GA4 says	Kehai says	The difference that matters
users	visitor days	GA4 counts people across days with a cookie. Kehai's visitor code is scoped to one UTC day; daily codes are counted again on the next UTC day and summed as visitor days.
sessions	visits	Both use a 30-minute inactivity rule, but Kehai also ends visits at midnight UTC. Consent configuration and delivery can also change which activity either tool records
views	pageviews	Near enough the same thing. Kehai counts SPA route changes without extra setup
engagement rate	bounce rate, inverted	GA4 calls a session engaged after 10 seconds, a conversion, or two views. Kehai's bounce is one definition: a visit with a single pageview
average engagement time	average visit	Both count visible time only. Kehai sums it per visit and divides by visits
events	custom events	In GA4 everything is an event, pageviews included. In Kehai a pageview is a pageview, and events are the things you chose to name
conversions, key events	goals	A goal is one outcome with a daily series and a conversion rate, configured on a screen rather than marked on an event
purchase revenue	net revenue	GA4's value is whatever the tag sent. Kehai's net_amount comes from your server, is validated against the items, and never mixes currencies
session default channel group	channel	Both classify per visit. Kehai's eleven channels have a stated evidence order, and AI assistants are their own channel ahead of search

Why the numbers differ

Put the same site in both tools and the counts will not match to the decimal. The reasons are structural, and knowing them beats staring at the gap:

GA4 estimates, Kehai counts. GA4's user counts are approximate by design, and on busy properties its reports can be sampled and its rows thresholded away for privacy. Kehai's counted numbers are exact sums over stored rows.
Consent changes the population. GA4 behavior depends on its consent-mode configuration and eligibility for modeling. Kehai does not model missing traffic. Your own consent assessment and collection setup determine which activity reaches it. Neither result is guaranteed to be higher.
Different definitions produce different rates. An engaged session and a non-bounce are not the same claim, so the two rates are not comparable digits, only comparable directions.
Days cut differently. Kehai cuts at your site's midnights everywhere. A GA4 property's time zone and processing delays can move a visit's day.

GA4 ecommerce payloads need an adapter before delivery to Kehai. In GA4, item price is after the per-unit discount. In Kehai, it is before the discount, and discount is the total for the line. The ecommerce guide provides the conversion and delivery contract. Similar field names do not mean a one-to-one payload mapping.

Where monthly uniques went

Kehai has no monthly unique visitors, for any origin, because its reports count daily visitor codes rather than a stable person across days. Visitor days name the daily counts, and their name tells you what summing them means. If your reporting genuinely needs a cross-day unique count, Kehai is the wrong instrument for that one number, and it says so here rather than letting you discover it in a meeting.