Search Fuel Space / Blog / AI keyword research
ArticleAI keyword research for Google Ads: why the list comes back as "low search volume"
Published
Ask an AI chat for Google Ads keywords and you get a clean, confident list in seconds. Upload it, and most of the keywords come back with the same status: low search volume. The list looked right. It was just never checked against what people actually search.
Written by Ericka Karen, who has planned Google Ads search campaigns since 2019 and wrote the rules behind Search Fuel Space.
- An AI chat has no access to Google search data, so the keywords it suggests often go live as "Low search volume".
- Inactive keywords cost nothing, but a structure and budget split built on them rest on searches that do not happen.
- Use AI for wording and Google Ads keyword data for demand: check every keyword before it enters the plan.
Why AI keyword research fails in Google Ads
A chat model writes the most likely words for a request. It does not look anything up. It has no access to Google's search data, so it cannot know that "affordable commercial solar installation for warehouses" is searched by almost nobody, while "commercial solar" is searched every day.
That is why the suggestions sound so good. They are specific, well written and on topic, the way a person would describe the service in a brochure. People searching Google write differently: shorter, plainer, often with a typo.
- It suggests phrases, not searches. Long, precise combinations that read well and that nobody types.
- It cannot see the country. The same term can have demand in the UK and none in Ireland. Without the data, it does not know.
- If you ask for volume, it guesses. A number from a chat is invented, however confident it sounds.
What "low search volume" means, and why it matters
In Google Ads, "Low search volume" is a keyword status. Google gives it to keywords with very little search history, and those keywords stay inactive: they do not trigger ads until searches pick up.
One inactive keyword is harmless. A plan where most keywords are inactive is not. The campaign structure was built around them, the ad groups were named after them, and the budget was split as if those searches existed. The account goes live and spends on the few keywords that do have volume, often the broadest and most expensive ones.
This is the story behind Search Fuel Space. Planning by hand used to be the slowest part of my work. AI chats made it fast, and then I watched plan after plan that looked complete fall apart the moment it reached the account, because the demand underneath was never real.
How to use AI for keyword research without the guesswork
- Let the AI brainstorm, not decide. Use it to list how customers might describe the service, the problems they have and the words on the website. That is a list of candidates, nothing more.
- Check every candidate against Google data. Run the list through Google Ads keyword data for the exact country and language of the account. Every keyword gets a monthly number, or it gets nothing.
- Keep the words people actually type. When a long, specific phrase has no volume, look for the shorter version people use. The demand is usually there, just in plainer words.
- Sort what is left by intent. Buying, comparing, brand, or just learning. Only the first groups deserve budget on a new account.
- Then bring the AI back. Grouping, ad copy, negative keyword ideas and the explanation for the client are all good jobs for it, once the keywords underneath are real.
Step two is the one that gets skipped, because Keyword Planner shows ranges instead of numbers on accounts that spend little, and checking a long list term by term takes hours. That is exactly the situation of a new client.
What AI is good at in Google Ads
None of this means keeping AI out of the account. It is useful for plenty of the work: reading a website and summarizing what the business can prove, grouping keywords by theme, writing and varying ad copy, suggesting negative keywords to check, and turning the plan into plain language for the client. Every one of those jobs works better when the keyword list it starts from has been checked.
The rule is simple: AI for wording, Google data for demand.
Where Search Fuel Space fits
Search Fuel Space is AI with access to the data an AI chat does not have. You send the website, the budget and the location. Every keyword in the plan comes with its monthly search volume from Google Ads keyword data, for that country and language, and what has no demand stays out, with the reason written down. You get the campaigns, the ad groups, the budget split, negative keywords and three responsive search ads per ad group, in a document with your brand. It plans, you execute, and it never logs into your account.
If you want to see what goes into a full plan, the Google Ads plan guide walks through it section by section.
Questions
Can I use ChatGPT for Google Ads keyword research?
For ideas, yes. It is quick at listing how people might describe a service. It cannot see how many people actually search each term, so every keyword it suggests still has to be checked against Google Ads keyword data before it goes into a campaign.
What does "Low search volume" mean in Google Ads?
It is a keyword status. Google marks a keyword this way when it has very little search history on Google Search. The keyword stays inactive and does not trigger ads until searches for it pick up.
Can an AI tell me the search volume of a keyword?
Not on its own. A chat model has no access to Google search data, so any number it gives is a guess. The volume has to come from Google Ads keyword data, through Keyword Planner or a tool that pulls that data for the country and language of the account.
Should I delete keywords marked as low search volume?
They do not cost anything while inactive, so there is no rush. The problem is counting them as demand: if half of a plan is low search volume keywords, the budget split and the structure were built on searches that do not happen.
Your next new account does not have to start on a blank page.
Join the waitlist. When the platform opens, you hear first and buy first, at the founding price, and you keep that price on future plans.
Get the founding price