You found a keyword research guide. It was published in 2023 or earlier. It told you to find high-volume keywords, check difficulty, and write content targeting them. Those guidelines for effective keyword research in 2023 were incomplete then. In 2026, following them without adjustment will point you at keywords that never produce traffic.
Most keyword research tutorials were written before AI Overviews rolled out to roughly 48% of tracked queries, before zero-click searches hit 68% of all Google searches in early 2026, and before ChatGPT reached 900 million weekly active users as a parallel discovery channel. The mechanics of keyword research have not changed. What changed is which keywords are worth targeting, which produce clicks versus impressions, and which you can realistically rank for with a low-authority domain.
The three pillars that define effective keyword research in 2026 are search volume, intent matching, and winnability. Search volume still matters because it tells you demand exists. But volume alone no longer predicts traffic. A keyword with 8,100 monthly searches and an AI Overview at the top of the SERP delivers a fraction of the clicks that same keyword delivered in 2023. Intent matching matters more now because AI Overviews satisfy informational queries directly on the results page. When Google can answer the question with a generated summary, your page competes against Google itself, not just other websites. Winnability is the variable most guides skip. Even if a keyword has volume and clear intent, can your domain actually rank for it? A site with a domain rating of 15 targeting a keyword with a difficulty score of 55 is not doing keyword research. It is doing wishful thinking.
Guidelines for Effective Keyword Research in 2023: What Still Applies
The 2023-era advice that still works: build seed lists from real customer language, expand with competitor gap analysis, target long-tail questions where top results are weak, and cluster keywords into topical groups. Those principles are sound and they predate AI Overviews by a decade. The 2023-era advice that is now dead: treating search volume as a traffic estimate, ignoring SERP feature overlap, and skipping the difficulty check against your own domain authority. Each of those assumptions breaks when an AI Overview sits above your result and absorbs the click.
Here is the test. Pull up any keyword research guide you bookmarked in 2023. Search for the words “AI Overview,” “zero-click,” or “winnability.” If none appear, the guide is missing the three forces that reshaped which keywords produce traffic in 2026. You do not need to throw away the framework. You need to add a scoring layer that accounts for them.
The 5-Step Keyword Research Framework That Still Works
This framework works for a solo operator with limited tools and a domain rating under 25. Each step includes what to do, what to skip, and how long it should take. The total time for a 20-keyword research sprint is 3 to 4 hours, which is less time than most people spend staring at a keyword tool dashboard without a process. For the broader content strategy this feeds into, our guide on creating a content strategy for SEO walks through the full publishing pipeline.
Step 1: Build a Seed List From Customer Language
Open your inbox, help desk, sales call notes, and review sites. Write down the exact phrases customers use to describe their problem. “Why does my AC smell like vinegar” is a better seed keyword than “HVAC services” because it reflects how real people search when they have a problem you can solve. Tool suggestions give you volume data for keywords people already rank for. Customer language gives you keywords nobody has optimized yet. Spend 30 minutes here. You should end up with 15 to 25 raw seed phrases.
Step 2: Expand With Competitor Gap Analysis and Question Mining
Take your seed list into a keyword tool. Run a competitor gap analysis against 2 to 3 competitors to find keywords they rank for that you do not. The Missing Keywords tab alone can fill a content calendar for months. For a full breakdown of what Semrush does well and where it hits its ceiling for WordPress users, see our Semrush Pro review.
Then mine People Also Ask boxes and Google Autocomplete for question-based variants. Type each seed keyword into Google, scroll to the PAA section, and click through 3 to 4 questions. Each one expands into 3 to 5 more. A single PAA seed query produces 30 to 50 question-based keywords in 10 minutes. These long-tail questions are where a low-authority domain can still compete, because the top-ranking pages for these queries are often forum posts and thin articles, not authoritative guides.
Spend 45 minutes on this step. You should end up with 60 to 100 expanded keywords from your seed list.
Step 3: Score Each Keyword on a Winnability Matrix
This is the step most keyword research guides skip, and it is the one that determines whether your content produces traffic or collects dust at position 47. Score each keyword on three factors, each worth one point.
Rankability. Can your domain rank for this keyword? Set a difficulty ceiling based on your domain rating. For a DR 8 site, target KD 0 to 12 for your first three months, rising to 0 to 18 by months four through six. Then check the domain ratings of the pages currently ranking in the top 5. If all five have DR 50 or higher, you cannot compete regardless of what the KD score says. If at least one has a DR within 15 points of yours, you have a shot.
