Most student recruitment teams reached 2026 with a report that starts halfway through the story. The historical series stops at a fixed date, the older numbers no longer open, and the question that is left stopped being how to migrate.
Google Analytics 4 is now the only version of the tool in operation, and the transition that dominated the 2023 calendar is already behind us. Anyone searching for migration today is usually looking for something else: what was lost, what can still be recovered, and how to configure the new property so that it answers business questions.
What happened to Universal Analytics and Google Analytics 4?
Universal Analytics went offline in two stages, and both are already behind us. Standard properties stopped processing hits on July 1, 2023, and during the week of July 1, 2024 access to the data, to the interface and to the API (Application Programming Interface) ended for most users.
Universal Analytics 360 properties were different only in the deadline. Accounts with a current 360 order received a one-time processing extension, which ended on July 1, 2024. Properties on 360 without an active order fell back to standard behavior and stopped alongside the rest, in 2023.
Caption: Google Analytics 4 only measures what it starts collecting, and the history locked in the old archive does not come back.
The practical consequence is blunt: there is no UA (Universal Analytics) property to keep running in parallel, and no export left to do after the deadline. The API many teams relied on to pull reports automatically stopped responding as well.
Here is how the dates line up:
The line that matters most is the third one. The 2023 cutoff stopped collection, and the 2024 cutoff stopped consultation, which is what turns frozen data into unreachable data.
Can you recover Universal Analytics data today?
Inside Google Analytics, no. Google stated that Universal Analytics data not exported by the deadline would be deleted permanently, with no way to recover it. Outside the tool, though, more material tends to survive than most teams expect, and it is worth looking before accepting the gap.
The first place to check is the set of files your own team produced at the time. Exported spreadsheets, PDF reports sent to the board, Looker Studio dashboards built on extracted data and BigQuery backups often hold usable slices of the earlier period.
The second place is the sources that never depended on UA at all. Google Search Console keeps its own search performance history, Google Ads keeps campaign history inside the account, and the institution's CRM (Customer Relationship Management) usually holds the series of leads and enrollments by source.
The third place is the most forgotten one: the reports that turned into decisions. Meeting minutes, results presentations and campaign briefs frequently record the numbers behind a choice, with the date and the segment attached, even without the raw data underneath.
None of this rebuilds a complete series, and it is honest to say so. What these sources allow is an approximate comparison, good enough to answer whether today's volume sits above or below the level of two years ago.
How do you create a Google Analytics 4 property from scratch?
Creating a Google Analytics 4 property today is an initial setup job, with no comparison step against UA. It comes down to three linked moves: opening the property inside the account, creating the data stream for the site or the app, and installing the tag with the Measurement ID that GA4 generates automatically.
The property is born in the admin area of the Google Analytics account. It pays to set the correct time zone and currency from the start, because changing those fields later does not fix data already collected, only future data.
The data stream is what separates web from app. An institution running an applicant portal, an institutional site and a student app normally needs separate streams, which keeps behaviors that do not compare from being mixed together.
The Measurement ID appears as soon as the stream is created, in the format that begins with "G-". It is the identifier that ties the installed code to the property, and it has to reach every page of the site, including campaign landing pages hosted outside the main domain.
Installation can happen directly in the code, through a tag manager or through the native CMS (Content Management System) integration. Whichever path you take, the test is the same: open the realtime report and confirm that your own visit shows up there within seconds.
One last adjustment almost always slips through: data retention. The retention setting starts at the shortest period, and leaving it there quietly costs you depth in exploratory analysis.
How do you configure events and key events in GA4?
Google Analytics 4 organizes everything as events, and that is where setup stops being technical and becomes a business decision. An event records one specific user action, such as a button click, a form submission, a video view or a step completion. Marking the ones that matter turns behavior into a metric you can follow.
Part of the work arrives already done. Enhanced measurement automatically captures page scrolls, outbound link clicks, on-site search and file downloads, with no extra configuration.
What takes real work is the custom event, and it is what separates a generic report from a recruitment report. Application started, registration completed, document uploaded and WhatsApp button clicked are examples that describe the actual journey of an applicant.
Among the events you collect, some become key events, which is what GA4 calls an action that matters especially to the business. That set is what feeds conversions in Google Ads and what shows up in performance reports as a result rather than as traffic.
One practical rule keeps the list from inflating: a key event has to be something the institution would use to decide where to put money. Mark every click as a key event and the report drifts back into volume, while the conversion funnel loses its hierarchy.
Naming deserves a written agreement. Lowercase event names, one shared convention and a dictionary kept between marketing and technology prevent the scenario where three different names measure the same action and no report ever reconciles.
