Keyword Research: How to Find, Evaluate, and Prioritize Search Opportunities

What is keyword research?

Keyword research is the process of identifying, organizing, and evaluating the searches people make so you can decide which opportunities are relevant to your website and what to do about them.

It helps you understand what your audience is looking for, how different searches relate to each other, and which resources could best serve those needs.

For example, imagine a company selling employee onboarding software.

Someone searching for "employee onboarding software" may be evaluating products. Someone searching for "employee onboarding checklist" may want a practical resource they can use at work. Both searches are relevant to the company, but represent different needs and may require different pages.

Keyword research helps you recognize those differences and decide what to do about them. That may mean improving an existing page, creating something new, consolidating overlapping content, maintaining a resource that already serves its purpose, or deciding not to pursue an opportunity.

How to conduct keyword research in 8 steps

A complete keyword-research process follows eight practical steps:

Together, these steps take you from a broad set of potential searches to a focused set of decisions about where to invest.

The process is useful whether you're researching keywords for a new website, improving an established content program, or investigating opportunities within a specific topic.

Keyword research in practice: Beacon

Bhavya leads marketing at Beacon, a B2B software company that helps growing businesses manage employee onboarding.

Beacon gives HR teams one place to assign onboarding tasks, coordinate work across managers and departments, and track each new hire’s progress from offer acceptance through their first weeks on the job.

The company's website includes product and feature pages, an employee onboarding checklist, and a small collection of educational resources for HR professionals.

Beacon already receives some organic search traffic, but Bhavya wants to attract more businesses that could become customers. Her objective is to increase qualified demo requests from businesses discovering Beacon through organic search.

She has resources to create or substantially improve only two pages during the next planning period.

Bhavya needs to determine which search opportunities are relevant, whether Beacon's existing pages can serve them, and where those two projects would provide the most value.

We'll follow her through the eight steps as she turns an initial keyword dataset into a prioritized content plan.

Beacon and Bhavya are fictional. All business details, keyword measurements, SERP observations, decisions, and outcomes in this example are illustrative.

Step 1: Define your objectives and search territory

Before opening a keyword-research tool, establish what your website is trying to achieve and which audience you want to reach.

Keyword research can reveal thousands of searches related to your industry, but not all of them will contribute meaningfully to your goals.

Start by defining the business outcome you want organic search to support.

Start with your business objective

Different organizations need organic search to contribute to different outcomes.

Your business objectives determine what makes a search opportunity valuable.

A publisher may prioritize keywords that attract readers interested in its subject matter. A software company may place greater emphasis on searches made by people who could benefit from its product.

This doesn't mean every keyword needs to lead directly to a sale.

Educational searches can help people understand a problem, evaluate their options, or discover a business before they're ready to take action. For example, an HR manager searching for “how to improve employee onboarding” may not be looking for software yet, but a useful guide could introduce them to Beacon while they’re still figuring out how to solve the problem.

The important question is how the search need relates to your audience and what your organization wants to accomplish.

Identify your audience and the problems you can help solve

Next, define who you want to reach and why they might search.

Instead of starting with your products or industry terminology, start with the problems, questions, and decisions your audience faces. What are they trying to understand, compare, or accomplish? Where can your business genuinely help?

For example, an accounting software company might serve small-business owners who need to send invoices, understand cash flow, or prepare financial reports. Those needs can lead to search terms around invoicing, cash-flow management, and financial reporting, beyond just keywords that describe the software itself.

Pay attention to the language your audience uses. Customer interviews, sales conversations, support requests, and internal site searches can reveal terms and questions that your team may not use internally.

The goal is to identify the needs your business can credibly serve, not every topic your audience might search for.

Define the scope of your research

A useful keyword-research project needs boundaries.

Decide which part of your business you’re researching and which audience or market the research should cover. If you offer several distinct products, for example, it may make sense to investigate one product area at a time.

Geography and language can also matter. A business operating in several countries may find that people search differently or encounter different results in each market.

Your existing website should shape the scope too. Consider which pages already serve the territory you’re researching and how much capacity you have to create or improve content once the research is complete.

Aim for a scope broad enough to uncover meaningful opportunities but focused enough to evaluate them properly and act on what you find. A focused research project that leads to clear decisions is more useful than an enormous keyword list with no path to action.

Example: Bhavya defines Beacon's research objective

Bhavya starts with Beacon’s business goal: increase qualified demo requests from businesses discovering Beacon through organic search.

For this project, a qualified demo request means an inquiry from a business that fits Beacon’s intended customer profile, rather than any visitor who submits a form.

She identifies HR professionals at growing businesses with 100 or more employees as the audience Beacon wants to reach, then defines three initial areas of research: employee onboarding software, practical onboarding resources, and guidance on managing the onboarding process.

Bhavya also reviews Beacon’s existing website. The company already has a product page, several feature pages, and an employee onboarding checklist, so some opportunities may be better served by improving existing resources than creating new ones.

