestamp.
Custom / data warehouseRun a SQL query against your events table, export as CSV. Any column names work — you map them during upload.

CSV format

Your CSV needs three columns (the names can be anything — you map them during upload):

  • User ID — a unique identifier for the user (email, UUID, internal ID)
  • Event name — what happened (signup, purchase, page_viewed)
  • Timestamp — when it happened (ISO 8601 format preferred, e.g. 2026-01-15T10:00:00Z)

Any additional columns are ignored. Here's an example:

user_id,event,timestamp
user-42,signup,2026-01-15T10:00:00Z
user-42,onboarding_completed,2026-01-15T10:05:00Z
user-42,invite_sent,2026-01-16T14:30:00Z
user-99,signup,2026-01-15T11:00:00Z
user-99,upgrade_clicked,2026-01-17T09:00:00Z

How to upload

Step 1 — Export from your analytics tool

Export event-level data as a CSV from your analytics platform (see the table above for platform-specific instructions). Include at least a few weeks of data for meaningful insights — a few months is ideal.

Step 2 — Upload and map columns

Go to Settings → Integrations and click Upload CSV. Drop your file (up to 10 MB) and the uploader will auto-detect standard column names like user_id, event, and timestamp.

If your columns have different names (e.g. Distinct ID, Event Type, Event Time), use the dropdowns to assign each column to the right field. A preview of your first few rows helps you verify the mapping is correct.

Step 3 — Import

Click Import events. The importer shows you how many events were added and how many duplicates were skipped (if you're re-uploading).

Step 4 — Ask a question

Your data is immediately available. Try asking the agent something like: "What are the most common events in the last 30 days?" or "Show me a signup-to-upgrade funnel."

Deduplication

Duplicate events are automatically detected and skipped — it's safe to re-upload the same CSV or overlapping date ranges. Deduplication uses a hash of the user ID, event name, and timestamp, so identical rows are never inserted twice.

Limits

  • File size: 10 MB per upload
  • No row limit: files up to 10 MB typically contain 50,000–200,000 events depending on column count
  • Multiple uploads: upload as many files as you want — they accumulate, duplicates are skipped

Combining with live data

CSV import works alongside all other data sources. A common pattern:

  1. Upload a CSV with 3–6 months of historical data to bootstrap the agent
  2. Connect PostHog or the PAI SDK for live event streaming going forward

The agent sees all events together regardless of source, so historical context from your CSV enriches real-time queries from day one.

Other data sources

For live event streaming (no CSV export needed), connect PostHog via webhook or add the PAI JavaScript SDK to your site.

CSV Import - Product Analyst AI