Using it

1. Reading the numbers
2. Time ranges
3. Terms, operators and topics
4. Locations
5. Search types
6. Good habits
7. Prompts

Reading the numbers

Everything Google Trends returns — apart from Trending Now’s search volumes — is an index from 0 to 100:

1. Google takes the share of all searches that a term had at each moment (or in each place).
2. It rescales the whole answer so the highest point in it is 100.

So a value is never a count of searches. What follows from that:

Only terms in one request are comparable.	Two separate calls are on two separate scales. compare_many, share_of_search, compare_periods and compare_locations build a single scale for you.
Across places it is a share of local searches.	A small country can outrank a large one. The place with 100 is where the term matters most, not where most people searched.
<1 and 0 are different.	<1 is a trace of interest; 0 is too little to measure.
The last point is usually partial.	It is still being collected. Tables mark it with * and averages leave it out.
It is a sample.	The same request on another day can differ by a point or two.

Time ranges

You write	You get
1h 4h	one point per minute
1d	one point per 8 minutes
7d	hourly
1m 3m, or any range under about 9 months	daily
12m 5y, or up to 5 years	weekly
all, or longer than 5 years	monthly, back to 2004

Also accepted: 6m, 2y, 45d · a year 2024 · a month 2024-03 · years 2019-2023 · explicit dates 2024-01-01 2024-12-31 · an hourly range 2026-09-28T00 2026-09-30T00 (at most 7 days).

For daily data over a longer range, use daily_history: it joins overlapping windows onto one scale.

Terms, operators and topics

A search term matches searches containing those words, in that language and spelling.

Write	Meaning
tesla model y	searches containing all three words, in any order
"tesla model y"	that exact phrase
tesla + byd	either
jaguar -car	jaguar without “car”

A topic is Google’s own grouping of every search about one thing, in any language. find_topic returns topic ids such as /m/0dr90d (Tesla, the company); pass one wherever a keyword goes. Do not compare a topic with a search term — they are measured differently.

A category narrows an ambiguous word: “jaguar” in Autos & Vehicles is the car. find_category gives the ids. With no keyword at all, a category id measures the whole category.

Locations

Locations are codes: US, IR, DE; regions US-CA, IR-07; US metro areas US-CA-807. find_location finds them by name or lists what is inside one. Leave geo blank for worldwide — except in the Trending Now tools, which need a country.

Set GTRENDS_GEO to make one location the default.

Search types

property is one of web (default), youtube, news, images, shopping.

Good habits

• Several keywords in one call. interest_over_time takes five; trend_momentum and content_calendar eight; compare_many twenty-five. One call each is slower and more likely to hit a rate limit.
• Start with keyword_overview. It is the whole Trends page in four requests.
• A term is matched literally. If people write it in more than one way, join the wordings with + in one term, or use a topic id.
• Use a long range for verdicts. Twelve months cannot tell a season from a trend; trend_momentum and seasonality use five years for that reason.

Prompts

Four workflows appear in your client’s prompt menu:

Prompt	Arguments	Result
trend_report	keyword, geo	Direction, seasonality, the events behind the spikes, rising themes, recommendations
newsjacking_brief	subjects, geo	Today’s trends that fit your subjects, with the story, the queries, an angle and a risk note
seasonal_content_plan	topics, geo, lead_weeks	A twelve-month publishing plan
market_comparison	brand, competitors, geo	Share of search, who wins where, what is rising around each brand