Choosing keywords without guesswork

"Keyword research" sounds like a job for software. For most small businesses it is mostly a job for memory. The best list of search terms you will ever get is the list of questions customers ask you on the phone, at the counter and in emails. Tools are useful for checking and sorting that list, not for inventing it.

This note sets out a simple method that works for a café, a builder or an accounting practice alike.

Step one: write down the questions

Spend a week noting every question a customer or prospect asks. Write it the way they said it, not the way you would phrase it. A builder might collect:

• "How much does a deck cost?"
• "Do I need consent for a carport?"
• "How long does a bathroom renovation take?"
• "Do you do work in Rolleston?"

Those four questions already tell you more than a page of high-volume keywords. They show the words customers use ("consent", not "building authorisation"), the worries they have (cost, time, rules) and the places they live.

Step two: sort by intent

Every search has a purpose behind it. A practical way to sort your list is into three groups.

Intent	What the person wants	Example	What page answers it
Learn	To understand something	"do I need consent for a carport"	An explainer or guide
Compare	To weigh up options or prices	"deck cost per square metre nz"	A pricing or options page
Act	To find someone and get it done	"deck builder rolleston"	A service page for that area

Mixing these up is the most common reason a page never ranks. A service page stuffed with "how to build a deck yourself" text serves neither the DIY reader nor the person ready to hire.

Step three: favour the longer searches

Short, broad searches like "builder" are hard to win and often vague. Longer, more specific searches are each less common, but together they make up a large share of what people type, and they usually signal clearer intent. Someone searching "replace rotten deck boards Lower Hutt" knows what they want and where.

For a small New Zealand business the long, local search is usually the realistic target. You do not need thousands of searches a month. Ten searches from people in your area who need exactly what you do can be worth more than a thousand from people who do not.

Step four: check with free data

Two free sources are worth knowing.

Google Search Console. Once your site is verified, the Performance report shows the searches that led to your pages appearing in results, with the number of impressions, clicks, the click-through rate and the average position. This is real data about your own site, and it often reveals questions you had not thought to answer. Google keeps up to sixteen months of it, which is enough to see seasonal patterns.

Google Keyword Planner. This sits inside Google Ads. Google's help page says you need to finish setting up an Ads account, including billing details, before you can use it. It suggests related searches and gives estimates of monthly searches and typical cost per click. Google describes the search figures as estimates, so treat them as rough guides, especially for small New Zealand towns where the real numbers are low.

Other places to look:

• The "People also ask" questions and related searches at the bottom of a results page.
• The search box's own suggestions as you type.
• Questions in local community groups and forums, phrased exactly as people write them.

Step five: one page, one main question

Once your list is sorted, give each distinct question or service its own page, and group close variants together. "Deck builder Rolleston" and "deck builders in Rolleston" belong on the same page. "Deck builder" and "pergola builder" probably do not, because the jobs are different.

A simple planning sheet has four columns: the main question, the intent, the page that answers it (existing or new), and the evidence you have that people ask it. Anything without evidence goes to the bottom of the list.

Common traps

• Chasing volume. A term with big numbers may be dominated by national retailers or information sites you cannot realistically outrank.
• Repeating the phrase. Writing "deck builder Rolleston" eight times does not help. Write naturally and cover the topic properly.
• Forgetting the brand. Many searches are for your business name. Make sure your homepage and Business Profile answer those well first.
• Ignoring seasons. Searches for heat pumps, gutters and lawn care rise and fall with the weather. Look at a full year in Search Console before deciding a topic is dead.

Done this way, keyword choice stops being guesswork. It becomes a record of what your customers already ask, sorted so each page has a clear job.