Learn · Data and content

How long should you keep analytics data?

Keep the aggregate reports for as long as you like and the raw visitor-level data for as short a time as answers your questions. GA4 lets you set event retention to either 2 or 14 months, and the longer setting is not automatically better if nobody ever queries past a year.

Retention is a privacy decision as much as a storage one. Under GDPR and the regimes modelled on it, personal data should be kept no longer than the purpose you collected it for requires, and "we might want it one day" is not a purpose.

The practical rule is to decide which questions you answer each year, keep enough to answer them, and export the summaries before the raw data ages out. Almost nobody does the export, and then the retention setting quietly deletes the comparison they wanted.

Updated September 2026

Learning Center More on data and content

First-party vs third-party analytics
Third-party analytics such as GA4 sends visitor data to somebody else's servers and gives you their reports. First-party analytics keeps that data on infrastructure you control. The trade is between depth of features and who holds the record of who visited you.
What is GA4 (Google Analytics 4)?
GA4 is the current version of Google Analytics, the free tool for measuring website and app traffic. It replaced the older Universal Analytics in 2023, so it's what every site uses now. It's event-based, tracks users across web and app, and is built around privacy changes like cookie limits.
SQL or NoSQL: which do you need?
For most business sites and apps, SQL. A relational database like PostgreSQL enforces the shape of your data and can answer questions you had not thought of when you designed it. NoSQL earns its place at very large scale, or when records genuinely have no fixed shape.
What is a data model, and why plan it?
A data model is the decision about what things exist in your system, what fields each one has, and how they relate to each other. It takes an afternoon to sketch and it governs what the software can do for years afterwards, which is why skipping it is the expensive option.
What is a content model?
A content model is the set of content types a site has and the fields inside each: a service with a name, a summary and three benefits; a case study with a client, a date and images. It is the structure of your content, separate from whichever tool you edit it in.
Structured content vs a page builder
A page builder stores a finished layout. Structured content stores the facts and decides the layout when the page is rendered. The difference only appears when you want that content somewhere else: a listing, an email, an app, or a rich search result, none of which a stored layout can supply.

Want this built, not just explained?

Start the conversation

How long should you keep analytics data?

First-party vs third-party analytics

What is GA4 (Google Analytics 4)?

SQL or NoSQL: which do you need?

What is a data model, and why plan it?

What is a content model?

Structured content vs a page builder

Want this built, not just explained?