To keep the research consistent, she focuses on English-language searches in the United States.

Finally, she accounts for Beacon’s capacity: the team can take on only two substantial content projects during the next planning period.

Bhavya now has the context she needs to evaluate opportunities properly: what Beacon wants to achieve, who it wants to reach, what content already exists, which search territory matters, and how much work the team can realistically take on.

Step 1 output: A defined business objective, audience, research territory, existing-content context, and resource constraint.

Step 2: Discover relevant search queries

Once you know what you’re trying to achieve, the next step is to identify the searches connected to your audience’s needs.

No single source will reveal the full picture. A useful keyword universe combines what you already know about your customers with evidence from search performance, competitors, keyword tools, and the language people use to describe their problems.

Start with seed topics

Seed topics are the broad concepts you use to begin keyword discovery. They usually come from your products or services, the problems you solve, important use cases, and the questions your audience asks.

For an employee onboarding software company, seed topics might include employee onboarding, new-hire experience, and HR workflow automation.

Use these topics as starting points for uncovering more specific searches, rather than treating them as keywords you should automatically target.

A single topic can contain several distinct search needs. Someone searching for onboarding software may be evaluating products, while someone searching for an onboarding checklist may simply want a practical resource. Those differences become important later when you analyze intent and organize the queries you’ve discovered.

Use Search Console to discover existing visibility

If your website already appears in search results, Google Search Console is one of the best places to start. Its Performance report shows the queries for which your website appears in Google Search, along with metrics like impressions, clicks, click-through rate, and average position. [1]

Start with the pages that matter to your business and examine the searches already bringing them visibility. Look for queries that:

• Generate meaningful impressions but relatively few clicks
• Already attract clicks to an important page
• Surface a page that only partly serves the apparent need
• Reveal adjacent searches you had not considered

Then look at the page receiving that visibility.

For example, if a product page appears for several related searches, those queries can reveal demand Google already associates with that resource. An educational page may also surface for searches beyond its primary topic, exposing adjacent needs worth investigating.

That existing visibility is useful evidence because you are not starting from a hypothetical opportunity. Your website is already appearing for the search in some capacity.

Search Console still provides only part of the picture. Some queries are anonymized and omitted from query-level tables, and the available table data may not include every recorded query. [2]

Use it to understand where your website already has visibility and what that visibility may reveal, not as a complete inventory of relevant search demand.

If your website is new or has little meaningful search history, rely more heavily on customer language, business knowledge, competitor research, and broader keyword-discovery tools.

Analyze relevant search competitors

Competitors can reveal search opportunities that your seed topics and existing performance data may miss.

Your search competitors aren't always the same as your direct business competitors. A software company may compete with other software providers for product-related searches, while publishers, specialist websites, and professional organizations may appear for educational searches.

Look at the websites that repeatedly appear for searches relevant to your audience, then use competitor-research tools to identify related queries and topics where they have visibility that you do not.

Treat those differences as leads to investigate, not gaps you automatically need to close. A competing website may rank for searches that have little relevance to your audience, product, or business objectives. Third-party keyword and traffic measurements may also be estimates rather than directly observed performance.

At this stage, the goal is simply to uncover potentially relevant search needs and add them to your working dataset. For a deeper framework for identifying search competitors and evaluating the opportunities they reveal, see our SEO Competitor Analysis guide.

Expand your research with keyword tools

Keyword-research tools can turn broad seed topics into more specific searches, questions, modifiers, and adjacent ideas.

For example, starting with employee onboarding, new-hire experience, or HR workflow automation might uncover searches related to onboarding software, employee orientation, workflow automation, templates, and specific onboarding challenges.

These tools can also provide useful context such as estimated search volume, keyword difficulty, geographic demand, competing pages, and related questions.

Google Keyword Planner can provide keyword ideas and historical search metrics, although it is primarily designed for advertising research. Its Competition metric measures advertiser competition rather than organic ranking difficulty. [3]

Use these metrics to understand the search landscape rather than decide which opportunities deserve investment. At this stage, the goal is to expand your working dataset. Evaluation comes later.

Incorporate customer language and search-result discoveries

Keyword databases are only one source of demand, and customers may describe their problems differently from your marketing team or the terminology used in your industry. Sales conversations, support requests, customer interviews, internal site searches, and product feedback can reveal questions and phrases that keyword tools overlook.

For example, Beacon may describe part of its product as HR workflow automation, while customers talk about assigning onboarding tasks, keeping managers on track, or making sure nothing gets missed during a new hire’s first week. Those phrases can lead to search needs worth investigating.

Search results can reveal additional ideas too. Related searches, recurring terminology, questions, and the subjects covered by prominent pages may expose adjacent needs that were not obvious from your original seed topics.

Add relevant discoveries to your working dataset even when reported search volume is low or unavailable. The next steps will determine whether the underlying need is meaningful, distinct, and worth pursuing.

