How to Do Keyword Research for a Brand New Website: Start Narrow

Only 1.74% of newly published pages reach Google's top 10 within a year for at least one keyword, according to Ahrefs' latest ranking-age data. That figure should change how a new site does keyword research.

An established site can research widely and publish across many topics. A brand new one has no authority, no history and no data, so it cannot afford shallow coverage of everything. It needs a narrow starting point, then room to widen.

This guide covers the three decisions that matter first: the territory you will own, how to check demand when you have no numbers, and which terms to go after. It ends with the order to publish your first 20 pages and what a realistic first year looks like.

What the First Year Looks Like for a Site With No History

Set expectations before you pick a single keyword. In a Google video on hiring an SEO, Maile Ohye said that in most cases SEOs need four months to a year to help a business implement improvements and see benefit. That was said years ago, but it still matches the data.

Ahrefs' statistics roundup reports that 72.9% of pages ranking in Google's top 10 are more than three years old. Time is part of the ranking picture, and a new site starts with none.

An earlier Ahrefs ranking-age study adds a useful nuance. Pages could reach the top 10 for low-volume keywords quickly, while high-volume keywords took close to a year, and only 0.3% of pages ranked top 10 for a high-volume keyword in under a year. Where you aim decides how long you wait.

So judge months one to six by different signals than traffic. Look for pages indexed, impressions rising in Search Console, and a growing list of queries that show your pages at all. Treat visitors and leads as a later result, not a month-two target.

This is also why keyword research for a new site is mostly a sequencing problem. The goal is not the biggest keyword list. It is a short list you can win first, so you collect evidence that your site works.

Also Read: SEO Fundamentals: Complete Guide to Ranking Higher on Google

Decision One: Pick a Territory Small Enough to Win

A territory is the slice of a topic where you will publish everything, at depth, before widening. It is defined by an audience, a problem and your edge. "Woodworking" is a topic. "Quiet woodworking for apartment dwellers" is a territory.

Here is a test I use as a rule of thumb, not an industry standard. Can you list 30 distinct, useful pages in the territory without repeating yourself? Below 30, it is too narrow. Past 300 without effort, it is too broad for a new site.

A hypothetical check for the apartment woodworking territory shows the idea. Sample pages might be workbench designs for small rooms, quiet hand tools, dust control without a garage, storing lumber in a flat, finishing indoors, and noise etiquette with neighbors.

Six topics came easily, and they clearly branch into dozens more. Your edge decides where to start inside a territory.

It can be expertise, access to people or data, a clear point of view, or just a willingness to document something others skip. Write it in one sentence. If you cannot, the territory is probably not yours yet.

Check three things about the territory. Is there a real audience asking questions, which forums, comment sections and Google's own suggestions will show? Do you have an angle, such as expertise, data or experience? And can a small site appear on page one, or do only huge brands appear?

Use Google Trends as a sanity check. It shows relative interest over time and lets you compare terms. A territory with steady or rising interest is safer than one that peaks once a year, unless seasonal demand suits your business.

Be willing to start smaller than feels comfortable. A narrow territory is easier to cover thoroughly, easier to link internally and easier for readers to remember. You can always widen once pages start ranking.

Decision Two: Check Demand When You Have No Data

You have no Search Console data yet, so you need to borrow signals. Start with Google's autocomplete suggestions and People Also Ask boxes. They show that a question exists, though not how many people ask it.

Next, read the results page. If small, specialized sites rank for the phrase, demand is real and the topic is reachable. If every result is a household brand or a marketplace, the phrase may be real but out of range for now.

Use keyword tools for ranges, not exact numbers. Ahrefs' own study found that Google Keyword Planner drastically overestimated volumes 54% of the time, and that Ahrefs' estimates were roughly accurate for 60% of studied keywords against Search Console impressions. Every tool estimates.

A practical habit follows from that. Check the same phrase in two sources and sort it into a bucket instead of trusting one number. Here are buckets I use, as my own convention for planning.

Tiny, under about 50 searches a month: only worth a page if it is close to what you sell or teach.

Small, about 50 to 500: the sweet spot for a first page, because competition is usually lighter.

Medium, about 500 to 5,000: a stretch goal once you have several pages in the same cluster.

Large, above about 5,000: a long-term target. Do not build a new site's plan around it.

Take a hypothetical example. One tool shows 300 searches for a phrase and another shows 90. Do not average them.

Place the phrase in the small bucket, because both readings land near it, and let real impressions settle the question later.

Do you need a paid tool for this? Not on day one. Autocomplete, People Also Ask, Google Trends, forums and a careful read of the results page cover the first pass. A paid tool earns its place when you are comparing hundreds of phrases and want to save hours.

Also Read: The Best Keyword Research Tools for Low-Competition Niches

Decision Three: Choose Terms Your New Site Can Realistically Win

Now match demand to ability. A phrase is worth targeting when you can answer it better than the pages already ranking, using low-competition keyword research to identify realistic opportunities, and when the pages that rank do not rely on authority you lack.

Read the top five results for each candidate and note what they are: forum threads, old posts, thin pages and small blogs are good signs.

Large publishers, tool vendors and official sites are not. Pay attention to freshness and depth as well as the site name.

Then build a launch portfolio instead of chasing one perfect keyword. As a guideline, not a rule, take your first 20 pages and aim for about 14 winnable terms, 4 stretch terms and 2 long-term pillars. The winnable pages give you feedback within months.

The stretch pages test your ceiling. The pillars are the big guides you will keep improving for a year or more. Having all three keeps you from either playing too safe or burning a year on a goal you cannot yet reach.

Sometimes the research says everything is out of reach. That usually means the territory is too broad or too crowded, not that SEO cannot work. Narrow it by audience, by location or by use case, until page one includes sites that look like yours.

