Reading a Facebook Ads export without double counting Reach.

How to export from Facebook Ads

1. Open Ads Manager and pick the level Go to the Ads Manager table and choose Campaigns, Ad Sets or Ads with the tabs above it. The level you are on is the level you export, and mixing levels between two files later will not join cleanly.
2. Choose your columns and breakdowns Use the Columns menu to build the metric set, then Breakdowns and Time to add a day breakdown. Without a time breakdown the whole date range collapses into one row per ad and there is no trend to chart.
3. Set the attribution window before exporting The attribution setting sits in the Columns menu as Comparing Windows or Attribution Setting. It is applied at export time, so the same date range gives different Results under a seven-day click window than under a one-day one.
4. Export the table data Use the Reports or Export control at the top right and choose the CSV format rather than XLSX. Ads Manager offers to export only what is on screen or the full table, and the full table is what you want.
5. Drop the file here Campaign name, Ad set name and Ad name become three levels of breakdown chips, and the spend line is drawn from Amount spent with the currency read from the header.

The Amount spent column carries the currency in its own header text rather than in the values, which means two ad accounts on different currencies produce headers that do not match. Rename one before merging or the merger will treat them as separate columns.

Try with sample Facebook Ads data (the same columns, 96 rows, no upload needed).

Ads Manager export traps

• Two date columns, not one. Reporting starts and Reporting ends both appear on every row. With a daily breakdown they are the same date; without one they span the whole range and there is no time series to build.
• Results is polymorphic. Its meaning is defined by Result indicator on the same row. A mixed-objective account cannot have its Results column summed, and Cost per result cannot be averaged.
• Attribution is baked in at export. Change the window and re-export the same dates and the Results change. Old files and new files are not comparable unless the window matched, and nothing in the CSV records which one you used.
• Reach is deduplicated people. It cannot be added across days, ad sets or campaigns, because the same person appears in several rows and Facebook has already removed the duplicates within each row only.
• Frequency is a ratio, not a metric to average. It is Impressions divided by Reach for that row. Averaging the column across rows gives a number that matches no real recomputation.
• The CPM header contains a comma. It is quoted in the file, which is correct CSV, but a naive split on commas turns one column into two and shifts every field after it.

Group by Result indicator before you total Results, treat Reach as a count of people rather than a number to add, and write the attribution window down somewhere outside the file. That is most of what separates a useful paid social report from a confident wrong one.

Frequently asked questions

Related tools