Should I compare GA4 sessions with Meta's link clicks or outbound clicks?

Outbound clicks. Meta's link clicks also count taps into its own full-screen formats, such as lead forms and Instant Experiences, which never reach your site. Outbound clicks only count clicks that lead off Meta. Even then, Meta says some traffic drops off before the page loads.

Why does DebugView show nothing after I click my own ad?

Usually consent. Google says events are not visible in debug mode when consent mode is on and the visitor has not accepted analytics cookies. Accept your banner, reload the page and make sure Tag Assistant started the session, since it adds the debug parameter.

Can a redirect remove the UTM tags from my Meta ads?

Yes. Google lists redirects, such as from one site to another or from https to http, among the ways UTM parameters get stripped. The visit then usually shows as (direct) / (none). Check the address GA4 recorded in DebugView, not the one in Ads Manager.

How to find missing Meta ad clicks in GA4, step by step

Count Meta's outbound clicks for one closed week, then GA4's sessions from every Meta source. Test-click your ad in debug mode to see if the tags survive. Then sort the rows by channel and check your cookie banner and landing domain.

By Joris van Huët, Founder & CEOUpdated 2026-10-019 min read

Run the numbers for your store: the free break-even ROAS calculator.

To find missing Meta clicks, compare Meta's outbound clicks with GA4's Meta sessions for one closed week. Then click your own ad in debug mode to see whether the tags survive. If they do, the gap usually sits in wrong channel labels, refused cookies, a second domain or shoppers who switch devices.

You need Ads Manager, GA4 and your Shopify admin. Two of the GA4 screens below, cross-domain measurement and reporting identity, need an Editor role.

Do steps 1 to 5 in one sitting. They take less time than an argument about whose dashboard is right.

Step by step

Count Meta's outbound clicks for one closed week. Link clicks also count taps into Meta's own full-screen formats, which never reach your site. Outbound clicks count only clicks that lead off Meta, which Meta calls a closer approximation of the traffic meant for your website. Menu path: Ads Manager > Campaigns > Columns > Customize columns > tick Outbound clicks > Apply.
Count GA4's sessions from every Meta spelling. Search the table for facebook, then instagram, fb and ig, and note each row's sessions. A search for facebook should also catch referral rows such as l.facebook.com. Menu path: GA4 > Reports > Acquisition > Traffic acquisition > dimension drop-down > Session source / medium > search bar above the table.
Sort each row by channel. Add Session default channel grouping as a secondary dimension, where Paid Social is the goal. Organic Social means the click had no paid medium, or arrived untagged from an address such as l.instagram.com, which Google lists as social. Paid Other means a paid medium with a source GA4 does not list, and Unassigned means no channel rule matched at all. Menu path: Traffic acquisition > the add button next to the dimension drop-down > Session default channel grouping.
Open a debug session on the address your ad sends people to. Meta appends the URL parameters to the ad's website URL, so join both into one address. Start Google Tag Assistant on it, then accept your cookie banner. Google says DebugView hides events when consent mode is on and analytics cookies were declined. Menu path: tagassistant.google.com, then GA4 Admin > Data display > DebugView.
Read the address GA4 recorded. Click the first page_view event and find page_location, Google's name for the page URL, among its parameters. Pass: every utm tag is still there. Fail: the tags are gone, so a redirect stripped them, one of Google's listed causes of (direct) / (none) traffic. Menu path: DebugView > Seconds stream > page_view > parameters list.
Check whether the ad lands on a second domain. Some ads open a quiz or a landing page on another domain before the store. Without cross-domain measurement, GA4 counts that shopper as two users and two sessions. The store then shows the first domain as the referral source instead of Meta. Menu path: GA4 Admin > Data collection and modification > Data streams > Web > your stream > Configure tag settings > Configure your domains.
Note where your cookie banner runs. In configured regions, Shopify collects non-essential data only after consent. If your GA4 tag respects that choice, Meta visitors who decline there never become sessions. Menu path: Shopify admin > Settings > Customer privacy > Cookie banner.
Show modelled data if GA4 offers it. The Blended reporting identity mixes in modelled behaviour for visitors who declined cookies, once your property qualifies. Switching it does not change what GA4 collects. Menu path: GA4 Admin > Data display > Reporting identity > Blended > Save.
Check boosted posts on their own. Meta applies URL parameters to clicks on the ad, not to clicks from people who see the same post organically. Those organic clicks arrive untagged and usually land in Organic Social. Menu path: Ads Manager > the ad > Tracking > URL parameters, then the table from step 3.
Compare devices on both sides. Meta's Impression device breakdown shows where people saw the ad, and GA4's Device category shows where they bought. GA4 cannot join two devices without a user ID, so a phone viewer who buys on a laptop never shows as Meta. Menu path: Ads Manager > your campaign > Breakdown > By delivery: Impression device; in GA4, add Device category as a secondary dimension in Traffic acquisition.