Example: Bhavya builds Beacon's keyword universe

Bhavya begins with Beacon’s three seed topics: employee onboarding, new-hire experience, and HR workflow automation.

She reviews Search Console to identify searches already generating visibility for Beacon’s product page, feature pages, and onboarding checklist. She also looks at recurring language from customer conversations, then uses competitor research and keyword tools to expand beyond the searches Beacon already knows about.

Her research produces the following representative selection:

These are discovery categories, not finalized keyword clusters. Some queries may ultimately belong together, others may represent distinct search needs, and some may fall outside Beacon’s relevant search territory.

Bhavya records each query, where it came from, and the available search-performance or demand data. She now has a working dataset to clean, organize, and investigate.

Step 2 output: A representative keyword dataset containing potentially relevant searches, their discovery sources, and available measurements.

Step 3: Clean and organize your keyword data

Keyword discovery often produces a messy dataset. Different sources may return duplicate queries, alternative wording, irrelevant searches, and metrics that are not directly comparable.

Before analyzing intent or grouping keywords, clean and organize the data so you can evaluate it consistently. The goal is to remove unnecessary noise without erasing differences that may matter later.

Remove duplicates and clearly irrelevant queries

Start with the easy cleanup: remove exact duplicates and searches that clearly fall outside your research territory. If several sources identify the same query, keep one working entry while preserving useful information about where it came from.

Be more careful with similar wording. Two queries that look alike may represent the same search need, but you should not assume that during cleanup.

For example, “employee onboarding checklist” and “new hire onboarding checklist” may ultimately belong together. That decision should come after you examine their intent and search results.

The reverse can also happen. “Employee onboarding software” and “customer onboarding software” share similar language but describe different business processes.

At this stage, remove what is clearly irrelevant and preserve the distinctions you still need to investigate.

Identify useful query characteristics

A query can be described in several ways at once. The following distinctions can help you organize your research.

You don't need to apply every label to every query. Only record the characteristics that help you understand meaningful differences in the dataset.

For example, “onboarding software for small businesses” includes an audience qualifier that may affect what the searcher expects. Preserve that distinction until you examine the search results, even if the query eventually belongs in a broader software cluster.

Understand search volume

Search volume estimates how often a query is searched within a defined market and reporting period. It helps you compare the relative size of potential demand, but is not a stand-in for how much traffic your website could receive.

Google Keyword Planner’s average monthly searches metric, for example, reflects factors including close variants, location, network, and the selected reporting period. [3]

Third-party platforms may use different datasets and estimation methods, so their figures should be treated as directional rather than exact.

Keep four limitations in mind:

• Historical averages can hide seasonality or emerging demand
• Different tools may report different estimates for the same query
• Closely related queries may share or overlap in reported volume
• Search volume measures searches, not unique people, website visits, or business outcomes

A high-volume query may generate relatively few organic clicks, while a page that serves several related searches may attract traffic beyond the reported volume of any single keyword. [4]

For the same reason, avoid simply adding every keyword’s volume together to estimate the size of a cluster. Overlapping measurements can make the total look larger than the underlying demand.

Understand keyword difficulty

Keyword difficulty is a proprietary metric used by SEO platforms to estimate how difficult a query may be to rank for. Different platforms calculate it differently.

Ahrefs, for example, bases its standard Keyword Difficulty metric on the number of websites linking to the top-ranking pages for a keyword. That doesn't fully account for the competitive position of your own website. [5]

A difficulty score can help you decide which opportunities deserve closer investigation, but it doesn't tell you whether your site can compete.

A low-difficulty query may still be a poor fit if the search intent does not match your offering or the results require a type of experience you cannot provide. A more difficult query may still deserve investment if it serves an important business need and your site has relevant expertise, existing visibility, or other meaningful advantages.

Use keyword difficulty as a signal, then judge the opportunity in context.

Don’t define long-tail keywords by word count

Long-tail keywords sit in the lower-frequency portion of search demand. The term refers to how rarely a query is searched, not how many words it contains. [6]

A short query can have relatively little demand, while a longer phrase can be searched frequently.

The useful distinction is often the specificity of the need. A more specific query may reveal a particular audience, problem, or requirement that a broader search does not.

However, specificity does not automatically make a keyword more valuable or easier to rank for. Focus on the underlying search need, not the number of words in the query.

Example: Bhavya organizes Beacon's data

Bhavya removes exact duplicates, records meaningful characteristics and modifiers, and separates clearly irrelevant searches from those that still need investigation.

She excludes “customer onboarding software” because it refers to onboarding customers rather than employees, placing it outside Beacon’s research territory. She keeps “onboarding app” for further investigation because the phrase is ambiguous. It could refer to employee onboarding software, but the query alone does not provide enough information to know.

Bhavya also keeps similar employee-onboarding queries separate for now. She does not assume that searches such as “employee onboarding checklist” and “new hire onboarding checklist” belong together just because they have similar wording. That comes later, after she examines their intent and search results.

