Do I need an attribution tool, or is GA4 enough?

I already have Google Analytics 4. Do I really need to pay for an attribution tool?

Google Analytics 4 is free, it is already on your site, and it has an attribution section. So before paying for another tool, the fair question is whether you need one at all. For a good share of stores the honest answer is not yet. For the rest, the answer depends on a few specific facts about how you spend and how you sell, and you can check all of them this afternoon.

What GA4 does well

GA4 is the best free tool for understanding behaviour on your own site: which landing pages convert, where people drop out of a funnel, how site search behaves, what returning visitors do differently. The data is first-party and no ad platform has a view of it.

It also includes attribution. Its Advertising reports assign credit for key events such as purchases with one of two models: data-driven, which uses the paths of converting and non-converting users in your property, or last click. Google retired first click, linear, time decay and position-based in November 2023, as its attribution settings page notes. The lookback window defaults to 90 days for most key events.

If most of your budget is in Google Ads, that is a reasonable place to start, because Google Ads and GA4 share click identifiers and GA4 can see the whole Google side of the journey.

Where GA4 stops being enough

The limits are not bugs. They come from how GA4 knows that a sale happened, and they matter more the more money a decision moves.

It counts purchases from the browser. GA4 registers a purchase when a purchase event fires in the visitor's browser. Ad blockers, tracking protection, a checkout that renders on a payment provider's domain and a customer who closes the tab before the confirmation page loads all remove the sale from GA4 while it stays in your store. We explain the mechanics in why GA4 disagrees with your ad platforms and your store.

It only subtracts refunds you report. GA4 lowers revenue for a refund when your store sends a refund event. Most implementations never add it, so returned and cancelled orders keep counting as revenue.

It sees non-Google cost only if you import it. To show cost and return on spend for Meta or TikTok, GA4 needs campaign data import, and the join depends on your UTM values matching exactly. "Facebook" and "facebook" are different sources to it.

It gives each sale to one channel in its acquisition reports. Traffic acquisition credits the last non-direct click of the session, so a buyer who clicked a Meta ad on Tuesday and searched your brand on Thursday is credited to search. That is a defensible rule, but it is not the answer to how much Meta produced.

Some rows are withheld. With Google signals active, GA4 applies data thresholds so a report cannot identify individuals, which can leave small segments incomplete.

A five-question test

Answer these about the last three months. Each yes is a reason GA4 alone will cost you money.

• Is more than a third of your ad budget outside Google, on Meta, TikTok or campaigns that end in WhatsApp?
• Are refunds and cancellations more than a small share of your orders, or concentrated in some channels?
• Is the gap between GA4 purchases and store orders large, or does it change from month to month?
• Has anyone moved budget between channels based on platform-reported ROAS in the last quarter?
• Do you sell across devices, or through leads that close later by phone or in a CRM?

Zero or one yes: keep GA4, tag your campaigns consistently, add the refund event and revisit in a quarter. Two or more: you are making budget decisions on numbers that nothing has checked against your orders, and that is the problem an attribution tool exists to solve.

What an attribution tool adds

The difference is the anchor. An attribution tool starts from the orders your store actually settled, not from a browser event, ties each order to the session and click that produced it, and keeps each ad platform's own claim beside that verified number instead of replacing it. That turns the question from which report is right into how much of each platform's claim holds up against sales that exist.

It does not replace GA4. The behavioural questions stay in GA4, where they belong. You can see how the options compare, including the free ones, in the best marketing attribution tools for ecommerce, and what it costs to add one in how much an attribution tool costs.

If you keep GA4 for now, do these three things

Use one naming convention for utm_source and utm_medium everywhere, in lowercase, so campaign data import can match. Send the refund event from your store. And once a month, compare GA4 purchases with store orders for the same days in the same time zone: if the gap is stable you understand it, and if it jumps something broke.