I stopped trusting keyword research tools years ago because their numbers are fabricated estimates, not real demand. Here's what I do instead, and why it lifted my traffic and sales by over 200%.
For years I did what every free keyword research tools guide tells you to do. I typed a phrase into a tool, looked at a volume number, and wrote the article. The traffic never showed up the way the tool promised. After enough of those misses, I dug into how these tools actually generate their numbers, and the answer is that most of them are guessing. They model search volume from sampled data, smooth it with an algorithm, and hand you a round figure that feels authoritative. It isn't. It's a statistical shrug dressed up as a fact.
So I quit relying on them as my first step. Now the first thing I do with any topic is search it myself, on Google, the way an actual reader would type it. Then I scroll to the bottom of the results page and read the related searches. That section isn't generated by a model predicting what people might want. It's built from what people actually typed, aggregated in real time. If Google is surfacing eight or ten related searches under a topic, real people are actively chasing that topic right now. If the related searches are thin or generic, there's no audience there worth my time, no matter what a volume estimate claims.
Only after I've confirmed that signal do I bring in tools at all. The two I use are Headline Pig and Content Cube, and I use them for a specific reason: they don't estimate demand, they pull it. Both surface actual search behavior sourced from what people typed into Google, Amazon, or YouTube, and both flag it with a green icon when the demand is verified against real query data rather than modeled. That distinction sounds small. It isn't. It's the difference between building on a guess and building on a fact.
The first thing I do inside these tools is run a head-to-head. Take two products, services, or ideas that sit close enough together that a buyer would naturally weigh one against the other. If people are actually searching both names side by side, comparing prices, specs, or outcomes, that's verified search demand, not a hunch. I ran this exact check last year on a pair of budget standing desks that were close competitors in price and features. Headline Pig showed the comparison query lit up green, meaning real people were typing both brand names into the same search. That single signal told me more than any volume estimate ever had, because it wasn't a projection. It was a record of what buyers were already doing.
Checking who else covers that exact comparison
Confirmed demand only matters if I can actually compete for it. So the second check is coverage. I search the exact comparison and count how many sites are already writing about it in that specific framing. Broad topics are almost always saturated. Specific comparisons often aren't. On that standing desk pair, I found two thin blog posts and nothing else treating the comparison as its own topic. Almost nobody was touching it directly, even though the demand was verified. That gap is where a smaller site actually has a shot. Skip this step and you're publishing into a category where sites ten times your size have already locked up the first page, and no amount of good writing fixes that math.
Checking buyer intent behind the keyword
The third check is the one most people skip entirely, and it's the one that decides whether traffic turns into sales. I look at who is actually running the search. Someone typing a brand name plus a vague word like "review" might just be browsing. Someone comparing two specific models, asking about warranty terms, or searching a product name with "vs" or "worth it" is standing at the checkout with a card in hand. That's buyer intent, and it's visible in the shape of the query itself. On the standing desk comparison, the related queries included price checks and shipping questions. Those aren't researchers. Those are buyers. Content built for that search doesn't just attract readers, it attracts people who were already going to spend money with somebody. The only question is whether it's with you.
What the fabricated numbers actually cost you
Here's the part that took me too long to accept. Every article written off a fabricated volume number carries a real cost: the hours of writing, the weeks of waiting for rankings, and the slow erosion of belief that content works at all. I've watched people publish thirty pieces into an empty room because a tool told them 2,400 people a month were searching for something nobody was searching for. The tool wasn't lying maliciously. It was estimating, and the estimate was wrong, and the writer paid for it.
So now nothing gets written until all three boxes are ticked: verified demand, manageable competition, and buyer intent. Hit all three and the headline has already won before I've written a single sentence of content. That's where the 200% lifts came from, and in a few cases the 10X and even 100X outliers. Miss one and you're publishing into an empty room, competing against sites ten times your size, or attracting readers who were never going to buy.
Most people skip this process entirely and then wonder why their content never takes off. It takes twenty minutes per topic. It's free. And it's the difference between writing for an audience that exists and writing for a number somebody's algorithm made up.
Also read: What Is an AI Browser and How Do Agentic Tools Like Comet Work • Best AI Video Generator for Marketing: A Founder's Hands-On Ranking • These Are the Best AI Coding Tools for Non-Technical Founders Right Now