A simplified extract from her organized keyword data looks like this:

All figures are illustrative estimates of monthly US searches, not actual market measurements. A dash indicates that no estimate is being used in this example.

The highest-volume query in the table is “customer onboarding software,” but its demand does not make it relevant to Beacon. Bhavya excludes it rather than allowing search volume to override the scope she defined in Step 1.

She also avoids adding the remaining volume estimates together. The queries may overlap in ways that are not yet clear, and she has not established which searches represent the same underlying need.

Bhavya now has a cleaner dataset that preserves meaningful differences between queries. The next step is to examine what those searches actually mean.

Step 3 output: A cleaned and organized keyword dataset with irrelevant searches removed, meaningful distinctions preserved, and important uncertainties flagged for further investigation.

Step 4: Understand search intent and analyze the SERPs

Once your keyword data is organized, investigate what people are trying to accomplish when they search.

Search intent is the underlying goal behind a query.

Two searches can concern the same subject while reflecting different needs. Someone searching for onboarding software may want to evaluate products, while someone searching for an onboarding checklist may want a resource they can use immediately.

Understanding that difference helps you judge whether an opportunity is relevant and what kind of resource could serve it.

The four common types of search intent

Search intent is commonly grouped into four broad categories. [7]

These categories are useful shorthand, but search intent doesn't always fit neatly into one box.

For example, classifying “employee onboarding checklist” as informational doesn't tell you whether searchers primarily want an explanation, a printable checklist, an editable template, or some combination of those resources. To understand the need more clearly, you need to examine the search results.

Analyze the actual search results

The search engine results page, or SERP, provides evidence about how a query is being interpreted and the kinds of resources currently serving it.

Google’s ranking systems consider several factors, including query meaning, content relevance, usability, and search context. Results can also vary by location, language, and other circumstances. [8]

For important opportunities, inspect the results under conditions that reasonably reflect your intended market, and consider:

For example, “employee onboarding checklist” may return practical resources people can copy, download, or adapt. If you were considering a purely explanatory article, that pattern would suggest searchers may expect something more usable.

That doesn't mean copying the ranking pages will make your page rank. SERP analysis helps you understand the search need and competitive environment. It's not a formula for ranking.

Account for mixed search intent

Not every query represents one clearly defined need.

A broad search may surface several types of results because different people use the same words for different purposes. For example, “onboarding app” could refer to employee onboarding, customer onboarding, or another onboarding experience.

If the SERP reflects several competing interpretations, do not force the query into a cluster yet. Investigate more specific searches and determine which need, if any, is relevant to your audience.

The same applies when several page types appear. A mixed SERP does not mean you need to create every format represented in the results.

Consider AI-generated search experiences

Search results may also include AI-generated experiences such as AI Overviews and AI Mode.

When evaluating an opportunity, consider whether these features appear, what information they provide, and how they change the search experience. A substantial search-volume estimate does not necessarily translate into a comparable number of website visits, particularly when users can get information directly from the results.

Google Search Console reports on visibility in supported AI-generated search experiences, including impressions from AI Overviews and AI Mode. [9] As with other Search Console metrics, an impression is not the same as a website visit or business outcome.

Don't dismiss an opportunity just because an AI-generated answer appears. Instead, ask whether your website can still provide something useful for the underlying need and whether that visibility is relevant to your objectives.

Example: Bhavya examines Beacon's search opportunities

Bhavya selects representative queries from the cleaned dataset, examines their search results, and records the following patterns:

She then examines several more specific searches.

For “employee onboarding checklist” and “new hire onboarding checklist,” she finds substantially similar result patterns. “Employee onboarding checklist template” places greater emphasis on editable or downloadable resources, but many of the same types of pages still appear.

That suggests the three searches may represent closely related needs.

By contrast, “employee onboarding process” produces a different pattern. The results focus more broadly on designing and managing onboarding rather than providing a practical checklist.

Bhavya now has evidence about what the queries appear to represent, which resources currently serve them, and where uncertainty remains. She has not yet decided which searches belong together or which pages Beacon should create or improve. Those decisions come in the next steps.

Step 4 output: An evidence-supported understanding of the search needs behind the queries, the result patterns associated with them, and any remaining uncertainty.

Step 5: Group keywords by shared search need

The next step is to group related queries into keyword clusters.

A keyword cluster is a group of searches that represent the same or substantially similar need and can be served by the same page.

The goal is not to group every keyword about the same subject. It is to determine which searches belong on the same page and which require something different.

Focus on whether one page can serve the need

Queries can use different words to represent the same search need, or similar ones to reflect different goals.

Ask yourself: Could one page satisfy these searches well?

If yes, the queries may belong together. But if they require meaningfully different information, functionality, audiences, or page purposes, keep them separate.

Use SERP overlap as evidence

Comparing the ranking pages for different queries can also help you determine whether they represent the same search need.

