Data thresholding: when analytics hides data to protect privacy

You run a report for traffic from a specific city and see "(not provided)" instead of actual numbers. Google is hiding the data because too few users came from that location. This is data thresholding—hiding data that might identify individuals.

What is data thresholding

Definition

Data thresholding is when analytics tools hide detailed data because the sample size is too small. If fewer than 3 people visited from a location or keyword, the tool shows "(not provided)" or "other" instead of revealing the specific data.

Why tools do it

Privacy protection. If the tool shows exact numbers for every city, someone could infer who visited from a small town. If your site gets 5 visitors from Smalltown USA, showing that number might reveal identities. Hiding it protects privacy.

It's increasingly common

Apple's privacy features, Google's differential privacy initiatives, and data protection regulations all push tools toward thresholding. Expect more data to be hidden over time.

Where you see thresholding

In keyword reports

"(not provided)" keywords are either encrypted by search engines or hidden by Google due to small sample size. You can't see which keywords brought traffic, only that they existed.

In geographic reports

Specific cities disappear. Only larger cities show. Small geographic areas are rolled into "other" to prevent identification of rare users.

In audience demographic reports

Age and gender data disappears if sample sizes are small. You see percentages for large groups, actual numbers are hidden for small ones.

In custom dimensions

Custom dimensions you create are thresholded too. If a user property is rare, it's hidden to protect privacy.

The impact of thresholding

Incomplete data

You can't see all your data. A portion is hidden. Analysis is always incomplete.

Lower totals

When thresholded data is removed, your total shows lower numbers. If 5% of traffic is hidden, your total traffic count is 5% low.

Broken attribution

Hidden keywords can't be tracked. If a keyword generates traffic but Google hides it, you can't attribute conversions to that keyword.

Changed segment behavior

Niche segments look worse because their data is hidden. A long-tail keyword that converts well might not appear at all.

How to work with thresholding

Expect it and plan for aggregate analysis

Don't try to see every small segment. Plan analysis at the aggregate level. Don't ask "which exact keyword." Ask "what's the performance trend across keyword groups."

Use larger groupings

Instead of reporting by city, report by region. Instead of specific keywords, report by keyword categories. Larger groupings avoid thresholding.

Combine tools

Use Google Search Console for keyword data (some searchc queries show there that don't in Analytics). Use CRM for customer geographic data (not thresholded). Combine sources to see the full picture.

Increase traffic to reduce thresholding

Larger sample sizes trigger thresholding less. As your traffic grows, more detail becomes available. This is one reason why analytics becomes more useful as sites scale.

Frequently asked questions