How to Do Keyword Research: The 8-Step Process We Use on Every Campaign

Keyword research is the process of finding the search terms your audience uses, evaluating their traffic potential and competition, and mapping them to content you’ll create. It’s the foundation of every SEO and content strategy. Without it, you’re guessing which topics to write about and hoping Google notices. With it, you’re building content that meets documented demand. The process hasn’t changed in its fundamentals, but the tools, data sources, and considerations have shifted in 2026. AI search engines like ChatGPT and Perplexity now influence keyword strategy because your content needs to be cited by both traditional search and AI models. Semrush’s Keyword Magic Tool now accesses over 27.9 billion keywords across 142 databases (Semrush, 2026). Ahrefs maintains 28.7 billion keywords across 217 locations (Ahrefs, 2026). The data available to you today is deeper than it’s ever been.

“Keyword research isn’t a task you do once. It’s a system you run every quarter. The brands that win organic traffic treat keyword research like financial planning: review the data, adjust allocations, and measure the returns.”

Hardik Shah, Founder of ScaleGrowth.Digital

Seed keywords are the 10-20 broad terms that describe your business, products, or services. They’re the starting inputs you’ll plug into keyword research tools to generate hundreds of more specific variations. Getting seeds right determines whether your final keyword list is relevant or full of noise.

A seed keyword is a broad, 1-3 word term representing a core topic in your industry that you expand into a larger keyword list using research tools.

There are 5 reliable seed keyword sources: 1. Your own product or service categories. What do you sell? What categories appear in your navigation? If you’re a SaaS company selling project management software, your seeds include “project management,” “task management,” “team collaboration,” “workflow automation.” List every category without filtering. 2. Customer language. Read support tickets, sales call transcripts, and chat logs. Customers describe your product differently than you do. A CRM company calls it “customer relationship management.” Their customers search for “track leads,” “sales pipeline tool,” or “keep track of clients.” Customer language reveals long-tail keywords your internal team would miss. 3. Competitor websites. Visit 3-5 competitor homepages and blog sections. Note the terms they use in navigation, H1 headings, and page titles. Plug their domains into Ahrefs’ Site Explorer or Semrush’s Organic Research to see what they rank for. Their ranking keywords become your candidate seeds. 4. Google’s own suggestions. Type each broad term into Google and note the autocomplete suggestions, “People Also Ask” questions, and “Related searches” at the bottom of the SERP. These are real search patterns. Google suggests them because actual people search for them. 5. Industry forums and communities. Reddit, Quora, industry Slack groups, and LinkedIn discussions reveal how practitioners talk about your space. The language used in discussions is closer to how people search than the language used on corporate websites. Aim for 15-25 seed keywords. Too few (under 10) and you’ll miss keyword clusters. Too many (over 40) and you’ll spend too long in the expansion phase without adding meaningful coverage. Our keyword research template has a dedicated tab for organizing your seed list.

Take each seed keyword and run it through your primary research tool. For each seed, you’ll generate 50-500 keyword variations depending on the topic breadth. The goal is volume first, filtering later. In Semrush Keyword Magic Tool: Enter your seed, select your target country, and review the results. Semrush groups results into subtopic clusters automatically. For the seed “email marketing,” you’ll see clusters like “email marketing tools,” “email marketing strategy,” “email marketing templates,” and “email marketing examples.” Export the full list including metrics. In Ahrefs Keywords Explorer: Enter your seed, choose “Matching terms” for exact variations and “Related terms” for semantic connections. Ahrefs shows keyword difficulty (KD), search volume, and traffic potential. The “Questions” filter surfaces question-based keywords ideal for FAQ content and featured snippets. Free method with Google Keyword Planner: Enter 3-5 seeds at once, filter by country, and download the results. Google Keyword Planner shows volume ranges (not exact numbers) unless you’re running ads. The “Refine keywords” panel on the left lets you expand into related categories you might not have considered. For each seed keyword, you should collect:

• All phrase-match variations (keywords containing your seed)
• Question-based variations (“how to,” “what is,” “why does”)
• Comparison variations (“[seed] vs [alternative]”)
• Long-tail variations (4+ word phrases with lower volume but higher intent)
• Modifier variations (“best,” “free,” “for beginners,” year-based terms)

After running all seeds, you’ll have a raw list of 500-5,000 keywords. This is normal. The next steps filter and prioritize. Don’t start filtering yet. A complete raw list prevents the common mistake of researching too narrow a keyword space.

Keyword difficulty (KD) estimates how hard it will be to rank in the top 10 for a given keyword. Every major SEO tool calculates KD differently, which is why the same keyword might show KD 45 in Ahrefs and KD 72 in Semrush. The number itself is less important than the relative ranking within each tool’s scale.

Keyword difficulty is a score (typically 0-100) estimating the competitive strength of the current top-ranking pages for a keyword, primarily based on their backlink profiles and domain authority.

