How to Find the Best Keywords for SEO in 6 Steps

Last updated: November 2026.

Finding keywords for SEO comes down to six steps: list seed topics from your services, expand them with free tools, check search intent, read volume against difficulty in an Australian location, compare what competitors rank for, then map one keyword to each page. DigiFix runs this method for Melbourne businesses every week.

What makes a keyword the best one to target?

A keyword is worth targeting when three things line up: people search it, they want what you sell, and you can realistically reach the first page. Plenty of keywords fail one of those tests. A high volume phrase that attracts researchers will not pay the bills, and a perfect intent match you cannot outrank is worse than a smaller term you can win.

So the goal is not the biggest keyword. It is the keyword with the best ratio of demand to competition that sits close to a buying decision. For a Melbourne accountant, “tax return Melbourne” carries buying intent, while “what is a tax return” carries research intent and belongs on a different page, if it belongs on the site at all.

Two practical filters save the most time. First, ask whether a page already answers the query, because two pages chasing one keyword compete with each other. Second, ask what a searcher would need to see to act, because that tells you whether the keyword wants a service page, a guide or a comparison. Our guide to search intent covers that second filter in detail.

How do you start a shortlist from your own services?

Start with what you already sell, not with a tool. A tool given no direction returns a list of generic terms your competitors also found, which is how sites end up with six pages that all say the same thing.

Write down each service you invoice for, then each problem a customer describes before they buy it. Those two lists are your seed topics. A plumber in Melbourne might write blocked drain, burst pipe, hot water system replacement, gas fitting, leaking tap and emergency plumbing. A migration lawyer might write partner visa, skilled visa, refused visa appeal and sponsorship obligations.

Then add the words customers actually use. Their vocabulary differs from yours. A customer searches “water coming up through the floor”, not “subfloor drainage remediation”. Search Console is the cheapest place to find that gap, because the Performance report lists the queries your pages already appeared for, including the long phrasing you would not have guessed.

Ten to fifteen seed topics is plenty. Fewer means you will run out of angles. More means you are still thinking like a business owner rather than a buyer.

Which free tools expand a keyword list fastest?

Four free tools cover the whole job, and each one answers a different question. Use them in this order rather than opening all four at once.

Free sources of keyword ideas and the setting that makes each one accurate for an Australian site.

Keyword Planner is the one most people misuse. It reports volumes for exact match and rounds them, so a keyword showing 50 searches a month may sit anywhere near that figure, and two different keywords can display the same number. Read the columns as bands, not measurements. DigiFix’s own free SEO tools will pull a page’s current targeting so you can see where the gaps sit before you add more.

How do you check what a searcher actually wants?

Before you judge a keyword’s numbers, read the page that already ranks first for it. The results tell you what Google has decided the query means, which is the intent you have to match or deliberately beat.

Three shapes cover most queries. Informational results are guides, definitions and videos. Commercial results are comparisons, reviews and pricing pages. Transactional results are service pages, booking forms and product listings. A keyword whose first page is all guides will not reward a sales page, however good the sales page is.

One more check catches mismatches early. If every result is a national brand or a directory, the query is probably informational even though it sounds commercial, and a local service page will struggle. If the results are local businesses with suburb names in their titles, you are looking at a commercial query with a local winner, and that is the kind worth chasing.

How should you read volume against difficulty?

Read the two numbers together, never alone. Volume tells you how many people ask, difficulty tells you how many sites are fighting for the answer. A keyword with 500 monthly searches and a weak first page is a better target than one with 5,000 searches and six national brands holding the top spots.

Difficulty scores are relative to the tool that produced them, so a 30 in one tool is not a 30 in another and neither maps to a ranking guarantee. Use the score to rank your own shortlist, then sanity check the top three results by hand: look at domain size, whether the page actually answers the query, and whether the content is thin. That manual look is what the score is trying to approximate.

How to read Keyword Planner volume bands. Figures are rounded and location dependent, so confirm each row for Australia on the day you build the list.

Set the location to Australia every time. A keyword that looks promising on global figures can have almost no local demand, and the reverse happens too. Spelling the location in your keyword list also keeps you honest about which pages should carry a suburb or city name.

What can competitor rankings tell you about gaps?

Your competitors have already done part of the research. Their rankings show which pages earn traffic and which keywords they have ignored, and the ignored ones are the cheap wins.

Take the three sites that rank for your main service term. For each, list the pages that appear for the same topic cluster, then compare against your own sitemap. Keywords where two or three competitors rank and you have no page are the clearest content gaps available. Keywords where nobody ranks a dedicated page are usually where a good page wins quickly.

Ignore their traffic estimates and look at their structure instead. A competitor with a page per suburb, a pricing page and three comparison guides is telling you what their market searches for. That pattern is more reliable than any third party volume estimate, and it costs nothing to read. If you want the tooling side of this, our round up of keyword research tools covers what each one is good for.

