Moving from another system

If your site used another analytics system before, you can bring its history into Looku with looku history:import. Your charts then continue without a gap.

You need: the earlier data as a file with one JSON object per line (JSON lines). Most systems can export to CSV or JSON; you convert that to the format below with a small script or a spreadsheet.

File format

{"type":"pageview","ts":"2026-10-29T08:00:00Z","path":"/","visitor_hash":"a1b2c3","referrer_host":"google.com","referrer_path":"/","utm_source":"","utm_medium":"","utm_campaign":"","utm_content":"","utm_term":""}
{"type":"engagement","ts":"2026-10-29T08:02:00Z","path":"/","dwell_ms":12000,"max_scroll_pct":80}
{"type":"event","ts":"2026-10-29T08:03:00Z","path":"/","name":"add_to_cart","target":"","value":null,"props":{},"items":[{"sku":"A1","quantity":1}]}

Field	For	What
type	all	pageview, engagement or event
ts	all	time in UTC (ISO 8601); Looku counts the day in the site’s time zone
path	all	the page path
visitor_hash	page views	the earlier system’s daily visitor identifier, never an IP address. Without it, every page view counts as a new visitor
referrer_host, referrer_path, utm_…	page views	optional
dwell_ms, max_scroll_pct	engagement	reading time in milliseconds and scroll depth 0–100
name, target, value, props, items	events	as in events

Device, browser and country stay empty for imported rows.

Import

Find the time of the first page view Looku received itself: in the dashboard, or ask yourself when you connected the site. Use it as --before, so that nothing is counted twice. Copy the file to the server’s deploy folder (from your own computer: scp history.jsonl you@192.168.1.10:looku/deploy/), and in that folder run:

docker compose run --rm -v "$PWD/history.jsonl:/tmp/history.jsonl:ro" ingest history:import myblog /tmp/history.jsonl --before 2026-11-02T12:00:00Z

You should see a summary like:

3 page views, 1 events, 1 engagement rows imported to myblog; 1 skipped at or after --before, 0 older than the retention; 1 days rolled up.

Rules

• All or nothing. If one line is invalid, nothing is imported, and the message says which line. Fix it and run again.
• The same file cannot be imported twice. A changed file counts as a new file, so do not import overlapping files.
• Put each day’s rows in one file. Visitor identifiers are hashed again with a key that exists only during one import, so the same visitor on the same day in two files would count twice.
• Rows older than the site’s retention period (default 25 months) are skipped.
• Finished days are added up right away; the last two days are added up by the next nightly job.

For what this means for privacy, see Privacy and the law.