This short Make automation uses AI to select keywords from Google Trending Now via SerpAPI, then has an AI agent review the keywords to measure relevance to your brand. When it finds relevant keywords, it will send them to you via email. It can be set to any time interval, but we recommend 24 hours maximum to 4 hours minimum.
It’s a neat starter agent for people in SEO and digital marketing to implement, as it has most of the basics:
API call
Flow control
AI agent selection
Use a tool to send information
The JSON blueprint is here. To get started you will need:
Free make.com account.
Free SerpAPI account.
OpenAI API access with $10 credit.
Integration between make.com and Gmail (can be via personal Gmail, but easiest through Google Workspace).
Module 1 - Make a SerpAPI API Call
The first step is the trickiest, but given there’s a native SerpAPI module in make.com, it’s not so bad. Use the ‘Make an API Call’ module at the bottom when you click add module and search for SerpAPI. You must have an account with SerpAPI for it to work.
For the URL all you need to put is /search.json.
Then various key/value fields make up the query string. The full documentation for Google Trends Trending Now API is linked here, but as an example, this is what I filled out:
Item | Key | Value |
1 | engine | google_trends_trending_now |
2 | geo | GB |
3 | hours | 24 |
4 | category_id | 4 |
This tells the SerpAPI module to use the Google Trends Trending Now API, in GB/UK location, in the last 24 hours for the Entertainment category. A full list of category IDs is linked here.
Once this is set up, it’s possible to add in more categories into the Make scenario by cloning the Category, Iterator and Array Aggregator steps and putting them before the OpenAI Create a chat completion module, like in the below:
You also don’t have to set a category - this will just take from the top 50 trending keywords in the selected time period.
Modules 2 and 3 - Flow Control
The first module will pull a reasonably big package of data from bigger categories. These steps just sort that data into something manageable. For the Iterator, you just need to select 1. trending_searches []
For the array aggregator, you want the Iterator as your source module, then Aggregated fields as query, search_volume and increase_percentage. This will give you a good sense of what keywords are trending and their velocity.
Module 4 - AI Agent
We might get told off for calling this an AI agent, given it’s just a chat completion - but this AI step does make probabilistic decisions and filters the list, which is pretty much an agent. Those semantics aside, there are two prompts. Using GPT 4.1 is recommended because GPT 5 doesn’t work so well in make.com just yet.
First is the Developer / System prompt. Note anything in {brackets} is for you to review and change.
You are a Google Trends (Trending Now) analyst.
Instructions:
- Review each "query".
- If you think it matches the guidelines of {YOUR BRAND}, include it in the output.
- Each item must be in the format:
<li><em>query - search_volume (volume) - increase_percentage</em>: Short explanation of why this is interesting to {YOUR BRAND}</li>
- Wrap all items inside a <ul>…</ul> (not each list item).
- Use a thousand separator on search_volume
- Ignore all queries that are not related to {YOUR BRAND} as per the brand guidelines and keywords below.
Return HTML only — no markdown, no plain text, no further explanations.
## {YOUR BRAND} Guidelines ##
{PUT BRAND CONTEXT IN HERE - this can be as long as short as you like}
## Preselected keywords for {YOUR BRAND} ##
If any keyword is vaguely related to these it must be included. This is also not a definitive list.
{INSERT KEYWORD LIST - recommend c.50 high interested keyword / topics, but can have fewer. Use entities rather than generic keywords, as this agent works on trends}
IMPORTANT: If uncertain, always include. It is more useful to have 20 listed than 4, as we can discard what we don't want. We cannot discard what we don’t know. User prompt:
Here is the JSON array of trending searches:
Use 7. Array []
Module 5 - Send Email
Configure a Google Workspace connection and select email address and subject line. The content should just be the result from the AI agent. Alternative you could set this up in a messaging app like Slack.
Now you just need to set the schedule. You can do this on the bottom bar. Run once for testing, or set to a daily (or other time period) schedule.
