Import a spreadsheet

If your team does not run a supported CRM, you can upload a spreadsheet of your deals and get a real forecast. The upload becomes events on the same append-only event log a Salesforce or HubSpot backfill writes to, so point-in-time reconstruction works exactly the same way.

Find it under Onboarding → Import from a spreadsheet.

Before you start

• You need to be an org admin. The endpoint is org-admin-gated; other members do not see the card.
• Your workspace must have no CRM connected. If a live Salesforce or HubSpot connection exists, the import is refused. Mixing CSV-sourced and CRM-sourced events for the same deal would corrupt reconstruction. Re-uploading CSV on top of CSV is the intended path and is allowed.

File format

Export CSV (tab-separated is also accepted) with a header row. Excel users: Save As → CSV — .xlsx is a binary format and is not parsed.

Your browser does the parsing, and the parser handles the usual awkward cases: double-quoted fields containing commas, escaped quotes (""), newlines inside quoted fields, and both \n and \r\n line endings. The delimiter is detected from the header line — tab if one is present, otherwise comma.

Server-side limits: 200 columns and 50,000 rows per upload.

Mapping your columns

Pipemetry guesses a mapping from your header names and shows it as a table for you to correct. Anything it cannot place is set to Don’t import.

Field	Notes
Opportunity ID	A stable per-deal id, so re-uploads update the same deal
Deal name	Used as the key when there is no ID column
Amount	Currency symbols and commas are fine — $1,234.50 parses
Close date	ISO (2026-09-30) or M/D/YYYY
Stage
Owner / rep
Forecast category
Created date	Optional — defaults to the import day when unmapped

You must map a key column

Map one column to Opportunity ID or Deal name. That is what lets Pipemetry tell deals apart and update the right one on a re-import. The Import spreadsheet button stays disabled until you do, and the server rejects a keyless mapping too.

Won / closed from stage

By default Pipemetry derives each deal’s won and closed flags from its stage value. Two comma-separated lists control it, pre-filled with Closed Won and Closed Won, Closed Lost. Matching is case-insensitive, and everything not listed is treated as open. Clear the checkbox if you do not want the derivation at all.

Preview and import

The card previews your first 5 rows, using only the columns you mapped, so you can catch a mis-mapped column before writing anything. Then click Import spreadsheet. The result line tells you how many deals were parsed and how many new events were written.

Re-uploading is safe

Each event’s identity is derived from the deal, the field and a hash of the normalised value:

• Re-upload the same rows → the same identities are re-derived and nothing new is written (0 new events). The app says so explicitly: nothing changed since your last upload.
• Re-upload with a changed value → a new event is appended. Your history accretes: the newer value wins from the upload instant onward, and the older value is still what you see if you look at an earlier as-of date.

This is why re-uploading an updated export weekly is a reasonable way to build point-in-time history without a CRM.

Without an ID column, renaming a deal starts a new one

When you map Opportunity ID, that value is the deal’s identity and the name is free to change. When you don’t, Pipemetry derives the identity from the deal name — so editing a name between uploads reads as one deal disappearing and a different deal appearing. The old one keeps its history and stops updating; the renamed one starts from scratch.

If your deals ever get renamed, map an ID column. Any stable value works — a CRM record id, a row number you don’t reshuffle, even a hand-assigned code — as long as it stays put across exports.

What you get afterwards

The imported deals feed the same forecast, waterfall, opportunity and accuracy screens a CRM connection feeds. One honest caveat: a spreadsheet snapshot carries no real change history, so your point-in-time views only become meaningful from your first upload onward — earlier days have nothing to reconstruct from.