
Your team migrated to GA4 because Universal Analytics stopped processing data, the numbers look roughly right, and nobody is sure why some reports disagree with your ad platforms. That is the normal state of a rushed migration. The good news is that a focused cleanup, done in an afternoon or two, removes most of the doubt.
Before opening the admin screen, write down the five to ten business actions that matter: a purchase, a lead form, a booking, a quote request. Every cleanup choice below flows from that list. If an event does not support a decision someone actually makes, it is a candidate for removal.
Open the events report and sort by volume. In most properties we see, three patterns show up: events that were recreated with slightly different names, events that fire twice, and events nobody has looked at since launch.
In GA4 any event can be marked as a key event or conversion. Resist marking everything. Each conversion feeds reporting, audiences and, if linked, ad bidding. Keep the list short and tied to the business actions you wrote down at the start. Then compare the count in GA4 against the source of truth, such as your order system or CRM, for one full week. Small gaps are expected. Large gaps mean a tagging or consent problem.
If customers move between your marketing site and a checkout, booking tool or payment page on another domain, configure cross-domain measurement. Without it, one visit becomes two, and the channel that started the journey gets replaced by direct or by your own checkout domain. Also review the referral exclusion list so payment providers and your own subdomains are not counted as traffic sources.
Event-level data retention in GA4 defaults to a short window in standard properties, and you can extend it in the settings. Extend it before you need it, since it does not recover data that has already aged out. While you are in admin, filter internal traffic, review who has access, and remove former agency and employee accounts.
The BigQuery link gives you raw event exports that you own, which matters for custom attribution, joining with CRM data and auditing what the interface shows. Exports start from the day you link, so the earlier you connect it, the more history you keep. Plan for storage and query costs by setting a budget alert and partition-aware queries from the start.

Consent banners shrink your data. Here is how consent mode, modeling and first-party setups keep reporting honest without crossing privacy lines.

Most dashboards get built, admired and ignored. Here is how to design metrics and review meetings that change what a team does next.

Time zones, contracting, quality and communication: how a U.S. front door backed by affiliate companies in Lima and San Jose actually works day to day.
A 45-minute working session, no slides.