Comparison
Keep GA4. It’s free. This is the layer you’ll actually read.
Side by side
Their column first.
Twelve rows, and GA4 wins three of them outright. A comparison that only lists its own wins is a pitch; this one is meant to be checked.
| GA4 | Pear Metrics | |
|---|---|---|
| Meta spend | Not available at any price. | Synced daily, in the same table as Google. |
| Google Ads spend | Available once the accounts are linked. | Synced daily. |
| CPA by campaign, across platforms | Export both, join by hand, repeat every Monday. | One column. |
| Individual user exploration | Yes. User Explorer, per-user paths. | No, and never. There is no user identifier to explore. |
| Remarketing audiences | Yes, and they push straight into Google Ads. | No. This is the reason to keep GA4. |
| Sampling and thresholding | Explorations sample above a limit; small segments get withheld. | Never. Every session is counted and every row is shown. |
| Ad blockers | A known share of traffic never arrives. | First-party subdomain, so collection survives domain blocking. |
| Safari and returning visitors | Client-side cookies capped at seven days, so returning users are partly fiction. | Sessions are counted server-side. Returning users are not claimed at all. |
| Consent banner | Usually required, and consent mode models the gap. | Not required for analytics. Nothing is stored on the device. |
| Where the data lives | Google's infrastructure, on Google's terms. | A database we run and nobody advertises from. None of it is passed on. |
| Setup | gtag, conversion config, cost data imports, BigQuery export for anything real. | One DNS record, one script tag, two ad accounts connected. |
| Price | Free. | Free to 25K events a month with one ad connection, then from $59. |
The layer
The table GA4 cannot build.
Spend from both platforms and the sessions it paid for, one row per campaign, with cost per conversion on the same axis. GA4 has half of the inputs and none of the join.
Figures shown are illustrative.
The objections
“I can already see individual users in GA4.”
Some of them. Safari forgets them after seven days, ad blockers hide the rest, and Explorations sample what remains. We’d rather count all sessions than model some people.
The better question is what you last did with it. When did you last open User Explorer and change a bid because of what you saw there?
“GA4 is free.”
GA4 is free the way a billboard is free for the advertiser’s competitor. You’re the data. Also: it has no Meta spend, so the table you actually need still gets built by hand.
“We already have a consent banner.”
Then you already know what share of visitors decline, and that GA4 models the rest. Analytics that stores nothing on the device does not need to be behind the banner, so the refusals are in the count instead of estimated.
“We would have to run both.”
Yes, and that is the recommendation. Keep GA4 for the audiences and the Google Ads plumbing. Read this one for the numbers.
The concession
If your strategy runs on GA4 audiences and BigQuery, we’re not your tool.
Teams whose media buying depends on user-level remarketing audiences, or whose analysts live in BigQuery, need the identity layer this deliberately does not have. Better to know that now than three weeks in.
Keep GA4.
Add the layer.
