Keyword research: what are we investigating and why?
Keyword research investigates which phrases an audience uses, why it searches for them and how those needs relate to a business’s content tasks. It connects customer questions, search data and result analysis. The outcome is a justified topic choice and page plan, rather than merely a large keyword list.
At a glance
- Prioritize the search need, the actual offer and the useful answer you can produce together.
- Volume, advertising competition and organic difficulty describe different things.
- Every number needs a source, target market and period; missing data must not be replaced with invented estimates.
On this page

What question does keyword research answer?
Start by asking what problem brings a person to search, rather than how many words can fit on a page. A keyword expresses that need. Research also examines the appropriate answer and whether the business can genuinely help.
“Coffee machine cleaning” may mean home maintenance, choosing a cleaner or arranging a repair. These are not automatically equivalent business opportunities. Interpreting search intent determines the page required.
Useful research therefore includes an editorial decision: what to explain, for whom, in what depth and alongside which existing page. A longer list helps only if it supports those decisions.
Where can initial phrases come from?
Recurring customer questions, wording in quote requests and accurate descriptions of the offer are useful starting points. These are observations, not measurements of total market demand.
Google Keyword Planner can suggest related ideas and estimates from keywords or a website. It is an advertising tool, so interpret its metrics accordingly. The Search Console performance report instead describes queries and performance associated with your own Google appearances.
Existing search queries provide feedback, but your site's report may reveal little about a completely new topic. These sources answer different questions rather than automatically substituting for each other.
A seed keyword, such as “coffee-machine repair,” can start the investigation. It can lead to distinct questions about faults, prices and supported models. The seed sets the research direction; it does not yet decide which question deserves its own page.
How should numbers be interpreted?
| Data | Context to record |
|---|---|
| Estimated monthly searches | Provider, country or location, period and network |
| Advertising competition | Advertising system and targeting settings |
| Trend index | Geography and comparison selection |
| Your search impressions | Report, page, dates and filters |
Keyword Planner documentation explains that monthly averages can include close variants and depend on selected settings. “Competition” concerns advertisers. It does not establish organic difficulty, and search volume is not your expected traffic.
Google Trends shows normalized relative interest from 0 to 100. A value of 100 does not mean 100 searches; zero with sparse data does not prove nobody searched. Trends can reveal timing patterns without supplying absolute counts.
Do not automatically dismiss a zero-volume keyword. If customers repeatedly ask about a part’s compatibility, a clear answer can be worthwhile even when a tool reports zero. Those customer questions support relevance, but they do not justify inventing a monthly search count.
Example: a coffee-machine repairer in Székesfehérvár
Imagine customers frequently asking why water leaks beneath their machines. The owner records “coffee machine leaking,” “coffee machine repair Székesfehérvár” and “coffee machine descaling.” These are ideas, not verified demand measurements.
The first could support safe initial checks and an explanation of when to contact a repairer. The second needs a local service page covering supported brands and drop-off arrangements. The third requires deciding which machine types can responsibly share instructions.
Research can then use suitable Hungarian-market settings and recorded sources. If a narrow phrase lacks usable data, its cell remains “no data.” A recurring customer question can still justify an explanation; it cannot justify an invented monthly count.
When should several phrases share a page?
Different phrasings of the same question can often share a thorough answer. Different decisions or actions may justify separate pages. Even a long-tail keyword does not automatically require another publication.
Examine the SERP: which content types appear, how deeply do they answer and what remains unclear? Record the query, location and time. Results support an inference about the engine's interpretation, not direct knowledge of every searcher's intent.
What belongs in a usable worklist?
Give each selected topic a primary reader question, planned page type and short rationale. Record related existing pages, required specialist information and uncertain data. This turns research into an editorial assignment.
Do not prioritize using one metric alone. A narrow question closely tied to the offer may be more useful than a broad, unmanageable topic. Refine decisions using real appearances and reader feedback; the initial list guarantees no result.
A content gap means an answer may be missing; keyword cannibalization concerns potentially overlapping pages. Keyword difficulty adds an estimate alongside offer relevance and result review.
Use SERP analysis to check content expectations and actual search competitors. Keyword Planner provides further estimates, but advertising data is not automatically a forecast of organic ranking success.
Begin with a defined target audience and the customer questions they actually ask. Customer research can reveal their wording and circumstances before a keyword tool supplies estimates.