How to Do Keyword Research for a Brand New Website
A new site has no search data of its own and no standing to compete with. Two disadvantages at once. The usual method assumes neither of them. This is a different starting position rather than the same exercise done earlier.
Two Disadvantages At Once
An established site doing keyword research can look at what it already appears for and build outward from evidence. A new site has neither of those. No reporting to read and nothing behind it to make search engines take a chance on it. Both problems have to be worked around rather than solved.
The first disadvantage. No data.
Every other guide in this series eventually says to check your own search reporting. A new site has an empty account, so external sources and your own knowledge carry the whole exercise.
The second disadvantage. No standing.
Nothing yet demonstrates the site is worth showing. That is not a permanent condition and it decides where you can realistically start.
What people do wrong. Aim at the main terms.
A new business writes its home page around the biggest search in its trade and waits. Nothing happens. The conclusion drawn is that search does not work.
What actually works. Start elsewhere.
Deliberately choosing searches nobody has covered properly, which is block three and is the whole strategy for a site with nothing behind it.
What a new site does have. Freedom.
No existing pages to work around, no accumulated mistakes and no competing pages of its own. Building correctly from the start is considerably easier than repairing later.
Forget The Terms You Actually Want
The searches a new business most wants to appear for are the ones it has no chance of winning. That is uncomfortable and it is the single most useful thing on this page, because accepting it early redirects a year of effort towards something that can actually work.
Why those terms are held. Years of accumulation.
The sites appearing for broad commercial searches have been there a long time, with everything that follows from that. A new site cannot shortcut it.
Why aiming there costs so much. No partial credit.
You either appear well for a search or you do not appear at all. Effort spent on an unwinnable term produces nothing, which is unlike most business investment.
What that effort looks like wasted. A year gone.
Twelve months of writing and rewriting a page aimed too high, with nothing to show and no way to tell whether anything was working.
What to do with those terms. A second list.
Keep them written down and dated. They become realistic later. Deleting them loses the direction the business is heading in.
How to explain it to an owner. Order, not ambition.
Nobody is saying the business will never own its main term. The argument is about which order to approach things in. That framing is usually accepted immediately.
What replaces them. Specific ground.
Searches with fewer serious contenders, which our long tail guide covers as the route rather than the limit.
Start Where Nobody Is Competing
Look for searches nobody has written a proper page about. They exist in every trade and they are usually specific and unglamorous. They are the only ground a site with no history can take.
How to recognise one. Weak results.
Search the phrase and look at what appears. If the answers are forum threads, pages that mention the subject in passing or listings with nothing useful on them, nobody has covered it.
Why the gaps exist. Large sites write broadly.
Established sites serve everybody, so they write general pages. That leaves the particular situations uncovered. Particular situations are what your customers are in.
Where the gaps concentrate. Three places.
Awkward specific problems. Questions people are slightly embarrassed to ask. Anything that varies locally. All three get asked constantly and answered badly.
What to check before committing. Real demand.
Nobody competing can also mean nobody searching. Confirm that customers actually ask this before writing, since your own experience is the evidence here.
Why these convert well. Relief.
Somebody who has searched three times and found nothing useful, then lands on a page that answers them properly, is considerably more likely to get in touch.
What this builds towards. Standing.
Winning specific searches is how a site accumulates the evidence that eventually makes broader ones possible.
Your Own Knowledge Is The Only Asset
A new site has no data, no standing and no history. What it does have is somebody who knows the trade. On a new site that is worth more than any research tool, because it is the only thing competitors cannot copy from a database.
What that knowledge contains. Real questions.
What customers ask, what they get wrong, what they worry about and what they are surprised by. All of it is searched and most of it is answered badly.
Why it beats a tool here. Nothing to check against.
With no reporting of your own, a tool's suggestions cannot be validated. Your own experience of customers can be, because you were there.
The exercise to do first. Before anything else.
Write down every question you have answered in the past month. That list is the foundation and it takes an evening.
What makes it defensible. Specificity.
Detail from actual jobs, actual prices and actual outcomes. A larger competitor writing generally cannot match a page written by somebody who did the work.
The mistake to avoid. Writing what you know.
Not everything you know is something anybody searches for. Filter it against real questions rather than publishing your expertise unfiltered.
Where this fits the method. The same first step.
Our method guide opens on the same instruction. On a new site it carries more weight than usual.
Difficulty Scores Know Nothing About You
A difficulty score is calculated without any knowledge of the site asking. For an established business that makes it a rough guide. For a new site it is close to useless, because the whole question is whether a site with nothing behind it can compete. The number was never asking that.