If two queries consistently produce many of the same pages, that overlap is useful evidence that one resource may serve both. If the results are substantially different, the queries may represent distinct needs. [10]

However, SERP overlap is not an absolute rule. Search results can vary by location, device, language, and time. Broad or ambiguous queries may also produce mixed results. Use overlap alongside your understanding of the searcher's need and the actual resources appearing in the results.

Don't treat a particular overlap percentage as a universal threshold for merging or separating queries.

Distinguish semantic similarity from search-need similarity

Semantic similarity describes how closely related two queries are in language or meaning. Search-need similarity asks whether the same resource can satisfy them.

That distinction matters when using automated clustering tools. A system may identify related terminology without establishing that the queries should be served by the same page.

For important opportunities, validate proposed clusters against search intent and SERP evidence.

You should be able to explain why the queries belong together and what the page serving them is meant to accomplish.

Example: Bhavya groups Beacon's keywords

Bhavya uses the illustrative SERP evidence to group searches that appear to represent the same underlying need.

Her working groups are:

The first three become keyword clusters because the queries within each group appear to share a search need that one resource could serve.

The software cluster includes the small-business variation because Bhavya’s illustrative results suggest it reflects the same underlying software-evaluation need.

The checklist cluster combines three searches that could be served by a practical resource with both guidance and a usable template.

The process cluster remains separate because its broader instructional purpose differs from the checklist need.

“Remote employee onboarding” and “employee orientation activities” remain standalone searches for now. Bhavya has not found enough closely related queries to justify grouping them with another need.

She excludes “customer onboarding software” because it falls outside Beacon’s offering and leaves “onboarding app” unresolved because its meaning remains ambiguous.

Bhavya now has a smaller set of meaningful search opportunities instead of treating every discovered keyword as a potential page.

Step 5 output: Keyword clusters organized by shared search need, with standalone, irrelevant, and unresolved searches handled separately.

Step 6: Map opportunities to existing and potential pages

Once related searches are grouped, determine which resource should serve each relevant search need.

Keyword mapping is the process of connecting a search need or keyword cluster to an appropriate existing or potential page.

Before creating anything new, examine the resources you already have.

Evaluate your existing coverage

For each search need, determine whether an existing page already serves it. Consider the page’s purpose, audience, content, functionality, and current search visibility. Ask:

• Does the page address the underlying search need?
• Is it intended for the right audience?
• Does it provide the information or experience searchers need?
• Does the need fit naturally within the page’s scope?
• Does the page already receive visibility for relevant searches?
• Could the existing resource serve the need better with meaningful improvements?

A page can be the right destination even if it is not performing particularly well. If it serves the correct need but lacks important information or functionality, improving it may make more sense than creating something new.

Conversely, a broad article may not adequately serve a more specific need, even when the subjects are closely related. The decision should follow from what the searcher needs and what the existing page is meant to accomplish.

Decide what should serve each need

Different situations call for different mapping decisions.

These are possible outcomes, not automatic rules.

Several pages receiving visibility for related searches does not necessarily mean they are competing unnecessarily. They may legitimately serve different needs within the same subject.

Consolidation makes sense when pages substantially overlap in both purpose and coverage, and one resource could serve the need more clearly. Likewise, a distinct keyword does not automatically justify a separate URL.

The goal is to create a clear relationship between search needs and the resources intended to serve them.

Check for unnecessary overlap before creating a new page

Before planning a new resource, ask whether an existing page could reasonably serve the need. A separate page is more appropriate when the opportunity requires a meaningfully different purpose, audience, information, or functionality.

For a deeper framework for evaluating existing pages, identifying overlap, and deciding whether to improve, consolidate, or create something new, see our Content Optimization guide.

Example: Bhavya maps Beacon's opportunities

Bhavya compares Beacon’s search needs and keyword clusters with the resources already on its website. She starts with two pages that already receive relevant search visibility.

The first is Beacon’s employee onboarding software product page. It explains the platform’s general benefits but provides limited detail about how its features help HR teams organize tasks, coordinate onboarding activities, and track progress.

Bhavya’s illustrative SERP review shows that several competing resources provide more detailed information for people evaluating onboarding software. Beacon already has a page serving this need, so she maps the software cluster to the existing product page and identifies an opportunity to improve it.

The second resource is Beacon’s employee onboarding checklist. It already attracts relevant searches, but it is outdated, provides limited guidance on assigning and organizing onboarding tasks, and lacks a practical format that HR teams can easily adapt.

Bhavya considers whether the template-related searches need a separate page. However, the illustrative SERP evidence suggests that the checklist and template needs could be served by a stronger version of the existing resource.

She therefore maps both to the existing checklist rather than proposing another page. Her initial keyword map looks like this:

All website details, SERP findings, and mapping decisions in this example are illustrative.

Bhavya now knows which needs have a natural destination on Beacon’s existing website, which may require a new resource, and which still need more investigation.

She still hasn't decided which projects deserve immediate investment. That comes next, when she evaluates the opportunities against Beacon’s objective, competitive reality, and available resources.

