How to Find Content Ideas People Are Actually Searching For

Most content ideas fail before you make anything. Not because the video, post or product is bad, but because nobody was looking for it in the first place.

Idea lists are easy to find. Ask any AI for "50 YouTube ideas about fitness" and you'll get 50 titles in a few seconds. They sound good. The problem is that nothing checked whether anyone is searching for them, who is already doing them, or what could go wrong.

This post covers how to check an idea properly, first by hand and then with a free tool I built to do it for you.

What a good idea actually needs

Before you spend a weekend on something, it should pass five checks:

1. Proof of demand. People are searching for it, asking about it or discussing it right now.
2. A gap. The results that exist leave something out: a question nobody answers, an audience nobody speaks to, or a format nobody uses.
3. A clear format. "Meal prep" is a topic. "What a night-shift nurse eats on 4 shifts in a row" is something you can actually make.
4. A first step you can take today. If the first step takes a month, you won't start.
5. An honest risk. Every idea has a weak spot. Knowing it up front is better than finding it after three months.

Most idea lists only give you the format. The other four are where the work is.

How to check an idea by hand

You don't need a tool for this. It just takes time.

Search it the way your audience would. Type the topic plus your platform ("meal prep for nurses YouTube") into a search engine. Look at what comes up. Are these big brands with polished pages, or small creators and forum threads? Small creators and forums usually mean there's room for you.

Look for the questions. Search "[topic] questions" and read the "People also ask" box. Every question that the top results don't answer well is a possible idea.

Check the communities. Search Reddit or Hacker News for the topic and sort by top posts. Points and comments are a rough measure of how much people care. A thread with hundreds of comments asking "is there a tool for this?" is close to proof of demand.

Name the competition. Write down 3 or 4 people already doing something close. If you can't find anyone, either you found a gap or there's no demand. Check which.

Write down the risk. One sentence: why might this not work? Too crowded, needs an audience first, hard to charge for, slow to build.

That's maybe an hour per idea if you're careful. Most people skip it, which is why most ideas don't go anywhere.

How IdeaSignal does it in about 10 seconds

I built IdeaSignal because I kept doing the steps above by hand. It's the first tool from GetRanked, and it's free.

You pick a niche (or type your own topic), a platform (YouTube, blog, X, TikTok, newsletter or SaaS) and a goal (grow an audience, make money, find an underserved market or get creative ideas). Then:

1. It searches what's showing up right now. Three web searches (the third one depends on your goal), plus popular Hacker News discussions from the last two years with their points and comments.
2. AI drafts ideas from those results only. Each idea has to cite at least two of the search results it came from. Vague ideas, repeats and ideas the results don't support are thrown out. Getting fewer ideas is fine.
3. A second model scores each one. Jev, a model from TypeSafe, answers the same five questions about every idea's sources: is there demand, is there a gap, is it dominated by big brands, are people paying for something related, and does it fit the platform. Those answers become the score.
4. You get the top 3, plus 2 that "have potential". Every idea comes with the sources it's based on, the competitors those sources name, its biggest risk and a step-by-step plan.

The AI writes the ideas but never scores them. The score comes from the same fixed questions and weights every time, not from the model that wrote the idea.

A real example

Here's one real run from this week. Custom topic: meal prep for nurses on night shift. Platform: YouTube. Goal: grow an audience. It found 17 sources.

The #1 idea, scored 77.8 out of 100:

What a Night-Shift Nurse Eats on 4 Consecutive Shifts

A YouTube series for night-shift nurses that follows one full stretch of consecutive shifts and shows exactly what gets eaten before, during, and after work. Each episode can focus on a different constraint, like no microwave, low appetite, or avoiding the post-shift crash.

Biggest risk: "This depends on viewers identifying with the same shift pattern, which narrows the audience compared with broader healthy-eating videos."

Already doing something close: two "what I eat as a night shift nurse" videos and two meal-plan articles, all named from the cited sources.

How to stand out: build the channel around repeatable "shift stretch" episodes instead of one-off recipes, so viewers can binge the format.

The plan:

1. Ask 10 night-shift nurses which part of consecutive shifts is hardest and use the most common answer for the first episode.
2. Record one full week of food prep and pack 4 shift days into labeled containers and snack bags.
3. Cut the footage into a structured episode: pre-shift meal, mid-shift food, caffeine timing, and post-shift recovery meal.
4. Publish the episode and a short teaser clip showing the full lineup of meals in one frame.
5. Measure comments, saves, and whether viewers request the next episode for another shift pattern or constraint.

Every idea in the run came out aimed at nurses, because the topic said so. That's the point: if your topic names who it's for, every idea has to fit them.

What it can't do

It's worth being clear about the limits:

• Scores are estimates, not guarantees. They compare ideas against each other from public search results.
• Search results aren't search volume. IdeaSignal doesn't have traffic or sales data, and it doesn't pretend to.
• The ideas are AI-written from search snippets. Read the linked sources before you commit to anything.
• It's free with limits: 3 runs a day per connection, 2 minutes apart. Each run costs me about 3 cents, and the site runs on donations.

Try it

Pick something you've been thinking about making and run it: signal.getseoranked.com. No account needed.

If something's missing or wrong, there's a feedback link at the bottom of the page. Feedback decides what I build next.

IdeaSignal is the first of a set of small tools I'm building under GetRanked. Each one does one job, works from real data and tells you what it doesn't know.