A worked example

Here is a made-up week, in round numbers, for illustration.

For illustration, one closed week	Count
Meta link clicks	2,000
Meta outbound clicks	1,600
GA4 sessions from Meta sources, all channels	1,000
of which Paid Social, as facebook / paid	550
of which Organic Social, as l.instagram.com / referral	300
of which Paid Other, as insta / cpc	150

For illustration, start from the 1,600 outbound clicks, not the 2,000 link clicks. The other link clicks include taps into Meta's own formats, which were never headed for your site.

In this worked example, 1,000 of the 1,600 became GA4 sessions, which leaves 600 with no session at all. Say step 5 shows your tags intact and step 7 shows the banner running in your biggest market. Then, in this worked example, visitors who declined cookies there are the first suspects for the missing 600. Meta itself warns that some traffic drops off between an outbound click and a page view.

Now the sessions that did arrive. For illustration, only 550 sit in Paid Social. In this worked example, the 300 from l.instagram.com / referral are untagged clicks. They may come from an old ad or from the organic side of a boosted post. For illustration, the 150 under insta / cpc carry a paid medium. But insta is missing from Google's list of social sources, so GA4 files them under Paid Other.

In this worked example, 450 sessions are Meta's but wear the wrong label. That part is a tagging job you can finish this week. For illustration, the 600 missing sessions are a different problem, and no tag will bring them back.

Leaks you never fix settle somewhere, and one store's export shows the usual place. On that store's Channels sheet, Direct holds 57.7% of revenue in all three views. Google describes (direct) / (none) as traffic without a clear referral source, which is what an untagged click with no referrer looks like. The export holds no spend, so it cannot say how much of that 57.7% began with an ad.

What to check when the report looks wrong

GA4 shows more Meta sessions than outbound clicks. Organic posts, shares and messages also send visitors from facebook.com and instagram.com. Only rows with your paid medium belong to your ads.
DebugView stays empty. Accept the cookie banner and reload the page. If it is still empty, restart from Tag Assistant, which adds a debug parameter to the address.
Your own domain shows up as a referral. A subdomain or a second domain is missing from Configure your domains. Google warns that self-referrals may mean your traffic is being attributed incorrectly.
A payment page shows up as a referral. List that domain as an unwanted referral, so shoppers who return from paying keep their real source. Menu path: Configure tag settings > Show all > List unwanted referrals.
Paid Social has sessions but almost no revenue. Buyers came back through email or search, which took the sale under last click. GA4's Key event attribution paths report, under Advertising, shows the journeys Paid Social opens.
Meta's clicks jump while GA4 stays flat. Go back to step 1. A new full-screen format raises link clicks without sending anyone to your site.

What to do this week

Run the test click today. Use steps 4 and 5 on your three biggest live ads. Pass: page_location keeps every utm tag on all three. Fail: a tag goes missing, so fix that redirect before anything else.
Write down the week's funnel. Note outbound clicks, GA4 sessions from Meta sources, and how many of those sit in Paid Social. Pass: most Meta sessions sit in Paid Social. Fail: Organic Social, Paid Other or Unassigned hold a big share, so retag those ads.
Repeat the funnel a week after your fixes. Use the same weekdays and the same campaigns. Pass: Paid Social's share rises and the mislabelled rows shrink. Fail: nothing moves, so hunt for boosted posts and older ads still running untagged.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Frequently asked questions

Should I compare GA4 sessions with Meta's link clicks or outbound clicks?
Outbound clicks. Meta's link clicks also count taps into its own full-screen formats, such as lead forms and Instant Experiences, which never reach your site. Outbound clicks only count clicks that lead off Meta. Even then, Meta says some traffic drops off before the page loads.
Why does DebugView show nothing after I click my own ad?
Usually consent. Google says events are not visible in debug mode when consent mode is on and the visitor has not accepted analytics cookies. Accept your banner, reload the page and make sure Tag Assistant started the session, since it adds the debug parameter.
Can a redirect remove the UTM tags from my Meta ads?
Yes. Google lists redirects, such as from one site to another or from https to http, among the ways UTM parameters get stripped. The visit then usually shows as (direct) / (none). Check the address GA4 recorded in DebugView, not the one in Ads Manager.

Go deeper: Incrementality testing, explained.

Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.

Keep reading

Terms in this article

Browse the full glossary