Attribution / measurement

Multichannel attribution — who to trust (Google, Meta, GA4)?

Jakub Pytlik | May 28, 2026 | 8 min read

In short Your Google Ads dashboard shows 40 sales, Meta shows 35, and Google Analytics (GA4) shows 28 — and every one of those numbers is "true." A gap of 20–40% between platforms is the norm in 2026, not a bug: each one counts in its own way (Meta credits ad views without a click, GA4 spreads credit across all channels, Google has its own time windows). The worst thing you can do is add the platforms together — you'll count the same sale two or three times and overestimate your results. I explain where the gap comes from, who to actually trust, how the big players measure this with tools like Northbeam or Triple Whale, and what makes sense for a small business.

Part of the full-funnel series Attribution is one of the pillars of thinking about the whole funnel. If you landed here out of context, start with the hub: full-funnel marketing — is the bottom of the funnel enough.

Why the numbers don't match (and why that's normal)

Picture a single customer. She saw your ad on YouTube, three days later clicked an ad on Instagram, and a week after that typed your brand into Google and bought. One sale — and Google, Meta, and your Google Analytics all claim credit for it. Each of them counts it differently, because each sees only its own slice of the story.

Three main reasons the numbers drift apart:

Different crediting models. Meta assigns the sale to its own ads (last-touch within Meta). GA4 by default spreads credit across all channels (data-driven). Google Ads has its own settings on top of that. Three different ways of counting the same thing.
Views without a click (view-through). Meta can credit a sale to someone who only saw the ad and didn't click — by default in a 1-day-after-view and 7-day-after-click window (you can change these settings in the dashboard). Google Ads doesn't do this on Search. That alone drives a gap.
Different windows, settings, and consent. Different conversion windows (how many days back to count), different time zones, iOS privacy limits, and on top of that Consent Mode — when someone rejects cookies in a GDPR banner, the platforms model their path instead of measuring it. Small details that add up.

In practice, a gap of around 20–40% between platforms counts as the norm in 2026, not a bug to fix. To put it bluntly: each platform grades its own homework — and usually gives itself an inflated score. (The most common, narrower case — the Google Ads vs GA4 gap — I cover separately in why Google Ads and GA4 data don't match.)

Attribution models in plain English

"Attribution model" sounds intimidating, but it's just a rule for who gets credit for a sale when there were several ad touches along the way. Two you need to know:

Last-click — all the credit goes to the last channel before the purchase. Simple and tempting, but misleading: it almost always rewards the bottom of the funnel (e.g. someone searching for your brand) and completely ignores what brought the customer in earlier. That's why the top of the funnel looks "useless" — I come back to this in the full-funnel article.
Data-driven — splits credit among the touches based on which ones actually pushed toward the purchase. Fairer. The catch: it only works within a single platform. Data-driven in Google can't see Meta touches, and in Meta it can't see Google. So a better model still doesn't merge two platforms into one truth.

Who to actually trust

Since every platform reports in its own favor, and you can't add them up — what do you do? Three rules I follow:

Pick one common denominator for budget decisions. Usually that's GA4, because it's the only one that sees all channels at once and measures them with a single model. When you're deciding "move budget from Meta to Google?", look at GA4 (or at real orders in your store/CRM), not at the sum of the ad dashboards.
Treat platform dashboards as a guide for optimizing within them. Google Ads will tell you nicely which Google campaign is performing better. But don't use it for "Google versus Meta" comparisons — because each one credits itself with the same people.
Your money is closest to the truth. The number of real orders, invoices, and signed contracts in a given month is undeniable. The platforms should fit to it, not the other way around.

And to be fair: GA4 isn't the absolute truth either — it models, it estimates part of the data, and in 2026 Google additionally rebuilt its attribution settings (more flexibility per conversion). It sees clicks from all channels, but it won't see Meta ad views without a click — if Meta sells through presence alone, GA4 won't credit it for that. Even so, among the free tools it's the closest to a shared, honest view. Don't chase a perfect match on a single sale — look for the trend: whether it's rising or falling month over month.

Tools: Northbeam, Triple Whale, and whether a small business needs them

Since no platform sees the whole picture, tools emerged that try to stitch it together. In e-commerce/DTC the best-known are Northbeam and Triple Whale. Both combine three approaches: multi-touch attribution (they track the whole path), marketing-mix modeling (statistical, cookie-free), and incrementality tests (they check whether turning off a channel actually reduces sales).

Triple Whale — more of an "all-in-one" (attribution + analytics + AI), easier to get started with, cheaper, with a module that combines all the methods into a single view (recalibrated weekly, according to the vendor). For most e-com brands in 2026, the more practical choice.
Northbeam — a powerful, specialized measurement engine, but it takes a lot of learning upfront and reports more slowly. In practice, many brands hire a dedicated analyst just to run it.

Reality check for a small business This is the league of brands that spend tens or hundreds of thousands a month on ads and have someone on staff who handles the numbers. If you spend 2,000–8,000 PLN a month, a tool like this is overkill — it only really pays off at budgets in the tens of thousands of zloty per month and with your own analytics person, because otherwise it costs more than it improves your decisions. I'm mentioning these so you know they exist and what they do, not as a recommendation. At your stage, a common denominator in GA4 plus solid UTMs do 90% of the job for free.

What to do on your own — the practical minimum

No tools costing thousands of zloty, within reach of any small business:

Set one common denominator (GA4 or real orders from your CRM/store) and compare everything else against it.
Never add up conversions from different ad dashboards. That's the first step toward overestimating your results and burning budget on a channel that "looks" good.
Tag your links with UTMs — everywhere you drop a link (posts, newsletter, partnerships). Without it, GA4 doesn't know where the traffic came from.
Close the loop on offline sales — if you finalize over the phone, by email, or in store, upload those transactions back into the system (conversion import). Otherwise the bottom of the funnel looks weaker than it is. How to set it up — in the article on conversions in Google Ads.
Accept a 20–40% gap as the norm and stop chasing a perfect match. Watch the direction, not the last decimal place.
For serious decisions (turn off a whole channel?) don't look at the dashboard — run an incrementality test. How — I describe it in the article on how ad channels affect each other.

Summary Google Ads, Meta, and GA4 show different numbers because each one counts only its slice of the path and does it its own way — a 20–40% gap is the norm. Don't add the platforms together (you'll count the same sale several times). Pick one common denominator (usually GA4 or real orders), and treat the ad dashboards as a guide for optimizing within them. Tools like Northbeam or Triple Whale are for the big league — a small business is fine with GA4, solid UTMs, offline conversion import, and common sense.

Your dashboards showing three different numbers? — I'll find where you're double-counting Tell me what conversion numbers you see in Google Ads, Meta, and GA4 for the same month (screenshots are fine). I'll reply with where you're most likely double-counting the same sale and which number I'd use as the basis for budget decisions. No strings attached — jakub.pytlik@jp-ads.pl or the contact form.

About the sources: The mechanics of the gap between platforms and "20–40% is the norm" — among others EasyInsights and Ruler Analytics; attribution models — Google Analytics Help. Tool comparison — materials from Triple Whale and QRY (2026). Assessments of how useful these tools are for small budgets, and examples marked "from my practice," come from my experience at JP Ads.

Keep reading in the series

← Blog Check your measurement