Data Export

The dashboard is great until you need to do something the dashboard doesn't do. Feed a custom model. Run a one-off analysis in a notebook. Hand a CSV to a partner who doesn't have access to your Routy account. Ingest the data into a warehouse you control directly. Sync a slice of conversion data into a tool that doesn't have its own webhook integration.

Routy's data export gives you the raw event data in the format your downstream tooling actually wants. CSV for spreadsheets and partner handoffs, JSON for code and APIs, Parquet for warehouses and notebooks. The export is async — you kick off a job, Routy processes it in the background (including very large historical windows), and the file is ready when you come back.

After running an export, the following is available:

• Every event you track — clicks, page views, conversions, payouts — exportable in full, at the event level rather than aggregated. The export gives you what's in the raw event log for your account, not a summarised view of it.
• CSV, JSON, or Parquet output. Each format suits different downstream tools: CSV for spreadsheets and partners, JSON for APIs and scripts, Parquet for analytical workloads (notebooks, warehouses, data-lake ingest) where the columnar storage and compression matter at scale.
• Day-by-day processing with historical days cached. The first export for a given date range processes from the raw log; repeat exports for overlapping windows complete dramatically faster because the per-day chunks are already prepared. This matters when you're iterating on an analysis that needs the same date range multiple times.
• Async jobs you can kick off and forget. Routy emails you when the file is ready, or you can watch progress in the UI. For very large windows (a full year of data on a high-volume account), the export might take minutes to complete, but it's always reliable for windows that would crash a "just download the CSV synchronously" approach.
• Resume on failure. If a long-running export hits a transient issue, it picks up from the last completed chunk rather than restarting from the beginning. For multi-hour exports this is the difference between "occasionally lose the whole run" and "always complete eventually."

The format choice deserves a note. CSV is the universal lowest-common-denominator and what most users default to, but it's lossy for nested fields and slow to process at scale. JSON preserves the structure but is verbose. Parquet is what an analyst on a notebook actually wants — typed columns, compressed storage, fast to load with pandas or DuckDB or whatever's at hand. For one-off partner handoffs CSV is right; for serious analytical work Parquet is right.

You open the exports section in the dashboard, pick the data type (clicks, page views, conversions, payouts), pick the date range, pick the format, and submit. The job goes into a queue and starts processing.

Routy's processing approach is day-by-day chunking. For each day in the requested range, Routy either reuses an already-prepared chunk (if you've exported overlapping data recently) or generates a fresh chunk from the raw event log. Chunks are concatenated into the final file as they complete, so progress is monotonic and visible.

When the export completes, you can download the file directly from the exports interface, or — if you've enabled email notifications — receive a link by email when it's ready. The file is available for download for a configurable retention period (typically 7-30 days depending on plan).

Worth knowing:

• Export sizes are unlimited in principle, but very large exports take time. A full year of click data for a high-volume account might be tens of gigabytes in CSV or a few gigabytes in Parquet.
• Authentication for the download link is account-scoped — anyone with the link who's also logged into your Routy account can download. Sharing the file with external parties means downloading it yourself first and re-sharing.
• Concurrent jobs are supported but rate-limited per account to keep one user's batch exports from starving another's. The limit is generous enough that most workflows don't hit it.

The export is the safety valve for everything the dashboard doesn't do. Most operations think they don't need it until they need it — until a partner asks for a CSV they can audit, until an analyst wants the raw data in a notebook, until a one-off question requires joining your conversion data against a separate data source that doesn't live in Routy.

The day-caching design matters more than it sounds. The standard alternative to "export day by day" is "export the whole range every time," which on a year of data and a high-volume account means waiting an hour for an export each time you tweak the analysis. With day-caching, the second iteration of an export over the same range completes in under a minute because the per-day chunks are already done. The third iteration is faster still. Analysts spending real time on Routy data tend to notice this within the first week and adjust their workflow accordingly.

The format choice matters too. Parquet exports for analytical work are 5-10× smaller than the equivalent CSV and load 10-50× faster into a notebook. For routine work this doesn't matter; for serious analytical work it's the difference between a workable workflow and an unworkable one.