You open a GA4 Exploration to review last year’s campaign, but the earlier months are missing. A standard report still shows figures for the same period. Before assuming tracking failed, check which reporting area you are using, the property’s retention settings and the Exploration itself.
This matters when you are deciding what to repeat next season. Missing detail should not be treated as evidence that a campaign produced no enquiries or sales. Equally, an old headline total does not prove you can still reconstruct the customer journey behind it.
Why reports can show dates that Explorations cannot
Google confirms that Explorations are limited by the property’s user-level and event-level retention window. Opening an older report in Explore does not bring data outside that window into the Exploration.
Start by writing down the actual question. “How much revenue did the campaign generate?” and “Which sequence of pages did mobile buyers follow before purchasing?” require different evidence. Try the simplest suitable report before rebuilding a complicated Exploration.
For example, a retailer preparing its Christmas range might still be able to compare overall trading periods, while finding that the detailed analysis it wanted to run for last December is unavailable. Record those as separate findings. Do not replace a missing journey analysis with a confident explanation based on totals alone.
1. Check the property and dates first
Record the property name and ID, the requested dates, the report or Exploration name and the exact symptom. Is the calendar restricted, is the table empty, or are only some rows missing? Those are different starting points for an investigation.
Check when this property actually began collecting the relevant website events. Ask whether a rebuild or agency handover introduced another property, and whether the event you need existed during the period being reviewed. Look for a dated implementation record or previous report rather than relying on memory.
Then inspect the date range inside the Exploration. Google’s Explorations guide explains that a saved preset such as “Last 28 days” moves forward when reopened, while a custom range remains fixed. A saved Exploration is therefore not automatically a frozen record of the period you originally reviewed.
Make a working copy before changing an analysis colleagues depend on. Use one recent, completed week as a control, then test the older period with the same simple setup. This gives your analyst a repeatable comparison.
2. Read the retention setting before changing it
The Explorations guide also identifies a two-month default and places Data retention under Admin, then Data collection and modification. Check the value actually saved in your property; do not assume somebody extended it during setup.
Google’s current data retention guidance distinguishes standard aggregated reports, which this setting does not affect, from Explorations and funnel reports, which it does. Standard properties normally offer two or 14 months for event data; properties classified as Large are restricted to two months.
Increasing retention applies to previously collected data that has not already been deleted. It cannot restore deleted detail. Changes take effect after 24 hours, and expired data is deleted through a monthly process. The “reset on new activity” option concerns user-level data; it does not keep every historical event indefinitely.
Have an Editor check any proposed change. Record the old value, new value and time saved, then verify after the implementation delay. Choose a period that fits your reporting needs and agreed data-handling arrangements. A longer setting needs a clear purpose.
3. Rule out an empty query
If the dates are available but the table is blank, simplify the query before concluding that the historical data has gone. Remove segments and filters from your working copy, then try a basic free-form table using Event name and Event count.
Google’s data compatibility documentation explains that incompatible dimensions, metrics or techniques can produce an empty report, unavailable fields or an “Incompatible request” warning.
Restore your settings one at a time. Keep a note of which addition makes the data disappear. That turns “GA4 is broken” into a useful request such as “this filter removes all rows for this date range”.
Check spelling and capitalisation too. Google’s reports-versus-Explorations guidance notes that Exploration filter expressions are case sensitive, whereas the search box above a standard report’s table is not. Copying what appears to be the same search between them may produce a different selection.
4. Read the data quality warning
A few missing rows require a different check from a whole missing year. Inspect the data quality indicator and record its message before altering the analysis.
Google explains that data thresholds can withhold information to protect users, including when demographic information is involved. These thresholds are system defined. Expanding a narrow date range may make previously withheld data available.
Test a broader period that remains inside the available retention window, and remove demographic breakdowns if they are unnecessary to the business question. Document whether the warning changes. Treat withheld data as unavailable, rather than entering zero into your planning spreadsheet.
If the indicator mentions sampling instead, record that separately. The Explorations guide describes both sampling and threshold warnings. Neither message, by itself, establishes that retention deleted your older data.
5. Use the history you can still verify
Build a short evidence list for the decision you need to make:
- Which standard reports still answer part of the question?
- Which dated reports or exports did your team retain?
- What do the shop’s order records or CRM say about actual outcomes?
- Which questions remain unanswered because the necessary detail is unavailable?
Compare equivalent trading periods, and annotate changes in offers, stock, website tracking and business processes. An apparent improvement deserves investigation if the definition of a successful enquiry changed between the two periods.
Keep each source’s meaning clear. CRM sales, website purchases and attributed GA4 events should not be silently merged into one total. Our guide to connecting conversion reporting with enquiries and sales explains how to build a more useful commercial view.
Would BigQuery solve this?
For businesses that repeatedly need detailed historical analysis, ask whether an existing BigQuery export covers the missing dates. Google’s BigQuery Export documentation describes exporting raw events for separate analysis. It also warns that export data can differ from the Analytics interface.
Check actual date coverage and completeness. Google’s export setup guidance describes data beginning to flow after linking, with daily exports containing the previous day’s data. Do not treat a newly created link as a recovery method for deleted historical events.
Consider this as a reporting project: decide what questions justify it, who will maintain it, what access is appropriate and how storage and query costs will be controlled. A smaller business may get more value from a dependable monthly decision report than an unused warehouse.
Make the next seasonal review easier
Agree which comparisons you want before the next campaign starts. Keep dated outputs with their property ID, period, filters, metric definitions and known limitations. Retain the conclusions and business context alongside the figures, so another person can understand why a decision was made.
After any repair, repeat the recent-week control and the older-period test. Record what is now available, what remains unavailable and why. A successful investigation may end with a corrected filter, an appropriate retention setting or a clearly documented historical gap.
If you need help separating those possibilities, BuzzBoost’s analytics and reporting service can help you turn the evidence into a practical measurement plan. Tell us which comparison you need to make and where the data appears to stop.
Featured image: AI-generated editorial artwork, not a photograph of a real BuzzBoost office, client or result.
