Google Trends for Keyword Research: How to Read It Correctly
Google Trends shows relative interest in searches over time, not search counts. That makes it excellent for comparing terms, spotting seasonality and checking direction, and misleading if you read the numbers as volumes. This guide explains what the data really is and how to use it.
- What 0-100 means
- Compare terms
- Seasonality
- Data limits
term A ######## 100 term B #### 45 (relative, not volume)
Trends uses a sample of real Google searches and turns each figure into a share of total searches for that place and time, rescaled to 0 to 100. The numbers are relative, not volumes. Compare terms against each other, check seasonality, and use a volume source for sizing. Low-volume terms show as 0.
What the data is
Per Google, Trends uses a lightly filtered sample of real searches with personal identities removed, queries assigned to topics and results grouped, allowing views by country down to city. Sampling is used because the volume of searches is too large to process in full; the sample aims to represent overall activity.
What 0 to 100 means
- Each figure starts as a share of total searches for that location and time, so high-volume places do not automatically rank highest.
- Shares are rescaled to a 0-100 range.
- Two regions with the same score can have very different total search volumes.
- Read the numbers as relative; Google’s page does not define the top value beyond the scaling.
Comparing terms
| Question | How to answer it (partly my own practice) |
|---|---|
| Which phrasing is more popular? | Compare the two terms in one chart |
| Is a topic rising or falling? | Use a long range and look at direction |
| Is it seasonal? | Look for repeating peaks across years |
| How big is it? | Compare against a familiar benchmark term; use volume tools for sizing |
Google stresses Trends is not a scientific poll: a spike shows search interest, not popularity or that a topic is winning.
Time ranges and time zones
- Ranges of 30 days or longer use UTC with daily, weekly or monthly granularity.
- Ranges of 7 days or less use the local time on your browser or device, typically hourly.
- The page does not address ranges of 8 to 29 days.
What is filtered out
- Low-volume terms appear as 0.
- Repeated searches by one person over a short period are removed.
- Queries with apostrophes or other special characters are filtered.
- Searches by Google’s own products, including AI Mode and AI Overviews, are excluded.
- Irregular activity is filtered, though some may remain, and statistical noise is added for privacy, most visible for low-interest queries.
Research step by step
Compare phrasings
Up to what the tool allows.
Set a long range
See seasonality.
Set the location
Match your audience.
Plan publishing dates
Ahead of peaks.
Using it for SEO (my own guidance)
- Choose between synonyms before settling on a focus keyword.
- Publish and refresh seasonal content before the usual peak (updating posts).
- Pair with autocomplete and People Also Ask.
Notes for WordPress sites
- Editorial calendar: note peak months beside post titles.
- Refresh before peaks rather than after.
- Check Search Console for your own seasonal curve.
Troubleshooting
| Symptom | Likely cause | Action |
|---|---|---|
| Term shows 0 | Low volume (per Google) | Try broader terms |
| Odd spike | Noise or event (per Google) | Check context; do not over-read |
| Term with apostrophe missing | Special characters filtered (per Google) | Remove the character |
| Hard to size demand | Values are relative | Use a volume tool |
Checklist
Setup
- Terms listed
- Region set
- Range set
- Category considered
Read
- Relative not absolute
- Compare terms
- Check seasonality
- Ignore tiny spikes
Cross-check
- Volume tool
- Search Console
- SERP intent
- Autocomplete
Act
- Pick wording
- Plan timing
- Refresh content
- Record decisions
What Google’s page does not say
It does not cover related queries and topics, category definitions, rising or breakout labels, the number of terms you can compare, or data refresh frequency beyond minutes after real-world events.
Key terms
- Relative interest
- Share scaled to 0-100.
- Sampling
- Using a subset of searches.
- Seasonality
- Repeating yearly pattern.
- Benchmark term
- Familiar term for scale.
Common mistakes
Reading scores as volumes
They are relative.
Treating spikes as popularity
Only interest.
Too-short ranges
Hides seasonality.
Niche terms
Show as 0.
Wrong region
Misleading shape.
Ignoring intent
Popular is not relevant.
Related guides
Google Trends questions
Does Trends show search volume?
No; values are relative.
What does 100 mean?
The peak of the scaled range; the page does not define it further.
Is the data a sample?
Yes, a sample of real searches.
Why does my term show 0?
Low-volume terms show as 0.
Do AI Overview searches count?
Google’s own product searches are excluded.
Which time zone is used?
UTC for 30 days or more; local for 7 days or less.
Is a spike proof of popularity?
No, only search interest.
How fresh is the data?
Analyzable within minutes of events; refresh details are not given.
Data behaviour comes from Google’s page. The comparison table, SEO use and steps are my own.