# Is Google Trends accurate? How sampling and noise shape the data

> Google Trends is accurate as a relative measure of search interest, and sampling and noise shape the data in 3 known ways: it uses a sample of Google searches, it adds statistical noise to protect privacy, and it reports terms with very low volume as 0. Google calls the data a largely unfiltered sample of actual search requests, not a perfect mirror of search activity.

Canonical: https://kwerix.com/trends/accuracy/

## How accurate is Google Trends data?

Google Trends data is accurate as a relative measure for popular terms, and less accurate for terms with little search interest. Google uses a sample of searches because it handles billions of searches a day, and the sample is large enough to represent all Google searches, according to Google Trends Help.

## Why does Google Trends data change between requests?

Google Trends data changes between requests for 2 reasons that Google documents.

1. **Sampling.** Google Trends draws on a sample of Google searches, not the full set, so two requests can read slightly different samples.
2. **Statistical noise.** Google adds small, random fluctuations to protect privacy, and the noise is most noticeable on queries with low or no search interest.

One-off spikes on low-interest terms can be noise rather than real search activity, and Google advises against reading them as evidence of real searches.

## Which searches are filtered out of Google Trends?

4 types of searches are filtered out of Google Trends, according to Google Trends Help.

1. **Searches made by very few people**, which appear as 0.
2. **Duplicate searches** from the same person over a short period of time.
3. **Queries with special characters**, such as apostrophes.
4. **Searches made by Google products and services**, including internal searches made by AI Mode and AI Overviews.

Google filters most irregular activity, such as automated searches, and in rare cases keeps some of it in the data as a security measure, so Google Trends is not a perfect mirror of search activity.

## When is Google Trends data unreliable?

Google Trends data is unreliable in 3 conditions.

| Condition | Why | What to do |
|---|---|---|
| Terms with little search interest | Noise and the low-volume threshold dominate | Compare with a common term or widen the location |
| Short ranges on small terms | Few searches per hour or day | Use a longer range, such as 12 months |
| Single spikes | Noise or irregular activity | Read the average and the trend, not one point |

## How to read Google Trends data reliably

Google Trends data reads reliably when a term is compared with common search terms, such as types of food, celebrities or weather, which is the scale check Google recommends. Averages over a full range and trends per year are steadier than single points, and [the Google Trends index](/trends/index/) explains what each value means.

## Google Trends accuracy FAQ

## Frequently asked questions

### Is Google Trends a poll?

No. Google Trends reflects search interest in topics, not opinions, and a spike shows that many people searched a topic, not that it is popular or winning, according to Google Trends Help.

### Does Google Trends include searches from AI Overviews?

No. Google Trends filters out searches made by Google products and services, including internal searches made by AI Mode and AI Overviews.

### Is Google Trends data free?

Yes. The Google Trends website is free to use; extracting the data at scale through an API has a cost, such as $1.50 per 1,000 keywords with the Kwerix Google Trends API.

## Sources

- [Google Trends Help. FAQ about Google Trends data](https://support.google.com/trends/answer/4365533)