Step 6 output: A keyword map connecting relevant search needs to existing or potential pages, with unresolved needs clearly identified.

Step 7: Evaluate and prioritize opportunities

Keyword research can reveal more worthwhile opportunities than your organization has the resources to pursue.

The next step is to decide which deserve investment and which should come first. Search volume and keyword difficulty can inform that decision, but neither tells you whether an opportunity makes sense for your business.

Evaluate each meaningful opportunity across four factors.

1. Relevance and fit

Does the search need fit what your organization can credibly provide?

Consider whether you can provide the information, product, functionality, or experience the search requires. If there is a fundamental mismatch between the search need and what your organization can appropriately offer, exclude the opportunity.

2. Business value

How could serving this need contribute to your objectives?

Consider who the searcher is, how closely the need relates to your offering, and what role the resource could play in their journey.

A commercially focused search may be closely connected to a purchase or inquiry. An educational search may help relevant people discover your business, understand a problem, or explore possible solutions. Both can matter, but they contribute differently.

3. SERP reality

What does the current search environment suggest the opportunity requires?

Use the evidence from your SERP analysis to judge what pursuing the opportunity would involve. The resources, functionality, competitors, and search features present in the results can help you understand what you would need to provide and the environment you would be entering.

4. Ability to compete

Does your website have a credible path to serving this opportunity well?

Consider your existing coverage and visibility, the strength of competing resources, your relevant expertise or advantages, and the capabilities required to produce something genuinely useful.

The goal is to judge the opportunity in the context of your website, not a generic difficulty score.

Compare value with effort

Once you know which opportunities are viable, compare what they could contribute with what they would require.

Improving an existing page with relevant visibility may take less effort than creating a new resource from scratch. Another opportunity may offer greater potential value but require research, functionality, expertise, or production resources you do not currently have.

There’s no universal scoring formula for resolving these tradeoffs. Instead, prioritization should answer three questions:

1. Is this opportunity worth pursuing?
2. What would pursuing it require?
3. Why should it come before the other work we could do?

Example: Bhavya prioritizes Beacon's opportunities

Bhavya returns to Beacon’s objective: increase qualified demo requests from businesses discovering the company through organic search.

She also returns to the constraint established in the Define step: Beacon can take on only two substantial content projects during the next planning period. Before comparing the opportunities, she reviews the existing performance of the two pages identified for improvement.

Beacon’s existing organic search performance

Search impressions and clicks are measured in Search Console. Qualified demo requests are measured separately from website activity and qualification records. All figures are illustrative and do not establish that either page caused the demo requests.

The product page receives fewer search clicks than the checklist, but it serves a more directly commercial need and is associated with four qualified demo requests in the illustrative period.

The checklist reaches a larger search audience and is associated with two qualified demo requests, but its primary job is to help HR professionals organize onboarding.

Both resources are relevant to Beacon’s objective, but they contribute in different ways. Bhavya compares the opportunities:

With capacity for two substantial projects, Bhavya prioritizes the two existing-page improvements.

Project 1: Improve Beacon’s employee onboarding software page.

The page serves an important commercial need and already attracts relevant search visibility. Improving its product information would address the weaknesses identified during the research and help prospective customers evaluate whether Beacon meets their requirements.

Project 2: Improve Beacon’s employee onboarding checklist.

The checklist already reaches a relevant audience, but it does not adequately serve several practical needs revealed by the research. Strengthening the resource and making it easier to adapt would better serve the existing keyword cluster without creating another substantially overlapping page.

Bhavya keeps the employee onboarding process guide as a potential future project. The search need is distinct and relevant, but the two existing-page improvements have stronger current evidence, align closely with Beacon’s objective, and fit the available capacity.

The research has now produced a focused set of priorities rather than an unfiltered list of potentially attractive keywords.

Step 7 output: Two prioritized projects supported by the available evidence, with other relevant opportunities retained for future evaluation.

Step 8: Turn your research into action and measure the results

Keyword research only becomes useful when it changes what your organization does.

Once you've selected the opportunities worth pursuing, turn them into a practical work plan. For each project, define what will change, why the work is justified, what outcome it should support, and how you will evaluate the result.

Define the action and intended outcome

For each selected opportunity, record the search need, relevant page, supporting evidence, planned action, and intended outcome.

The action should follow from your research, while the outcome should reflect the job of the resource. A commercial page might help prospective customers evaluate an offering and support qualified inquiries. An educational resource might better serve a specific need, reach the right audience, or help people take an appropriate next step.

Avoid assuming that every SEO project should increase both traffic and conversions.

Establish a baseline before making changes

Record what is happening before you implement the work, then choose measurements that reflect the opportunity and intended outcome:

For an existing page, record the relevant page and query performance over an appropriate period. For a new resource, document the starting conditions, intended search need, and the measures you will use to determine whether it begins serving that need.

Don't expect every metric to move together. More impressions do not necessarily mean more relevant visitors, and more clicks do not establish that more people completed a meaningful action.

