Upload offline conversions from Google Sheets or CSV

The sheet layout that works

One row is one sale or one step in a sale. Keep the header row exactly as below, in plain lowercase, and format every column as plain text before you paste anything in. That single habit prevents most upload problems.

1. email: the customer's email address, no spaces.
2. phone: with the country code, for example +14155550137. Format the column as text, or the plus sign and leading zeros disappear.
3. event: the stage, such as lead, booked, showed, won or paid.
4. date and time: when the event happened, in one consistent format.
5. value: the amount as a plain number, no currency symbol or thousands separator.
6. currency: a three-letter code such as USD, GBP or EUR.
7. order_id: your own ID for the sale. The same order ID sent twice counts once, so this is what protects you from doubles.
8. click ID columns: gclid, wbraid or gbraid for Google, fbclid, fbc or fbp for Meta, ttclid for TikTok, li_fat_id for LinkedIn. Format as text. A click ID that a spreadsheet turns into a number or scientific notation cannot be matched.

Past customers or new sales: which route?

This is the decision that matters most, so make it deliberately.

Use the Ads Warmup CSV for past customers. It takes up to 20,000 rows on a paid plan and is dated when it is sent, not when the sale happened. That makes it right for a customer list you want the ad platforms to learn from now, and wrong for pretending old sales happened this week.

Use a live feed for new sales. Add each new row to the sheet, and let Google Sheets, Zapier, Make or n8n post it to your private sales webhook. DataCops sends it the same day, matched to the ad click it kept on your own domain. The sale arrives well inside every platform's window, and nobody has to remember to upload.

Why a monthly upload loses rows

This is a made-up example to show the arithmetic, not a customer result. A shop closes 40 sales in June. It uploads one CSV on 30 June.

The 8 sales from the last week of June are inside Meta's 7-day limit and are accepted. The 32 earlier ones, closed between 1 and 22 June, are more than 7 days old, so Meta rejects them. Google Ads is kinder: all 40 are inside 90 days of the click, provided the click ID survived. If half the sales came from Safari browsers that lost the click ID, those rows have nothing to match on.

So one upload gives Meta 8 of 40 sales, or 20 percent, and Google fewer than it should. Sending each sale on the day it closes gives every platform all 40. The fix is not a better spreadsheet. It is a shorter gap between the sale and the send.

Works with the rest of DataCops