Why does Google only keep your performance data for 16 months?

What exactly does Google store for 16 months?

Google archives every click and impression recorded on its results pages for your site. This includes the queries typed by users, the displayed URLs, average positions, click-through rates, countries, and devices. All of this data feeds into the Search Console performance reports.

This 16-month retention starts when the event occurs. An impression from January 1 will be accessible until April 30 of the following year, after which it will disappear. No permanent archive exists on Google's side for site owners — after this period, the data is permanently purged.

Why is there a limit of 16 months and not more?

Google typically cites storage constraints and GDPR compliance. Processing billions of daily results generates a colossal volume of indirect personal data (queries, behaviors). Limiting retention reduces legal risks and infrastructure costs.

For SEO professionals, this duration is sufficient to observe complete seasonal trends (12 rolling months + margin). However, it becomes problematic for long historical analyses, site recovery audits, or comparisons with periods prior to the past year.

How can you concretely access this 16 months of data?

The performance reports in Search Console allow you to select any range within the last 16 months. You can compare two periods, filter by query, page, country, device. The interface displays up to 1,000 rows by default; the CSV/Google Sheets export removes this limit.

The Search Console API offers programmatic access to the same data, with a limit of 25,000 rows per query. To extract the entire data of a large site, you need to split API queries by date range or dimension. This is the API that third-party tools (Semrush, Ahrefs, Data Studio) use to centralize and archive your data beyond 16 months.

• Rolling 16-month window: each day, the oldest day disappears
• Raw data available via interface and API: clicks, impressions, CTR, average position
• Manual or automated export is essential to retain history beyond Google's limit
• No recovery possible once the 16 months have passed — permanent loss
• Daily granularity maintained: you can analyze day by day over the entire period

Is this statement consistent with observed practices in the field?

Yes, the 16-month limit has been empirically verified for years. All SEO practitioners who try to compare data beyond this time frame find it impossible to access previous periods. Google does not offer any "archiving mode" or paid extension of this window.

However, one point that is rarely highlighted: the quality of data sometimes degrades on older periods within this window. Some filters or dimensions may become unavailable, and aggregations might present minor discrepancies. [To be verified]: Google has never officially documented whether the data from the initial months of the window undergoes different processing (sampling, compression).

What concrete issues does this limit pose for SEOs?

The most frustrating case: the audit of a reclaimed site. You acquire a client whose site has existed for three years, but you only have access to the last 16 months. It's impossible to know if a traffic drop observed 14 months ago is due to a penalty, a failed migration, or normal seasonality that's absent from your visible window.

Another concern: year-over-year comparison beyond 12 months. If you want to compare February N to February N-2, it's impossible via Search Console. Only tools that continuously archived your data can do this — assuming you connected them before the period you wish to analyze.

Should you systematically archive this data yourself?

For a professional site generating organic revenue, yes, it is an operational obligation. The traffic history is a strategic asset: it documents the impact of your SEO actions, proves your ROI, and alerts you to anomalies. Losing this history means losing your SEO memory.

Several approaches coexist: monthly manual export (time-consuming, risky), automated scripts via the Search Console API (requires dev skills), or outsourcing to third-party tools like Google Data Studio with connector, Semrush, Ahrefs, or dedicated solutions like Search Analytics for Sheets. The choice depends on your data volume and technical stack.

What should you implement immediately to avoid losing your data?

First action: connect Search Console to an automatic archiving system. If you're using Google Sheets, install the "Search Analytics for Sheets" add-on and schedule a weekly or monthly extraction. If you have access to BigQuery, create a daily export via Cloud Functions or Apps Script — you will then have a queryable SQL database indefinitely.

Second reflex: document your SEO actions with timestamps. Log in a shared file (Notion, Google Docs, Trello) every migration, redesign, link-building campaign, structural change. When you analyze your curves in six months, you'll be able to correlate traffic variations with your interventions. Without this memory, you'll be navigating blindly.

What mistakes should you avoid in managing this history?

Classic error: only exporting monthly aggregates. You then lose daily granularity, which is essential for detecting spikes and abrupt drops. Always export at the day level, even if you aggregate it yourself later. The opposite is impossible.

Another trap: not versioning your exports. If you overwrite the same file each month, you risk losing data in case of a bug or misoperation. Date your files (e.g., gsc_export_2024-03.csv) or, better, version them in a Git repository or a Drive with history.

How can you check that your archiving is working correctly?

Test the continuity of your data: select a period covering two successive exports and verify that the figures overlap without gaps or duplicates. Compare the sum of clicks over a week in your archive with the figures displayed in Search Console — they should be identical (allowing for freshness: Search Console updates data with a 2-3 day latency).

Also check the dimensional coverage: if you export by query, ensure you capture anonymized queries, data by device, by country. Some homemade scripts miss essential dimensions. It's better to have a modular architecture where each dimension is exported into its own table, then joined in an SQL query.

• Install an automatic export system (API, Sheets, BigQuery) before the end of the 16-month window
• Export at the daily level, not monthly, to maintain maximum granularity
• Document each SEO action with date and context in a shared logbook
• Version your exports to prevent any accidental loss of historical data
• Test the continuity of data between successive exports to detect gaps or duplicates
• Audit dimensional coverage (queries, pages, countries, devices) to avoid missing anything