Seasonality deserves a note too. Pages need months to rank, so publish seasonal topics well ahead of the peak that Google Trends shows. A page published in the peak month usually arrives too late for that year's demand.

Here is a hypothetical walk-through for a new site on quiet woodworking for apartment dwellers. All volumes are made up for illustration. The point is the reasoning, not the numbers.

"apartment workbench ideas": small bucket, about 300 searches. Small blogs and forum threads on page one. Winnable, so it goes in the first cluster.

"quiet woodworking tools for apartments": small bucket, about 90. Two thin pages on page one. Winnable and close to the territory.

"how to cut plywood without a table saw": medium bucket, about 1,200. A mix of videos and blogs. A stretch page once the cluster has support.

"best beginner woodworking projects": large bucket, about 40,000. Big publishers dominate. A long-term pillar, built last.

"woodworking": a head term. Skip it entirely for now.

Notice what the example does. Two phrases are winnable now, one is a stretch, one becomes the pillar and one is ignored. That is a plan a new site can follow, instead of a list of keywords it cannot reach.

Avoid choosing by volume alone. Ahrefs notes that search volume does not always translate into traffic and that its Traffic Potential metric predicts organic traffic more accurately, which is a reminder that the top page's real traffic tells you more than a keyword's headline number.

Factor in AI summaries too. Ahrefs found that AI Overviews were associated with a 58% lower average click-through rate for the top organic result. For a simple informational phrase, an answer box may satisfy readers before they click.

Check each candidate's results page for a summary at the top. Favor phrases where your page can offer more than a short answer, such as a template, a comparison, a worked example or local detail. Those pages give readers a reason to click through.

Sequence the First 20 Pages and Match Their Intent

Order matters as much as selection. Publish your pages in clusters rather than scattering one post per topic. Five connected pages on one sub-topic give readers and Google a clear picture of what your site covers.

A sensible order starts with the winnable pages in your first cluster, then adds a stretch page that links to them, then moves to a second cluster.

Each new page should link to related ones, using descriptive anchor text and a clear internal linking strategy, from the day it goes live.

For the hypothetical woodworking site, the first cluster might run in this order: apartment workbench ideas, quiet woodworking tools, dust control without a garage, storing lumber in a flat and finishing indoors. Then comes the stretch page on cutting plywood without a table saw, linking back to all five.

Use the pillar as the hub. Even a rough pillar draft can wait until the supporting pages exist, because it can then link to a finished set. That is usually easier than writing the pillar first and forcing the others to fit.

Match each page to the intent behind its phrase. Informational searches want an explanation or a how-to. Commercial searches compare options.

Transactional searches want to buy or sign up. A page that ignores intent will struggle even when the keyword is perfect.

For a new site, start with informational pages in your first cluster, because they are easiest to make clearly useful. Add one commercial or transactional page per cluster once the supporting pages exist. Comparison pages in particular need credibility, so earn it first.

Our piece on topical authority explains why this depth matters for how search engines judge a site. For a new site, the practical takeaway is simple: finish one cluster before opening another.

Also Read: E-E-A-T Explained: How to Build Topical Authority

If Your New Site Sells Something: Where the Research Changes

Everything above assumes you are building an audience first. If you sell a service or a product, the research starts closer to the sale, because a small number of the right visitors can matter more than a large number of curious ones.

A new local service business should pair what it does with where it does it. Phrases that combine a service with a city or area reflect clear intent. Check who appears for them, and keep your business profile details consistent with your site.

A new online store should research category-level phrases before individual products. Category pages can hold more related phrases and give you something to link into.

Be wary of terms that large marketplaces own, since a new store rarely beats them head on. A new software or product site should start with the problem, not the product name.

People who have never heard of you search for what hurts. Comparison and alternative phrases come later, once you have the credibility to write them fairly.

Use Search Console to Correct Your List

Verify your site in Search Console on day one, so data starts accumulating. Give it a few weeks, then open the Performance report and look at the queries and pages that appear, even with tiny numbers.

Two views help most. First, queries where your page appears at positions 8 to 30. Those are phrases Google already associates with you, so a stronger section or a clearer heading may lift them. Second, queries you never planned for, which often reveal wording your readers use.

Seeing nothing after a few weeks? Check the basics before you worry. Confirm that the pages are indexed with the URL Inspection tool and that your sitemap is submitted. New sites are slow to appear, and an empty report is often just a young site.

Treat the first month as exploratory. A handful of impressions is a signal, not a trend. Wait for patterns across several weeks before you change course, and add the new phrases to your list.

Keep a simple research log with one row per phrase: the phrase, its bucket, your verdict on the results page, the page you assigned it to and the date. Within a few months it becomes a record of what you tested and what the data showed.

Then re-rank your list every quarter. Move proven phrases up, drop ones that attract nothing and promote a stretch term when a cluster has built enough support. Keyword research for a new site is a loop, not a one-time task.

Narrow First, Then Widen

A new site wins by being specific, patient and organized. Pick a territory you can cover in depth, check demand with several imperfect signals, choose terms you can win, and publish in clusters.

Here is a four-week plan before you publish anything, using my own time estimates. Week one, define the territory and run the 30-page test, about three hours. Week two, check demand for your candidate phrases, about four hours.

Week three, read the results pages and build your launch portfolio, about five hours. Week four, outline the first cluster, about five hours. That is roughly 17 hours of research before the first page, and it saves you from writing pages nobody will find.

Each ranking page makes the next one easier. Start small enough to finish, and widen only when the data tells you the first area is working.

Also Read: The Marketing Strategy for Global Organic Traffic Dominance