GA4 cost data import now has native connectors for Meta, TikTok and Snapchat, plus Pinterest and Reddit, under Admin > Data import > Campaign data. They pull cost, clicks and impressions daily and backfill at least 24 months. GA4 joins that cost to sessions on utm_source, utm_medium and date, so each platform needs one consistent source and medium. For X, LinkedIn or Microsoft Ads, upload a file in the same shape.
- Google now calls it "Import campaign data"; since July 2026 cost rows need a currency.
- Matching is exact and case-sensitive: Facebook is not facebook.
- utm_id and utm_campaign are optional but needed for campaign-level cost.
- Imported cost shows in Reports > Acquisition > Non Google campaign and Advertising > Planning > All channels.
- Cost is joined when the report runs, so it can’t be used in audiences or exploration segments.
The reason to do this is simple: once cost sits next to GA4 sessions and key events, you can show cost per engaged session, cost per key event and ROAS for every platform under one set of rules. The catch is that GA4 only matches cost to sessions whose UTMs agree with the import exactly, so this job is half setup and half UTM discipline.
Before you start
Check these first
- Editor role on the GA4 property
- Analyst access or higher on the TikTok and Snapchat ad accounts (the connectors ask for it)
- One utm_source and one utm_medium per platform across every live ad
- utm_id set to the platform campaign ID, and utm_campaign to the campaign name
- No earlier manual cost uploads for the same platform, or delete them first (GA4 does not de-duplicate)
If your UTMs are inconsistent, fix them before connecting anything (the paid social UTM fix). The connector asks for a single source and medium per platform and only matches sessions that use them.
Native connector or file upload?
Choose the import method
| Platforms | History | Upkeep | Use when | |
|---|---|---|---|---|
| Native connector | Meta, TikTok, Snapchat, Pinterest, Reddit | At least 24 months on setup | Runs daily on its own | The platform is supported. Almost always the right choice. |
| CSV upload | Any platform | Whatever you upload | You re-upload, up to 120 times a day per property | X, LinkedIn, Microsoft Ads or a one-off backfill |
| Scheduled source | Any platform | Whatever the source holds | Set once | You already keep cost in Google Sheets, SFTP, BigQuery or a warehouse (open beta) |
Set up the Meta, TikTok or Snapchat connector
Five steps per platform
- Create the import sourceAdmin > Data collection and modification > Data import > Create import source > Campaign data, then pick Meta, TikTok or Snap. Meta and Snap can also be linked from Admin > Product links.
- Connect the ad accountSign in to the platform and choose the business and the ad account.
- Enter source and mediumType the exact values your URLs use, such as facebook and paid_social. Meta lets you enter one source per publisher platform (Facebook, Instagram, Audience Network, Messenger, Threads) if you split them.
- Review and finishCheck the schedule and filters, then finish. The first import takes up to 30 minutes and data reaches reports within 24 hours.
- Validate the matchOpen Reports > Data import > Campaign data import validation (added August 2026) and look for campaigns with missing cost, clicks or impressions.
The ad account, source and medium are fixed once the connector exists. To change them, delete it and create a new one, and check the history comes back as expected.
The file template for every other platform
For platforms without a connector, export daily campaign data and shape it into these columns. One row per campaign per day.
Campaign data columns
| Required? | Notes | |
|---|---|---|
| utm_source | Required | Exactly as in your URLs, lowercase |
| utm_medium | Required | Exactly as in your URLs, for example paid_social |
| date | Required | YYYY-MM-DD, not in the future |
| currency | Required with cost | ISO code such as USD, or set one fixed value when mapping |
| utm_id | Recommended | The platform’s campaign ID |
| utm_campaign | Recommended | The campaign name, matching utm_campaign in URLs |
| cost, clicks, impressions | At least one | Daily values; cost is the one you want |
utm_source,utm_medium,utm_id,utm_campaign,date,currency,impressions,clicks,cost
<source>,<medium>,<campaign id>,<campaign name>,2026-11-18,<currency>,<impressions>,<clicks>,<cost> - utm_term and utm_content can’t be in the file, and dynamic values such as {{campaign.name}} aren’t supported. Use the rendered names.
- Uploading the same source, medium, campaign and date again overwrites the row, so re-uploading the last 7 days each morning picks up the platforms’ late restatements.
- Limits: 1 GB per import source and 120 uploads per property per day.
Read GA4 ROAS honestly
GA4’s ROAS is revenue GA4 credited divided by cost the platform reported. The revenue side follows GA4’s own rules: no view-through, no platform modelling. So expect GA4 ROAS for paid social to differ from what Ads Manager shows, often by a lot for awareness-heavy campaigns. That isn’t an import error. Show both, and explain the gap with a reconciliation table (GA4 vs ad platform conversions). Google Ads cost needs no import; it appears once the accounts are linked.
Where this goes wrong
- Two mediums on one platform. Ads tagged cpc and paid_social: the connector matches one, and the other half of the sessions shows no cost.
- Connector plus old uploads. GA4 doesn’t de-duplicate between them, so cost doubles. Delete the manual dataset before connecting.
- Day edges. If the ad account and the GA4 property use different time zones, daily cost and sessions won’t line up at the edges. Compare by week.
- Renamed campaigns. utm_campaign no longer matches the cost rows. Keep utm_id on every ad so the campaign still joins.
- Expecting cost in audiences or segments. It’s joined at report time only.
Do it faster with Whathead
For Meta, TikTok and Snapchat, Google’s connectors do the job; use them. Whathead helps with the platforms around them and with checking the match. The Google Sheets add-on can pull daily campaign cost from X, LinkedIn or Microsoft Advertising into a sheet laid out like the template, ready to upload as a CSV or connect as a Sheets import source. With the Whathead MCP, Claude can do the same in a chat:
- “Pull daily cost, clicks and impressions by campaign from my X Ads account for the last 90 days in GA4 campaign-data columns, with utm_source [your value] and utm_medium paid_social.”
- “Compare GA4 sessions by session source / medium / campaign for last week with the campaigns in that file, and list anything that won’t match.”
Whathead doesn’t upload into GA4’s data import; you do that step in Admin. Whathead is our product.
GA4 cost data import: common questions
Can GA4 import Facebook ad costs?
Yes. GA4 has a native Meta connector under Data import that brings in cost, clicks and impressions daily, matched on utm_source, utm_medium and date.
Does GA4 cost import work retroactively?
The native connectors backfill at least 24 months on setup. For uploads, you can include past dates; the match-rate check looks at the most recent two years.
Why is my imported cost not showing?
Usually the source or medium in the import doesn’t match the URLs exactly, including case. Data can also take up to 24 hours to appear. Check the Campaign data import validation report.
Do I need utm_id for cost import?
No, the required keys are source, medium and date. Without utm_id or utm_campaign you only get cost at source/medium level, not per campaign.
Can I use imported cost in explorations?
You can report on it in the cost reports, but campaign data is joined at query time, so it can’t be used in audiences or exploration segments.
Sources
- GA4 Help (Checked 19 Nov 2026): import campaign data, data import limits, Meta connector, TikTok connector, Snap connector, exporting cost from other platforms, release notes
Written by Ahmed Shahin, founder of Whathead. Last reviewed 19 November 2026.
Pull cost from every other platform
X, LinkedIn and Microsoft Ads cost in a sheet shaped for GA4, refreshed on a schedule.



