Keyword research: choose queries for the US, UK and Canada

Start keyword research with a decision: what problem can your publication solve, and what useful action should follow the answer? For a blog serving online stores, a clear procedure is more valuable than an unreviewed list of thousands of variations. This workflow helps choose topics for the United States, United Kingdom and Canada, then assign each task to a suitable page.

Download the free keyword decision worksheet. It contains three fictional examples and fields to complete. It is a working template, not a volume export or a traffic forecast.

1. Define the task before choosing the phrase

Write the problem in ordinary language: make a store discoverable, prepare a promotion, or understand a product-feed migration. Identify the reader, their system, market and decision. “Ecommerce SEO” alone is a loose brief. “Check why Shopify product pages are not indexed” already identifies a task, a platform and the evidence to investigate.

Turn that task into a few natural phrases. Preserve actual customer wording when available. An AI suggestion remains a hypothesis until checked. Record what the proposed answer cannot solve, too. A web-indexing tutorial does not establish that Google Shopping will approve a product. Product eligibility and a web page’s indexing status require different checks.

Google’s SEO Starter Guide recommends understandable, useful content. Keywords should describe that content. Repeating every variation in the title or producing a page for each variation does not establish relevance.

2. Separate country, language and search engine

Create a row for each query and market. “English” is a language, not a country. A US estimate does not become a Canadian estimate because the wording matches. Within Canada, French and English searches can also concern different readers and results.

For a contribution-margin example, use the case’s actual currency and units: USD, GBP or CAD when applicable. For a country-independent sitemap procedure, a shared guide may serve all three markets. A local version needs a meaningful difference in the process, conditions or evidence. Appending a city list to a general article does not create that difference.

Keep a separate engine column. An estimated Google volume does not measure Bing searches, Brave searches or AI prompts. Inspect those results where they matter to your audience and retain the observations independently.

3. Distinguish an estimate from a measured outcome

Our October 3, 2026 review includes these three Semrush observations. They illustrate how to interpret a candidate, rather than proving our blog ranks or forecasting its clicks.

The provider defines search volume as an average monthly measure for a geographical database. Averages smooth seasonal peaks. KD is a provider indicator; a low value does not guarantee that your page will outrank existing results. An unavailable value remains blank instead of becoming zero.

Keep four distinct fields: idea source, dated estimate, current page status and measured result. When Search Console has no usable data, record “unavailable.” A tool returning no rows does not show that nobody asks the question.

4. Inspect the answer a reader expects

Search the phrase in the intended geographical context and record the date. Inspect the formats offered: product listing, category, comparison, tutorial, official documentation or discussion. Read the sections that actually answer the task. This can change the brief; it is not a ranking formula.

Write one sentence describing what your own answer lacks. For example: “We describe a sitemap, but do not explain how a successful fetch differs from an unknown URL in the index.” That gap may justify a useful revision. “A competitor has twenty headings, so we need twenty headings” describes no reader problem.

Shopify’s keyword research guide treats search intent as a selection criterion. Also check the evidence required by your chosen format. A comparison needs defensible criteria. A tutorial needs executable steps. A claim about an outcome needs actual measurements.

5. Group variations by the decision they support

Group phrases requiring the same procedure. “How to plan a Black Friday email” and a close variation can share an answer if the reader must establish the same offer, deadline and destination. Keep variations in the research file for later observation; they do not require separate published articles.

Separate requests that change the work. Choosing a Shopify agency needs evaluation criteria and commercial evidence. Diagnosing an unindexed URL requires a technical process. Combining them merely because both contain “Shopify” weakens the answer.

Do not add the volumes of close variants to announce an audience. One person can perform several searches, and providers can group or estimate related wording. This worksheet supports a task and destination decision, rather than manufacturing a total of future visitors.

6. Prioritise with explicit criteria

First decide whether the topic fits your work, whether you can answer it and whether a relevant page exists. Then compare effort, available evidence, seasonality and the reader’s next action. A homemade score may organise the queue, but it is your convention rather than a Google model.

The worksheet offers three decisions: update an existing answer, create a distinct answer, or verify before writing. For Q4, seasonality can make a correction time-sensitive without guaranteeing that a new page will be discovered before the campaign. Improve the resource a visitor can use immediately: a worksheet, calculation or set of checks.

A useful writing brief records the central question, expected format, sources, any explicitly fictional scenario and limitations. The writer can then identify what must be explained and what requires checking before publication.

7. Connect the answer to a resource, then measure

The next step should help a reader apply the answer. Our Q4 dossier connects promotion planning with free resources. Our sitemap diagnostic protocol addresses technical states. Pick the destination according to the task and label the link with what the reader will find.

After publication, observe discovery, indexing, impressions, clicks, resource use and visits to an offer separately. The Search Console performance report offers queries, pages and countries when available. Use consistent periods and definitions. A Semrush volume is not a denominator for calculating your page’s click-through rate.

More clicks do not establish a sale. Orders need an attribution method that is actually configured. Leave missing values blank and record editorial changes so that later observations can be interpreted.

Make the first three decisions

Choose three concrete tasks, using a different market for each where the context justifies it. Complete the worksheet, find a suitable existing destination and list missing evidence. Then choose one answer to improve. A useful research register grows through reviewed decisions rather than a rising row count.

Continue with our search engine, keyword and page-discovery library. It connects free methods with official sources and the checks needed to turn an idea into a publishable answer.

SEO Foundations

Explore this related DIY Marketing Guide guide to work further on the method. DIY is part of our business group; contents, language and price are shown on its store.

See the English guide