SEO Fundamentals: Keyword Research and Search Intent

An online houseplant shop wants to rank for "monstera". It is a short, popular word, and the shop imagines thousands of buyers. Search the term and the results are mostly care guides, plant encyclopedias and images. People typing that single word are mostly curious, not shopping. Meanwhile "buy variegated monstera cutting" has a fraction of the volume and a much higher chance of ending in a sale. The skill in keyword research is not finding big numbers, it is finding the words your actual audience uses and understanding what they want when they use them.

Search intent: the question behind the query

Search intent is the goal behind a query. Search engines work hard to match it, so a page that does not fit the intent will struggle however well it is optimised. A rough, widely used classification has four types, and the table shows what page tends to satisfy each.

A quick heuristic you can automate is looking for tell-tale words. The script below is deliberately crude, since real intent depends on context and on what the results page shows, but it is a useful way to sort a long list into rough buckets before you review it by hand.

INTENT_HINTS = {
    "informational": ("how to", "what is", "why", "guide", "vs", "ideas"),
    "commercial": ("best", "review", "top", "compare", "alternatives"),
    "transactional": ("buy", "price", "cheap", "order", "discount", "near me"),
    "navigational": ("login", "sign in", "official site", "contact"),
}

def guess_intent(query: str) -> str:
    q = query.lower()
    for intent, hints in INTENT_HINTS.items():
        if any(h in q for h in hints):
            return intent
    return "unclear - read the results page"

for q in [
    "how to repot a monstera",
    "best low light houseplants",
    "buy pothos cutting online",
    "leafy login",
    "monstera",
]:
    print(f"{q:<30} -> {guess_intent(q)}")

how to repot a monstera        -> informational
best low light houseplants     -> commercial
buy pothos cutting online      -> transactional
leafy login                    -> navigational
monstera                       -> unclear - read the results page

The last line matters most. When a keyword is ambiguous, the right move is to read the actual results page, because it shows what the search engine currently believes people want: a wall of blog posts means informational, product grids mean transactional.

Where keyword ideas come from

• Your own customers: support emails, chat logs, sales calls and reviews contain the exact phrases people use, which are often not the ones your team uses internally.
• Search Console: once a site has data, the Performance report shows real queries that already trigger impressions. Pages ranking just below the first page for a good query are the cheapest opportunities.
• Autocomplete and related searches: typing the start of a phrase, and scrolling to the related searches and question boxes, reveals common variations.
• Competitors' pages: read their headings and the questions they answer, then look for what they miss rather than copying them.
• Forums and communities: the questions people ask in their own words, including confusion and complaints.
• Keyword tools: free and paid tools give volume and difficulty estimates. Treat those numbers as relative guides, not facts, because the tools model them and often disagree with each other.

Turn the list into a page map

A list of keywords is not a plan. Group keywords that share intent and could be answered by the same page, then assign one primary page per group. If two pages target the same query, they compete with each other and neither is clearly the best answer, a problem sometimes called keyword cannibalisation. A simple worksheet keeps this honest.

PAGE: /guides/repot-monstera           TYPE: guide
  Primary query   : how to repot a monstera
  Variations      : when to repot monstera, repotting monstera soil mix
  Intent          : informational
  Goal            : newsletter sign-up; link to potting-mix category
  Notes           : results show step lists with photos -> include both

PAGE: /plants/monstera-cuttings        TYPE: category
  Primary query   : buy monstera cutting
  Variations      : variegated monstera cutting, monstera cutting for sale
  Intent          : transactional
  Goal            : add to basket
  Notes           : keep price, stock and delivery info visible

PAGE: /guides/best-low-light-plants    TYPE: comparison
  Primary query   : best low light houseplants
  Intent          : commercial investigation
  Goal            : click through to product pages
  Notes           : explain who each plant suits; do not list 40 items

NOT TARGETING (for now): "monstera" alone (mixed intent, high competition)

Judging value and difficulty

Pick targets by combining three things: relevance to what you offer, intent that matches a page you can make, and a realistic chance of competing. For a new site, long, specific phrases (often called long-tail queries) are usually the practical starting point, because the results are less crowded and the visitor knows what they want. Look at the current results page before committing. If the top ten are all large, established publishers with deep content, and your page would be a thin summary, choose a narrower angle where you have real experience to add.

Common mistakes

• Targeting broad head terms first. Symptom: months of work and no impressions for a one-word query dominated by large sites. Fix: start with specific phrases tied to what you sell or know.
• Ignoring intent. Symptom: a product page ranks poorly for a query where every result is a guide, and visitors who do arrive leave at once. Fix: match the page type to the intent, or choose a different query.
• One page for everything. Symptom: a long page stuffed with many unrelated queries that answers none properly. Fix: one page per intent group.
• Trusting tool volumes literally. Symptom: forecasts of thousands of visits that never appear. Fix: use volumes to compare options and confirm with real Search Console data.
• Writing for the keyword, not the reader. Symptom: awkward phrasing and repeated terms. Fix: write naturally and cover the topic fully; related terms will appear on their own.

How this shows up in real projects and interviews

In practice, keyword research is an hour-long exercise repeated each quarter, not a one-time project, and the most useful inputs are free: customer language and Search Console. In interviews for marketing, content or product roles, a strong answer to "how would you decide what to write about?" starts with the audience's problems, uses intent to choose the format, and ends with how you would measure results. If you are building a personal site to attract recruiters, apply the same idea: think about the phrases a hiring manager would type, and see how writing a resume that passes ATS uses the same matching logic.