How Ahrefs calculates KD: Primarily based on the number of referring domains linking to the top 10 results. A KD of 50 in Ahrefs means you’d need approximately 50 referring domains to have a chance at the top 10. It’s a backlink-centric metric. How Semrush calculates KD: Factors in referring domains, page authority, content quality signals, and SERP features. Semrush’s KD tends to run higher than Ahrefs for the same keywords because it weighs more factors. Beyond the score: manual SERP analysis. The most accurate difficulty assessment comes from looking at the actual SERP. For your target keyword, check:

• Who ranks? If positions 1-5 are held by Wikipedia, government sites, and major publications (DA 80+), it’s harder than the KD score suggests
• What’s the content quality? If the top results are thin or outdated, you can outrank them even with lower authority
• How many backlinks do the top results have? If position 1 has 300 referring domains and position 5 has 15, there’s a breakable entry point
• Are there featured snippets? Featured snippets can elevate lower-authority pages to position 0. If the current snippet answer is weak, that’s your opportunity

Practical KD guidelines for different site authorities:

Your Domain Rating	Target KD Range	Timeline to Page 1
DR 0-20 (new sites)	KD 0-20	3-6 months
DR 20-40 (growing sites)	KD 0-35	2-4 months
DR 40-60 (established sites)	KD 0-50	1-3 months
DR 60+ (authority sites)	KD 0-70+	1-2 months

These are guidelines, not guarantees. Content quality, topical authority, and internal linking all influence how quickly you rank. But targeting keywords outside your realistic KD range wastes time and budget.

Keyword clustering groups related keywords that should be targeted by a single page. Without clustering, you risk creating multiple pages targeting the same keyword (cannibalization) or spreading one topic across too many thin pages. A well-clustered keyword strategy means fewer, stronger pages instead of more, weaker ones.

Keyword clustering is the process of grouping related keywords that share the same search intent and SERP overlap into a single content target, so one page captures traffic from multiple search queries.

The SERP overlap method (most reliable): Two keywords belong in the same cluster if 3+ of the same URLs appear in both SERPs. Search for “email marketing tools” and “best email marketing software.” If the same 4 articles appear in both results, Google considers these the same topic. One page should target both. If the SERPs are completely different, they’re separate clusters requiring separate pages. The semantic method (faster): Group keywords that describe the same concept from different angles. “Email marketing tools,” “email marketing platforms,” “email campaign software,” and “email automation tools” are semantically related and likely share SERP overlap. This method works for initial grouping but should be verified with SERP analysis for your top-priority clusters. Cluster structure:

• Primary keyword: The highest-volume term in the cluster. This goes in your title tag and H1
• Secondary keywords: 3-8 related terms to include naturally in headings and body content
• Questions: “How to,” “what is,” “why” variations for FAQ sections and H2 headings
• Long-tail variants: 4+ word phrases for paragraph-level targeting

We consolidated 43 planned articles down to 16 pages for a B2B SaaS client using SERP overlap clustering. Those 16 pages captured the same keyword space with higher authority per page because content and backlinks weren’t diluted across redundant pages. The result: 340% more organic traffic than their previous unstructured approach. Tools for automated clustering: Semrush’s Keyword Strategy Builder creates clusters automatically. Ahrefs’ Traffic Share by Pages report shows which of your pages compete for the same keywords. For a manual approach, our keyword research template includes a clustering worksheet.

After running keyword research on 40+ campaigns, these are the patterns that consistently produce better results. 1. Research your competitors first, not your seeds. Plug 3-5 competitor domains into Ahrefs or Semrush before you start seed brainstorming. Their ranking keywords reveal keyword opportunities you’d never have brainstormed on your own. We discovered 30-40% of our best-performing keyword targets by starting with competitor analysis rather than internal brainstorming. 2. Use Google Search Console for existing sites. GSC shows you the actual queries your site appears for, including ones you didn’t target. Filter by impressions (high) and position (11-20). These are keywords where you’re almost on page 1. Small optimizations to existing content can push them up. This is the fastest ROI in SEO. 3. Don’t ignore zero-volume keywords. Keyword tools report “0” for many keywords that actually get searched. If a keyword is specific, has clear commercial intent, and matches your offering, target it even with reported zero volume. We’ve seen “zero volume” B2B keywords drive 20-30 highly qualified visits per month, which for enterprise sales, can be worth six figures annually. 4. Check AI visibility for every primary keyword. Search your primary keyword in ChatGPT, Perplexity, and Google AI Overview. If AI models already answer this query comprehensively, your content needs to be the source they cite, not a duplicate of what they say. Structure content with clear definitions, original data, and named sources. Our GEO guide covers the full framework. 5. Refresh your keyword research every 90 days. Search patterns shift. New competitors enter. Seasonal trends change. Quarterly keyword reviews ensure your content strategy stays aligned with actual demand. Add this to your content operations calendar, not your “someday” list.