How do I know whether the fix actually worked?

You need before-and-after evidence in the browser and in the downstream report. A clean-looking dashboard without validation is not enough.

When should this become a full GA4 audit instead of a quick fix?

If the issue touches attribution, consent, revenue, campaign quality, or data trust for more than one workflow, it is usually safer to audit the surrounding implementation than patch only the visible symptom.

GA4 Data Retention Pitfalls: The 14-Month Default That Breaks Year-on-Year Reporting (2026)

How long does GA4 keep data?

GA4 has two separate retention concepts. Standard reports retain aggregated data indefinitely — the Acquisition, Engagement, Monetisation reports always show all your historical data. Explorations and user-level data retention is configurable: default 2 months, maximum 14 months on standard tier (50 months on Analytics 360).

New properties default to 2 months — most teams discover this when they try to build a year-over-year Exploration and find the prior-year data is gone. The fix is one toggle in Admin → Data Settings → Data Retention. Set to 14 months immediately on every new property; the change applies forward, not retroactively.

The 2-month default is the single most common silent setup error in 2026.

What "data retention" actually controls

GA4's retention setting affects only user-level and event-level data used in Explorations. It does NOT affect:

Standard pre-built reports (Acquisition, Engagement, Monetisation, etc.) — these aggregate at the property level and persist indefinitely
Audiences (defined by user behaviour rules — they continue working with new traffic)
Lookback windows for Smart Bidding (governed separately)
Goals/conversions tracked at property level

What retention DOES affect:

Explorations beyond the retention window
User-level cohort analysis
Path exploration with custom date ranges beyond the window
Funnel analysis with comparison periods beyond the window
Some segment-level reports

The practical implication: most stakeholder-facing reports work fine with default 2-month retention. The teams that get burned are analysts trying to do quarterly cohort analysis or year-over-year deep-dives.

The retention options

GA4 Admin → Data Settings → Data Retention shows two settings:

Setting 1 — Event data retention

Options on standard tier: 2 months (default), 14 months. On 360: 2, 14, 26, 38, 50 months.

Default is 2 months. Change to 14 months immediately on every new property. No downside — Google doesn't charge for the longer retention. Forgetting this is the most common GA4 setup oversight.

Setting 2 — Reset user data on new activity

Default: ON. Recommended: ON.

When a user has new activity, their identifier window resets. So a user active in month 13 still has their full retention window. With this off, identifiers expire on their original schedule regardless of new activity — meaning long-term users get cycled out of cohort analysis prematurely.

For consent-related reasons, some GDPR-strict implementations turn this off. If you do, document why — analysts will eventually run into the consequences and need context.

Why the 14-month limit matters

14 months is just enough for year-over-year comparison plus the trailing month. Beyond 14 months (on standard tier), Explorations cannot reach the data — even though it's still present in GA4's backend.

For longer historical analysis, your only path is BigQuery export. BigQuery preserves data indefinitely (subject to your own table-level retention policies). A property with BigQuery export running from day one has a permanent historical record that's queryable forever.

Want to see which hidden implementation gaps are affecting your GA4 data quality?

Start free audit

The strategic implication: enable BigQuery export early, even if you don't immediately need it. Storage is cheap (typical $20–$200/month for active GA4 properties); the historical data becomes invaluable when stakeholders ask "what did Q4 look like 3 years ago?"

What changing retention doesn't do

Three patterns that look like retention should fix but don't:

Pattern 1 — Retroactively recover old data. Changing from 2 to 14 months today does NOT bring back data older than 2 months. The setting applies forward only. Data already aged out is gone.

Pattern 2 — Fix Standard report inconsistencies. Standard reports retain aggregated data indefinitely already. If your Acquisition report shows weird historical numbers, retention isn't the cause — usually it's a sampling, channel-rule, or attribution model change.

Pattern 3 — Extend lookback windows for attribution. Attribution lookback is a separate setting (Admin → Property settings → Attribution settings). 30 days for acquisition events, 90 days for other conversions. Retention doesn't control this.

Standard reports: always available

The most reassuring part of GA4's retention: standard reports work even without long retention. If your stakeholders only look at:

Traffic Acquisition (sessions by source/medium over time)
User Acquisition (new users by source/medium over time)
Engagement (pages, events, key events)
Monetisation (revenue, ecommerce metrics)
Demographics (age, gender, interests)

These work for years of data regardless of retention setting. Retention only matters for analyst-driven deep-dives via Explorations.

So the 2-month default is genuinely fine for stakeholder reporting. It's just bad for any analyst work — and analysts find out at the worst possible moment.

The audit pattern

For any GA4 property you inherit:

GA4 Admin → Data Settings → Data Retention.
If "Event data retention" shows "2 months", change to "14 months" immediately.
Verify "Reset user data on new activity" is ON unless there's a specific compliance reason it's OFF.
Document the change date in your annotation log. Analysts will eventually want to know which historical data has 14-month retention vs which has 2-month.
Plan for BigQuery export. Even at low traffic, the long-term value of permanent historical data is significant.

This audit takes 90 seconds. It's the single highest-ROI configuration check on any GA4 property.

When 14 months isn't enough

Three scenarios that genuinely need beyond-14-months data:

Scenario 1 — Annual deep-dive analysis. "How does Q1 2026 compare to Q1 2024?" requires 24+ months of data. BigQuery is the only path on standard tier.

Scenario 2 — User cohort analysis with long behaviour windows. B2B with 18-month sales cycles needs to track cohort behaviour beyond GA4's standard retention. BigQuery query-based cohorts work regardless of retention.

Scenario 3 — Compliance audit trails. Regulated industries may require multi-year data retention as part of compliance. BigQuery with appropriate retention policies satisfies this; GA4 standard tier alone doesn't.

For all three, the answer is the same: BigQuery export from day one. It's not optional infrastructure for properties that will need historical depth.