Search Console measures performance in Google Search, while website analytics measures activity after people arrive. The two systems measure different parts of the journey, so their figures are not interchangeable. [2]

Evaluate what changed

After implementing the work, allow enough time for meaningful evidence to develop. Google notes that changes can take anywhere from a few hours to several months to be reflected in Search and generally recommends waiting a few weeks before assessing their effects. [11]

Compare subsequent performance with your baseline using comparable reporting periods, pages, and query groups. Then return to the outcome you intended to improve.

Did relevant visibility change? Are people finding and using the resource? Are they taking the appropriate next step? Has the business outcome you care about changed?

Interpret those results carefully. Search demand, competing resources, search-result features, other website changes, and external factors may all affect performance.

A change in performance after your work does not, by itself, prove that your work caused it. Record what changed, what remains uncertain, and what the evidence supports doing next.

Revisit your assumptions when the evidence changes

Keyword research is not a one-time exercise. Search behavior, competing resources, your website, and business priorities are always changing. A previously unattractive opportunity may become more worthwhile. A cluster may need to be reconsidered if the SERPs begin reflecting a different need. An existing page may no longer be the best resource for a search it once served.

When new evidence challenges an earlier decision, return to the fundamentals: What does the searcher need? Which resource should serve that need? Does the opportunity still support the business objective?

The answer may be to improve an existing page, create something new, change your priorities, investigate further, or leave the current approach alone.

Example: Bhavya turns Beacon's research into a plan

Bhavya turns Beacon’s two selected opportunities into a practical work plan.

Beacon’s content plan

Before implementing the work, Bhavya records the relevant baseline measurements and defines how each project will be evaluated.

She also keeps the employee onboarding process guide in the keyword map as a potential future project rather than commissioning it now.

The plan gives Beacon’s team a clear record of what will change, why the work was selected, what it is intended to accomplish, and how the result will be evaluated.

What happens after the changes?

After the updates have been published and enough time has passed for meaningful evidence to develop, Bhavya reviews a comparable subsequent 28-day period.

Her illustrative results are:

Both pages received more search clicks during the later period, while qualified demo requests associated with those landing pages remained unchanged.

Bhavya does not conclude that the updates caused the additional clicks or treat the traffic increase as evidence that Beacon achieved its business objective.

Instead, she asks whether the additional visitors are relevant, how they are using the improved resources, whether they are reaching appropriate next steps, and whether the available evidence justifies another change.

She records what changed, continues monitoring the two resources, and uses the evidence to decide what deserves investigation next rather than automatically commissioning another piece of content.

The keyword research has now done its job: it helped Beacon move from a broad set of possible searches to two justified investments, measurable outcomes, and a clearer basis for the next decision.

Step 8 output: An actionable work plan, a baseline and subsequent performance record, and a clear basis for deciding what to investigate or prioritize next.

Putting it all together: Bhavya's keyword research plan

Bhavya began with a broad collection of searches related to employee onboarding, a goal of increasing qualified demo requests from businesses discovering Beacon through organic search, and capacity for only two substantial projects.

By examining the needs behind the searches, the current SERPs, Beacon’s existing resources, and what each opportunity would require, she turned that keyword universe into a focused plan.

Beacon's completed keyword map

All findings and decisions in this example are illustrative.

The map preserves opportunities Beacon may reconsider as new evidence emerges rather than forcing every relevant search into a project.

This is the practical output of keyword research: an understanding of where relevant demand exists, what those searches represent, which resources should serve them, and where the business should invest.

Your keyword research template

Use Bhavya’s plan as a model for your own website. A spreadsheet or document is enough. What matters is that each decision can be traced back to the underlying search need, supporting evidence, and business objective.

Define your research scope

Start by recording the context for your research.

These fields establish the context you will use to evaluate the opportunities you discover.

Record and evaluate your keyword opportunities

For each meaningful search need or cluster, record:

Keep relevant URLs, reporting periods, geographic settings, and supporting observations with the research so each decision can be traced back to the evidence.

Do not combine unrelated search needs simply to reduce the number of entries or create separate projects for every keyword variation when one resource can adequately serve the shared need.

Before finalizing your plan

Check that each selected opportunity supports your business objective, that your clusters reflect shared search needs, and that proposed pages do not unnecessarily overlap with existing resources.

Confirm that the selected work is realistic given your available resources and that you understand the limitations of the keyword metrics and SERP observations informing the decision.

A completed keyword-research plan should make it clear what deserves attention, what action is justified, and why the remaining opportunities should wait, be excluded, or be investigated further.

Common keyword research mistakes

Prioritizing search volume over business value

High search volume does not automatically make a keyword valuable. Evaluate whether the underlying search need matters to your audience and objectives before treating estimated demand as an opportunity worth pursuing.

Treating every keyword as a separate page

Different queries can represent the same search need. Group related searches by need, then determine which resource should serve them rather than creating a page for every variation.

