Yandex Metrica or GA4: What to Install on a Project in Kazakhstan (and Why the Answer Is Both)
People ask “Metrica or GA4” as if you had to pick one. That is the wrong framing. These are two different systems with different philosophies, and on most serious projects in the CIS both are installed. Not out of greed — out of necessity.
Let's go through how they actually differ, rather than the way comparison tables do it with ten identical bullet points.
Different philosophies of measurement
Metrica thinks in sessions and visits. It is classic old-school web analytics — visits, bounces, page depth. GA4 thinks in events and in users across platforms. That is a completely different data model.
Which is why comparing the numbers head-on is pointless. Your visits in Metrica will never line up with sessions in GA4, and that is not a bug — they are different units of measurement. People burn days trying to reconcile a discrepancy that by definition should not reconcile.
Attribution: this is where the bodies are buried
Metrica gives you last-click by default plus a set of simple models. GA4 pushes data-driven. And so when you look at the same conversion in the two systems, the sources don't agree — because the models are different.
The practical conclusion: don't be surprised that Direct credits itself with more in Metrica and less in GA4. That is the models arguing, not the data lying. Pick one system as your source of truth for decisions and stop second-guessing.
Where Metrica is stronger
Webvisor. Session recordings, heatmaps, form analytics. GA4 has none of it. If it matters to you to see how a person actually pokes around a page and where they abandon a form, Metrica does it out of the box, for free. For UX analysis of landing pages it is irreplaceable.
Where GA4 is stronger
Raw data in BigQuery and cross-platform tracking. If you have web plus an app and need to follow a user across devices, GA4 can do it and Metrica cannot. Plus the free export to BigQuery, which opens the door to any custom analytics you want.
Why you end up with both
The prosaic reason: Yandex Direct works properly with Metrica, Google Ads with GA4. Each ad platform pulls its own system for optimization. Drop Metrica and you lose proper optimization in Direct. Drop GA4 and the same happens with Google Ads.
A contentious take: install both, but don't try to “reconcile” them at the level of numbers. Let each one serve its own ad channel and its own job. Metrica for Direct and UX. GA4 for Google Ads and raw material in BigQuery. Bringing it all together is only possible in BigQuery, and even then with the understanding that it will be an approximation, not an exact match. Anyone promising you “one unified, reconciled analytics setup out of two systems” either doesn't understand or is selling something.
Cross-Domain Tracking in GA4: How to Stop Losing the Source Between Your Own Domains
Set up cross-domain tracking in GA4 for multi-domain projects: linker, referral exclusions, one client_id. Why GA4 sees self-referrals and how to fix it.
Read →SKAdNetwork and SKAN 4.0: Why Your iOS Attribution Lies, and What to Do About It
SKAN 4.0 without the fluff: conversion values, three postback windows, the lock window. How to set up iOS attribution for a fintech app.
Read →Enhanced Conversions for Leads: How to Stop Paying for Leads That Never Make You Money
Set up Enhanced Conversions for Leads in Google Ads via GTM and server-side upload. Send CRM deal status back so Google optimizes for money, not forms.
Read →