How many keywords should each page target?

One primary keyword per page, supported by a handful of close variants. That is the whole rule. Two pages targeting the same phrase split the internal links and the ranking signals, and neither tends to win.

The variants matter more than the count. A page about hot water system replacement can reasonably answer “how much does a hot water system cost” and “hot water system replacement Melbourne”, because the same content satisfies all three. It cannot reasonably answer “gas fitting certification”, which needs its own page.

This is where a keyword map earns its keep. Write one row per page with the primary keyword, the intent, and the action the page should drive. If two rows share a primary keyword, merge the pages or pick a different keyword. Before publishing, check the page’s keyword density reads naturally, because a page optimised to a percentage reads badly and ranks no better.

What does a keyword map look like for a plumber?

Here is the shape a map takes for a Melbourne plumbing business, built from the six steps above. The volume column is deliberately blank of invented figures: pull it from Keyword Planner with the location set to Australia on the day you build the map, because volumes move and Google rounds them.

A keyword map for one business. Each row is one page with one primary keyword, one intent and one action.

Two things fall out of a map like this. The urgent terms need the phone number within the first screen, because those searchers do not read. The price terms need a real table, because a vague answer sends that visitor straight back to the results. Both lessons come from the same place: the intent column decides the layout.

Do keywords still matter with AI Overviews?

Yes, and Google is unusually direct about it. Search Central states there are no additional requirements to appear in AI Overviews or AI Mode, no special optimisations needed, and that the best practices for SEO remain relevant. Keyword work has not been replaced, it has been re-priced.

What changed is the reward for a vague page. An AI answer assembles a specific reply, so a page that answers a precise question clearly is easier to lift from than one that circles a topic for 2,000 words. Question keywords, the phrasing you find in People Also Ask and Autocomplete, are now the highest value items on the list, because they match how people phrase a request to an assistant.

The 15 per cent figure explains why the list never closes. Google reports that 15 per cent of the queries it sees each day are new, so any keyword list is a snapshot of known demand, not the whole of it. That is also the argument for a fast publishing loop over a perfect plan: the page that answers a new phrasing first owns it.

If AI visibility is a priority, our guide to answer engine optimisation covers the structure that makes a page quotable. None of it works without the keyword work underneath.

If you would rather hand the whole job over, our digital marketing services cover keyword research, content and the technical work that lets a page rank. You can see the Australian businesses we have done it for in our portfolio, and read how the team works on the about DigiFix page. Everything else we have written sits in the DigiFix blog.

Search marketing changes faster than most business owners can track, and a keyword list built two years ago is usually out of date. The fix is a short, repeatable loop rather than a bigger spreadsheet.

What do people ask about finding keywords?

How many keywords should a single page target?

One primary keyword per page, plus two or three close variants the same content genuinely answers. Anything more and the page has no clear subject, which makes it harder to rank and harder for a reader to trust. If two pages in your plan share a primary keyword, merge them or move one to a different phrase before either gets published. This is the most common structural fault we find on small business sites, and it is the cheapest to fix.

Are free keyword tools good enough for a small business?

For most Australian small businesses, yes. Keyword Planner, Autocomplete, People Also Ask and Search Console together cover demand, phrasing, questions and current rankings, which is everything the six steps need. Paid tools earn their cost when you are tracking thousands of keywords or need historical volume trends. Start free, and only pay once a specific question cannot be answered without it.

What is a good keyword difficulty score?

There is no universal number, because each tool calculates difficulty its own way and a score from one does not translate to another. Treat the score as a way to rank your own shortlist against itself, then open the top three results by hand and judge whether those pages actually answer the query well. A weak first page is a better signal than any score, and it costs nothing to check.

Should a local business target long tail keywords?

Yes, and it should start there. Long tail phrases carry clearer intent, face less competition, and match how people describe a problem rather than name a service. Ahrefs reports that keywords with fewer than 10 monthly searches make up almost 93 per cent of its database, so the small terms are where most of the available demand sits. Build the small wins first, then use them to support the broader terms.

How often should keyword research be refreshed?

Once a quarter is enough for most sites, with one exception. Check Search Console monthly and act immediately on any query with rising impressions and a weak average position, because that is demand you are already visible for and are failing to convert. A full refresh makes sense when you add a service, enter a new suburb, or a competitor launches a page that outranks three of yours.

Can you rank for a keyword with no volume data?

Often yes. Tools show no volume when a phrase is too new, too specific or too rare to measure, and Google still serves results for it. If the phrase describes a real service you sell and one page can answer it properly, publish the page and let Search Console tell you whether impressions arrive. The cost of a small page is low, and unmeasured demand is where the 15 per cent of new daily queries shows up first.