Clustering keywords by wording alone

Similar-looking queries may represent different needs, while different wording may point to the same one. Use search intent and SERP evidence rather than language similarity alone.

Trusting keyword difficulty without examining the results

Keyword-difficulty scores are estimates, not complete assessments of whether your website can compete. Use them as a signal, then evaluate the actual search environment and what pursuing the opportunity would require.

Copying competitor keywords without filtering

Competitors may rank for searches that have little relevance to your audience or objectives. Use their visibility to discover potential opportunities, then evaluate each one against your own business.

Ignoring existing website coverage

A new keyword opportunity does not necessarily require a new page. Check whether an existing resource already serves the need and could be improved instead.

Automatically rejecting low- or zero-volume keywords

A low or zero search-volume estimate does not prove that the underlying need has no value. If the need is relevant, decide whether to address it within an existing resource, create something distinct, or retain it for later consideration.

Relying on outdated keyword optimization advice

You don't need to force exact-match wording throughout a page or create separate content for every variation of how people might search. [12]

Avoid treating “LSI keywords” as a special list of terms that must be inserted into your content. Focus instead on serving the search need with accurate information, natural terminology, and an appropriate resource.

Continuing to optimize without reconsidering the opportunity

If a page does not perform as expected, do not assume it simply needs more content or more optimization. Revisit the search need, current SERP, intended audience, and role of the page before deciding what to change next.

Frequently asked questions

What is the difference between keyword research and keyword analysis?

Keyword research is the broader process of discovering, organizing, evaluating, and prioritizing search opportunities. Keyword analysis is the evaluation of individual queries or groups of queries using evidence such as search intent, estimated demand, existing visibility, competitive conditions, and business relevance.

Analysis is one part of the broader keyword-research process.

How do you choose the right keywords for your website?

Start with your business objectives and intended audience, then identify search needs your website can credibly serve.

Evaluate those opportunities against their business relevance, existing coverage, competitive requirements, and the investment needed to pursue them.

How do you decide whether two keywords belong on the same page?

Ask whether the searches represent substantially the same need and whether one resource can adequately serve both.

SERP overlap can support that decision, but it should not be treated as a universal rule.

Should you target keywords with low or zero search volume?

Low or zero reported volume should not automatically disqualify a keyword.

If the underlying need is real and relevant, determine whether it belongs within an existing resource, warrants something distinct, or should be retained for later consideration.

How many keywords should one page target?

There is no universal number. A page can be relevant to many queries that express the same or closely related search need.

Focus on the purpose of the page and the need it serves rather than an arbitrary keyword count.

How often should you conduct keyword research?

Revisit keyword research as your business, audience, website, or search environment changes.

Broader research may be useful when developing a new strategy, entering a market, launching a product, or reviewing your search program. More focused research can support individual opportunities or performance reviews.

Can you conduct keyword research without paid tools?

Yes. Business knowledge, customer feedback, Google Search Console, Google Keyword Planner, and manual SERP analysis can provide a strong foundation.

Paid tools can make discovery, analysis, and monitoring more efficient, but they do not replace understanding the search need, business opportunity, and resource required to serve it.

Final thoughts: Turn search demand into better content decisions

Keyword research helps you understand what your audience searches for, what those searches represent, and where your website has opportunities to serve them.

Finding relevant keywords is only the beginning. The real value comes from deciding which resources should serve those needs and where your organization should invest.

Some opportunities call for improving existing content, while others justify creating something new. The rest may be better investigated further, deferred, or excluded.

By connecting search demand with your business objectives, existing website coverage, and available resources, you can turn a broad keyword universe into a focused plan.

Done well, keyword research gives you a clear basis for deciding which opportunities deserve investment and what your website should do next.

Sources

• [1] SEO Starter Guide: The Basics | Google Search Central  |  Documentation  |  Google for Developers developers.google.com ↗
• [2] How Does Google Determine Ranking Results - Google Search google.com ↗
• [3] Use Keyword Planner - Google Ads Help support.google.com ↗
• [4] What are impressions, position, and clicks? - Search Console Help support.google.com ↗
• [5] Google Search's Guidance on Third-Party SEO Tools & Advice | Google Search Central  |  Documentation  |  Google for Developers developers.google.com ↗
• [6] Long-tail Keywords: What They Are and How to Get Search Traffic From Them — Ahrefs ahrefs.com ↗
• [7] Search Intent in SEO: What It Is & How to Optimize for It — Ahrefs ahrefs.com ↗
• [8] How Does Google Determine Ranking Results? — Google Search google.com ↗
• [9] Generative AI Performance Report (Search) — Google Search Console Help support.google.com ↗
• [10] How To Do Keyword Clustering the Easy Way — Ahrefs ahrefs.com ↗
• [11] SEO Starter Guide: The Basics — Google Search Central developers.google.com ↗
• [12] Optimizing Your Website for Generative AI Features on Google Search — Google Search Central developers.google.com ↗