Is there a Google Trends API?

What Google does and does not offer, why Trends numbers break when you compare them wrong, and the working routes.

The short answer

Google does not offer an open, self-service Trends API that you can sign up for and start calling. Access has been limited, which is why every practical route ends up being either the public endpoints behind the Trends website or a provider who calls them for you.

The mistake almost everyone makes

Trends numbers are not absolute search volumes. They are relative values from 0 to 100, scaled inside the request you made. That has a consequence people discover too late: if you fetch "electric car" in one request and "hybrid car" in another, the two numbers cannot be compared. Each was scaled against its own peak. Put both terms in the same request and you get one shared scale, and only then does a comparison mean anything.

This is the single most common reason a Trends dataset turns out to be wrong after someone has already built a dashboard on it. If your question is "which term is bigger", the terms must travel together.

Route 1: an open-source wrapper

There are Python wrappers around the same endpoints the website uses. Free, popular, and they work until they do not. Two recurring problems: the endpoints change without notice, which breaks the wrapper until someone patches it, and sustained calling from one address gets rate limited quickly. For a one-off analysis that is fine. For anything that has to run tomorrow as well, you are now maintaining someone else's reverse engineering.

Route 2: a provider

Several providers sell Trends data through an API. You trade money for not maintaining the plumbing. What to compare: whether they let you put several terms in one request (see above), whether you get the regional breakdown, and how they price a request that returns nothing.

Three different questions, three different shapes

Worth separating, because they are not the same job:

• Interest over time for many terms. A long list, each on its own scale, usually to spot a rise or a collapse. Billed per search term delivered, $3.00 per 1,000.
• A comparison between terms. Up to five terms on one shared scale, which is the only way to answer "which is bigger". Billed per comparison, $5.00 per 1,000.
• Where the interest is. The regional breakdown for a term, which is what you want before deciding where to sell or advertise. Billed per keyword mapped, $2.00 per 1,000.

What to watch for in your own numbers

• Low-volume terms return noisy or empty series. Treat a flat zero as "not enough data", not as "no interest".
• The time window changes the shape. A five-year view and a ninety-day view of the same term tell different stories, and both are correct.
• Category and region filters change the scaling too. Keep them fixed across a project or your series stop being comparable to each other.

The datasets behind this