How to find keywords you can actually win

Keyword research goes wrong in the same way almost every time. Somebody opens a tool, sorts by search volume, picks the biggest numbers they can find, and spends a year writing pages that will never reach the first page for any of them.

Volume is not value. A keyword is worth what the visitor is worth to you, multiplied by how likely that visitor is to act, multiplied by your realistic chance of ranking. Two of those three have nothing to do with how many people search it.

Three questions, in this order

1. 01Does this describe a problem you actually solve? If somebody searching it would not want what you sell even after reading a perfect page, the traffic is decoration.
2. 02Can you plausibly rank for it? Look at who is on the first page now. This is the question people skip and it decides everything.
3. 03Is there enough of it to matter? Volume comes last, and the threshold is lower than you think.

How to judge difficulty without a tool

Search the phrase and look at the first page honestly. Difficulty tools give you a number out of a hundred, which is useful, but the results themselves tell you more.

• If the first page is national brands, government sites and Wikipedia, you are not getting in. Pick something else.
• If it is a mix of known names and sites you have never heard of, there is room.
• If it is forum threads, a video and two thin pages that half answer the question, you have found something worth writing.
• If the results do not really answer the question at all, that is the best signal available. Somebody searched it, nobody served it.

A first page, read properly

An example of the judgement in practice. Say you set up stock ordering for independent restaurants and you are weighing two phrases.

For the broad one, restaurant inventory software, the first page is four software companies with pricing pages, two review-site round-ups and a trade publication. Every result is a business with a marketing budget and years of accumulated links. You are not getting in, and the people clicking mostly want to compare software rather than hire somebody.

For how to count stock in a small restaurant, the first page is two blog posts that stop at generalities, a forum thread where somebody describes their actual routine, and a video. Nothing there is a complete answer from somebody who has done it. That is a page you can write and win, and the person reading it has exactly the problem you fix.

The second phrase has a fraction of the volume. It is worth more, and it is worth more in both directions: you can reach the first page, and the visitor is closer to wanting you.

Where to find them

The best sources are the ones that do not come from a tool, because everyone has the tools.

• Your own customer conversations. The words people used when they described the problem to you are the words they type.
• The suggestions that appear as you type a search, and the related searches at the bottom of the results page.
• Questions in communities where your customers complain. The phrasing is unpolished and specific, which is exactly what you want.
• The pages your competitors rank for that they clearly did not plan. Those are often accidental and easy to take.
• Your own search data, once you have any. Search Console shows searches you nearly ranked for, which is the shortest path to a win.

Mining your own near misses

That last source is the most underused thing in small-site SEO, and it costs nothing. Once a site has a few months of data, Search Console is already telling you where you are close.

1. 01Open Performance, set the range to the last three months, and switch on both Average position and Impressions.
2. 02Sort by impressions and look for queries sitting between roughly position eight and twenty. Those are phrases where you appear, nobody clicks, and a small improvement moves you onto the visible part of the first page.
3. 03For each one, check which of your pages is ranking for it. Often it is the wrong page, and the fix is to give the phrase a page of its own rather than to improve the accidental one.
4. 04Look for queries with impressions and almost no clicks even at a decent position. That is usually a title problem rather than a ranking problem.

Improving a page that already ranks twelfth is far cheaper than writing a new page that starts from nothing, and the result arrives in weeks rather than months. Most sites have a dozen of these and have never looked.

Group them rather than listing them

Twenty phrases is not twenty pages. Many of them are the same question asked differently, and splitting them produces several thin pages that compete with each other instead of one good page that wins.

Sort your list by what the searcher actually wants. If two phrases would be satisfied by the same page, they belong together and the more natural one becomes the page's main phrase. If they would need different pages to be satisfied properly, they are different pages, whatever the wording suggests.

Specific beats broad, at the start

A phrase with four words and a clear situation attached is worth more to a new site than the two word version with fifty times the volume. It is easier to win, the person searching is further along, and the page almost writes itself because the question is specific.

Broad terms are not off limits forever. They are what you earn after a cluster of specific pages has given the site a reason to be taken seriously on the subject.