Analytics and Tracking

Filter

Also called data filter, view filter

A rule that keeps or drops rows of data, permanently when applied at collection or temporarily when applied to a report.

Quick facts: Filter

Category
Analytics and Tracking
Also called
data filter, view filter
Level
Intermediate
Affects
Data accuracy, internal traffic, historic comparability
Where to see it
GA4 data filters, Google Tag Manager, Looker Studio report filters
In this article4
  1. How a filter works
  2. Why filters matter
  3. Where filters go wrong
  4. How to act on it

How a filter works

A filter is a rule deciding whether a row of data is kept or dropped. The distinction that matters is when it is applied. A collection-level filter runs before the data is stored, so anything it excludes is gone permanently. A reporting-level filter runs when you open a report, so it only changes what you are shown and can be lifted at any time.

Typical collection filters exclude internal traffic, either by network address or by a flag set in the tag, and exclude known developer or staging visits. Typical reporting filters narrow a table to one country, one campaign or one page path while you investigate something specific, and are removed the moment you are finished.

Why filters matter

Internal traffic is the usual reason people need one. On a small business site where the owner, the staff and the developer all visit daily, a meaningful share of sessions comes from the business itself, and every rate calculated from those sessions is wrong. Excluding them makes the remaining data describe customers instead of colleagues.

Filters also keep test and staging environments out of live reporting. Once the same measurement setup exists on a staging copy of the site, excluding by hostname is what stops rehearsals being counted as business.

Where filters go wrong

Permanent exclusion is the real danger. Data removed at collection cannot be recovered, and a badly written rule can empty a property silently — you discover it weeks later with nothing to restore. Where the tool offers a testing mode, run every new filter in it first, and keep one unfiltered copy of the data as insurance.

Changing network addresses are the second problem, and a common one in Nepal, where home and office connections are frequently reassigned. An address-based internal filter goes stale without warning and starts letting internal visits back in. A flag set in the browser or in Google Tag Manager survives that far better.

The third is forgetting a reporting filter is switched on. A figure quoted from a filtered view and then repeated as a company-wide number is a mistake that travels quickly and is embarrassing to unwind.

How to act on it

Filter as little as possible at collection, and as much as you like at reporting. If a question can be answered by narrowing the view rather than deleting the data, do that instead: a segment is reversible and a collection filter is not.

Document every filter that is live — what it excludes, why it exists, who added it and when. When numbers later look wrong, an undocumented filter is one of the hardest causes to track down, because nothing in the report announces that it is there. A short note in the measurement plan costs minutes and saves days.

Do and do not

Do

  • Test a new collection filter before making it active
  • Exclude internal traffic with a tag flag
  • Document every live filter and who added it

Do not

  • Exclude data at collection when segmenting would do
  • Quote figures from a filtered view as totals
  • Rely on a fixed network address in Nepal

Questions people ask about this

Should I filter out my own visits?

Yes, for most small sites. Staff, owners and developers visiting daily distort rates and can make a quiet week look busy. Exclude them where your tool supports it, ideally with a browser flag or a tag-based rule rather than a network address, and keep the exclusion documented so the gap in the data is explainable later.

Can I recover data that a filter removed?

Generally no. A filter applied when data is collected prevents the row from ever being stored, so there is nothing to restore afterwards. That is why every new filter should be tested first where the tool allows it, and why keeping one unfiltered copy of the data is usually worth the extra property.

Is a filter the same thing as a segment?

No. A segment narrows a report while leaving the stored data intact, so you can apply and remove it freely without consequence. A filter can act at collection time and permanently exclude rows from the property. Use segments to explore, and reserve filters for data you are confident you will never need.

Related terms

Found this useful?

Share it, or ask an AI to summarise it

Back to the glossary

Knowing the term is the easy part

Applying it to your own site and budget is the work. Book a call and I will tell you what actually applies to you.