How to Do Keyword Research for a New Blog
A new blog should target specific searches that few sites cover well, not the big head terms everyone fights over. Start with 30 narrow questions your reader types into Google, pick the ones where the current results are weak, and write those first.
That is the whole idea. The rest of this guide shows how to build the list, how to judge each candidate, and how to avoid the most common mistakes.
Start with the reader, not the tool
Open a blank document and write down who you are writing for and what they are trying to get done. “People who want to start a vegetable garden on a balcony” is a usable reader. “People interested in gardening” is not.
Then list the problems that reader has. Each problem is a potential search. A balcony gardener might ask which pots drain well, how much sun tomatoes need, or why basil leaves turn yellow. Aim for 50 raw ideas without filtering. You will cut most of them later.
If you are short on ideas, look at the questions in forums, comment sections, and product reviews in your topic. Real wording matters because search engines match on how people actually phrase things.
Expand the list with free sources
Type each idea into Google and watch the autocomplete suggestions. Scroll to the “People also ask” box and the related searches at the bottom of the results page. Copy the phrases that look like something a real person would type.
Free keyword tools help here too. Most of them take a seed phrase and return a list of related queries with rough monthly search volumes. Treat the volumes as ballpark figures. They differ between tools and are estimates, not counts.
Once you have a blog with some impressions, Google Search Console shows the exact queries you already appear for. For a brand-new site it is empty, so lean on autocomplete and related searches for now.
Judge each keyword in three checks
You do not need a formula to decide which keywords to keep. Run three checks on every candidate.
First, look at intent. Search the phrase and see what kind of pages rank. If the results are all product pages, a how-to post will struggle. If they are all guides, you are in the right place.
Second, look at the competition. Are the top results from large publishers with years of backlinks, or from small blogs and forum threads? A page one filled with thin forum answers or outdated posts is a good sign for a newer site.
Third, look at your fit. Can you write a better answer than what is there? If you can add a clearer explanation, a better example, or fresher information, the keyword is worth a post. If you have nothing to add, skip it.
Prefer specific over popular
New sites have little authority, so they rarely beat established sites on broad terms. Specific phrases usually have lower search volume but also lower competition, and the people searching them know what they want.
Compare “gardening tips” with “how to keep basil from wilting on a sunny balcony.” The second gets far fewer searches, but a small blog can plausibly rank for it, and the reader who finds the post has a real problem it solves.
Many specific posts also add up. Ten posts that each bring a modest number of visitors can outperform one post that never leaves page three.
Organize the list so you can act on it
Put your shortlist into a spreadsheet with a few columns: the keyword, the intent, a note on the current top results, an estimated difficulty (low, medium, or high is enough), and a status.
Sort by the ones where intent matches your content and the competition looks beatable. Group similar phrases together, because one post can often cover several close variations. Do not write separate posts for “best pots for tomatoes” and “best tomato pots,” since both are answered by one page.
Add a column for the working title. Writing the title early forces you to confirm the post has one clear job.
Common mistakes to avoid
Chasing volume is the usual one. A keyword with thousands of monthly searches looks attractive, but if the first page is full of well-established sites, you may wait a long time for little return.
Another mistake is writing for the keyword and ignoring the person. Stuffing a phrase into every paragraph reads badly and does not help rankings. Use the phrase naturally in the title, the first paragraph, and a heading or two, then write for the reader.
A third is skipping the results page. Tool scores can be misleading, and only the live search shows you what Google currently thinks a good answer looks like.
A worked example of the three checks
Suppose your list includes “how to repot a snake plant.” You search it and find that page one has a few large home and garden sites, plus two small blogs and a video. The intent is clearly informational, so a how-to post fits.
The large sites cover the basics but skip details such as what to do when the roots are tightly bound, or how to handle a cracked pot. That gap is your angle. You can write a post that answers the basic question quickly and then spends real space on the cases other pages ignore.
Now compare it with “indoor plants.” The results are dominated by big publishers and retailers, the intent is mixed, and you have no obvious gap. You would park that keyword for later and move on. Doing this for each candidate takes a few minutes and saves weeks of writing the wrong post.
How many keywords you need before you start
You do not need hundreds of keywords before publishing. A list of 30 with 5 ready to write is plenty. Research that never ends is a common way to avoid publishing.
Plan to revisit the list after your first ten posts. By then you will have a feel for how long posts take, which topics come easily, and which readers respond. Your second round of research will be better than the first because it will rest on real data from Search Console.
What to do this week
Pick one reader and write down 50 problems they have. Run each through autocomplete and the People also ask box, and keep the phrases that sound natural. Check the top results for your best 15, and cut any where the intent does not match or the page one is dominated by large sites you cannot yet compete with.
From what remains, choose the first five and write their working titles. Start drafting the one where you know the most. When you have a few published posts and some impressions in Search Console, come back and add the real queries to your list.