What do you do with the history stuck in Universal Analytics?
With no importable historical series, the way forward is to build a new baseline and be explicit about where it starts. What works in practice is treating the day collection began in Google Analytics 4 as a declared zero mark, documenting that in the report, and comparing periods only inside that window, without stitching numbers from different tools.
The comparison between UA and GA4 does not reconcile even when someone holds both spreadsheets. The measurement models differ, the session definition changed and attribution works another way, so a gap between the two tools is expected rather than a setup error.
That is why the recommendation is to compare behavior instead of absolute values. If organic traffic accounted for close to half of visits before and holds a similar share now, the strategic reading survives, even with different totals.
Protecting the history you are building right now is the other half of the job. Skipping that step is what puts the same institution back in this position a few years from now.
Three habits protect the new history: exporting core reports on a fixed schedule, keeping a dashboard with the closing numbers of each recruitment cycle and, once volume justifies it, turning on the BigQuery export. All three are cheap today and impossible to replace later.
Institutions that already work in a data-driven way feel the gap less, because decisions never rested on a single source. When a number lives inside one tool only, any discontinuation turns into a blackout.
How does Google Analytics 4 support CRO and conversion rate?
Google Analytics 4 is the evidence layer for CRO (Conversion Rate Optimization), the discipline of improving the conversion rate. The tool shows where the applicant gives up, how often that happens and which segment concentrates the problem, which turns an opinion about the page into a testable hypothesis.
The first move is building the funnel exploration. Tracing the path from course page visit to application start to form submission exposes the step that loses the most people, and that step is the natural candidate for the first test.
The second move is segmentation. The same page can convert well on desktop and poorly on mobile, or convert well on brand traffic and poorly on paid media, and segmenting your audience keeps the average from hiding the real problem.
The third move is comparing channels with judgment. Reading the conversion rate by channel shows that organic, paid, email and social rarely convert at the same level, and prioritizing the highest-volume channel is not always the choice that produces the most enrollments.
None of it works without a stable definition of conversion. If a key event changes criteria mid-cycle, the series breaks, and the CRO method loses its basis for comparing before and after the test.
The gains here tend to compound. Small fixes to a form, to a page promise or to the clarity of the next step often return more than a budget increase, because they act on traffic the institution already pays to receive.
How do you track organic traffic and user experience in GA4?
Google Analytics 4 splits acquisition and behavior into separate reports, and the useful reading comes from crossing the two. Acquisition answers where the person came from, behavior answers what they did after arriving, and it is the combination that tells you whether the content delivered what the headline promised.
On the acquisition side, the channel filter isolates organic traffic, which is the visit arriving from an unpaid search result and the central indicator of SEO (Search Engine Optimization).
Tracking organic traffic by landing page shows which content sustains recruitment without paid media, and building an SEO report in GA4 turns that tracking into a routine, with no manual export every week.
On the behavior side, the metrics that say the most about user experience are engagement per session, average time on page and scroll events. A page with heavy organic entry and low engagement tends to signal a mismatch between the query that brought the person in and the content they found.
User experience, usually shortened to UX, also shows up in quieter signals. A conversion drop concentrated on mobile, a form abandoned at the same field every time and on-site search repeating the same term all point to friction that no acquisition report reveals on its own.
Acquisition, behavior and experience form a single set. Techniques that combine CRO, SEO and UX treat acquisition and experience as the same problem, and Google Analytics 4 is where both ends meet as numbers.
One note of technical honesty: GA4 shows what happened, not why. To reach the cause, quantitative data needs session recordings, an A/B test or a conversation with the admissions team, who hear the objection before it becomes an abandonment.
Frequently asked questions about Google Analytics 4
Where do you start with Google Analytics 4 today?
Start with what is broken now, not with what was lost in 2023. If the Google Analytics 4 property already exists, the first task is auditing what it measures: confirm the tag is on every page, review the list of key events and set data retention to the longest period available.
If the property does not exist yet, the cost of waiting grows every month. Each recruitment cycle without measurement is a cycle that will not enter next year's basis for comparison, and that gap cannot be filled in afterwards.
Once collection is in order, the next step changes in nature. It stops being configuration and becomes method: pick one funnel step, state a hypothesis, test it and measure the effect on conversion rate, with a conversion optimization cycle running continuously instead of as a one-off project.
It is also worth reviewing what changed conceptually between UA and GA4 before training the team, because much of the resistance to the tool comes from looking in the wrong place for a report that now exists in another shape.
If your institution has the property live but no confidence in the numbers it shows, the problem is usually configuration and event definition rather than the tool itself. That is a quick diagnosis to run, and it tends to unlock everything downstream.
Portugal