What the score is measuring. The competition.
How strong the pages currently appearing are, expressed as a single figure. Reasonable information, though only half of what you need.
What it is not measuring. You.
The same search is achievable for one site and impossible for another. A number attached to the search alone cannot express that difference.
Why that hurts a new site most. The gap is widest.
For a site with no history, the difference between what the score implies and what is actually achievable is at its largest. Everything looks harder than the number says.
What to look at instead. The actual results.
Two minutes reading what appears tells you more than any estimate. Look for whether businesses like yours are there or whether it is all national brands.
The signal worth acting on. Weak answers.
Low quality pages ranking is a better opportunity signal than any score, because it means nobody has bothered rather than that everybody has tried.
What to do with the scores. Break ties.
Use them to choose between two otherwise equal options and nothing more, which is the same treatment we give volume figures throughout this series.
Depth In One Place Beats Breadth
A new site covering one subject completely is in a stronger position than one touching six subjects partially. That is true generally and it matters enormously here, because a new site has limited effort and no margin for spreading it thinly.
Why depth works. Something to demonstrate.
A site that answers everything about one subject has shown it knows that subject. A site with one page each about six subjects has shown nothing about any of them.
What the mechanism is. Our advanced material.
Why grouped, interlinked coverage outperforms scattered pages belongs to the advanced SEO material rather than being re-argued here.
What breadth feels like. Progress.
Publishing across several subjects feels like covering more ground. It produces a site that is thin everywhere and strong nowhere.
How to choose the first subject. Commercial and winnable.
The one that matters most to the business among those you can realistically take. Both conditions, which usually rules out the obvious answer.
When to start the second. When the first is finished.
Meaning every reasonable question about it has a home. Our content strategy guide covers that discipline.
What this does to the launch plan. Narrows it.
Fewer subjects, covered properly. That is a smaller and more achievable launch than most businesses plan for.
What Happens First
Everybody launching a site wants to know how long it takes. We will not tell you, because any figure would be invented and it would be quoted back at us when reality disagreed. What can be described accurately is the order things happen in.
First. Being found at all.
Pages appear for searches somewhere, usually for very specific phrasings nobody targeted. Not visible yet, though it is the first evidence anything is working.
Second. Unplanned queries accumulate.
Your reporting starts showing searches you never chose. That is the tail arriving and it is the most useful early signal available.
Third. The specific pages start working.
The narrow subjects you deliberately chose begin appearing properly. This is where the first enquiries usually come from.
Fourth. Nearby searches follow.
Having covered one subject well, related searches become achievable without new work, because the coverage already supports them.
Last, if at all. The terms you wanted.
Broad commercial searches come after everything else. For a good many small businesses they never come at all. That is not a failure if the specific ones produce work.
Why we describe it this way. It is checkable.
You can look and see which stage you are at. A date on a calendar tells you nothing about whether the approach is working.
When Your Own Data Arrives
At some point a new site stops being new. The moment that happens is not a date, it is when your own search reporting contains enough to correct the assumptions everything so far was built on. From then the method changes.
What changes. The best source.
External estimates stop being your primary evidence. What your site actually appears for takes over, because it describes you rather than any site.
How to know it has happened. Patterns, not entries.
One search appearing once is noise. The same searches appearing steadily across weeks is a picture. That is the threshold rather than any elapsed period.
The first thing to look for. Unplanned searches.
What you appear for that nobody chose. That tells you what the market thinks your pages are about, which is frequently not what you intended.
The second thing. Near misses.
Searches where you appear without being found. Those are the cheapest next wins and they are how a new site accelerates.
What to do with the original plan. Amend it.
Add what the data revealed, reorder what remains and drop anything the evidence says nobody wants. Our guide to reading that data covers the exercise.
What not to do. Start again.
The subjects rarely move. It is an amendment rather than fresh research. Everything sits on the keyword research guide.
The searches you
most want are
the ones you
cannot win yet.
Effort aimed at a term held by sites with years behind them produces no partial credit. You either appear well or you do not appear, so a year can disappear into a page aimed too high with nothing to show and no way to tell whether anything was working. Nobody is saying you will never own it. The argument is about order.
How we start a new site:
We will not tell you how long it takes. Anybody who gives you a figure is guessing. It will be quoted back at them when reality disagrees.
Every guide.
One subject.
What the words mean, what the numbers in a keyword tool actually are, how to find the searches worth targeting, how to turn a list into a publishing plan and what to do when a page you relied on slips.