Clickability. Will this keyword produce clicks? Search the keyword in Google and check whether an AI Overview appears at the top. If it does, that AI summary will capture most of the clicks. According to Ahrefs’ December 2025 study of 300,000 keywords, AI Overviews reduce the click-through rate for the top-ranking page by 58%. If an AI Overview is present, score this factor 0 for click traffic. You may still target the keyword for citation value, which we cover below.
Convertability. Will the clicks convert? Commercial and transactional intent keywords score 1. Informational keywords score 0 unless you can funnel readers to a conversion page through internal links.
Add the three factors for a winnability score from 0 to 3. A score of 0 means skip. A score of 1 means target only for citation value or save until your domain rating grows. A score of 2 means target now. A score of 3 means priority target. Here is a worked example for a local HVAC company with a domain rating of 8.
| Keyword | Monthly Vol. | KD | Top 5 DR | AI Overview | Intent | Winnability | Verdict |
|---|---|---|---|---|---|---|---|
| hvac repair | 8,100 | 67 | 50 to 85 | Yes | Commercial | 0/3 | Skip |
| hvac repair near me | 4,400 | 45 | 30 to 60 | No | Transactional | 0/3 | Skip |
| why does my ac smell like vinegar | 1,300 | 11 | 12 to 28 | No | Informational | 2/3 | Target |
| how often should i replace hvac filter | 720 | 9 | 10 to 25 | Yes | Informational | 1/3 | Citation play |
| hvac maintenance checklist for homeowners | 480 | 14 | 15 to 30 | No | Informational | 2/3 | Target |
| best smart thermostat for two stage furnace | 260 | 17 | 18 to 35 | Yes | Commercial | 1/3 | Save for month 4 |
Notice the pattern. The two keywords with 8,100 and 4,400 monthly searches are unwinnable for a DR 8 domain. The keywords you can win have KD under 15, top-ranking pages with domain ratings close to yours, and long-tail phrasing that signals specific intent. You should end up with 20 to 30 winnable keywords from a starting list of 100. That is more than enough for your first 90 days of publishing.
Spend 90 minutes on this step. It is the most time-consuming part of the framework, but it is also the part that prevents you from wasting months publishing content for keywords you can never rank for.
Step 4: Group Keywords by Intent and Funnel Stage
Sort your scored keywords into four buckets. Informational keywords are top of funnel: the searcher is learning, not buying. Commercial keywords are mid-funnel: the searcher is comparing options. Transactional keywords are bottom of funnel: the searcher is ready to act. Navigational keywords are brand searches: someone is looking for you specifically. Each bucket maps to a different content type, and mismatching them is the second most common reason content fails to rank.
Spend 20 minutes on this step. You should end up with 4 to 5 clusters of 4 to 6 keywords each.
Step 5: Map Each Cluster to a Content Type and Publishing Cadence
Informational keywords get spoke posts: focused 800 to 1,200-word articles that answer one question in the first paragraph, then elaborate with causes, fixes, and prevention. Commercial keywords get comparison posts with tables, specific criteria, and a recommendation. Transactional keywords get service or product pages. Navigational keywords get your homepage and About page.
Publish 2 to 3 spoke posts per week and one comparison post per cluster per month. Update each pillar page quarterly. Total time for this step: 15 minutes to map, then the publishing cadence runs for 90 days.
What Changed Since 2023: AI Overviews and Answer Engine Visibility
The biggest shift since 2023 is not a new ranking factor or a Google algorithm update. It is a structural change in how search results deliver information. AI Overviews now appear on roughly 48% of tracked queries, up from 31% a year earlier. When an AI Overview is present, it reduces the click-through rate for the top-ranking page by 58%, according to Ahrefs’ December 2025 analysis of 300,000 keywords. Zero-click searches reached 68% of all Google searches in the first four months of 2026, up from 60% in 2024, per SparkToro’s analysis of Similarweb clickstream data.
The practical effect for keyword research: keywords that used to deliver clicks now deliver impressions that go nowhere. A keyword with 2,400 monthly searches and an AI Overview at position 0 will send you a fraction of the traffic that same keyword sent in 2023. Pew Research found that when an AI summary appears, only 8% of users click a traditional result link, compared to 15% when no AI summary is present.
Commercial Intent Keywords That AI Overviews Do Not Satisfy
Not every query triggers an AI Overview. AI Overviews appear most frequently for informational queries: definitions, how-tos, explanations. They appear less often for commercial comparison queries and transactional queries where the searcher wants to see pricing, reviews, or a buy button. This means commercial intent keywords that AI Overviews do not satisfy are now more valuable than they were in 2023, because the competition for clicks has thinned out at the informational level while remaining intact at the commercial level.
Research suggests that AI Overviews on commercial-intent results pages have grown significantly through 2025 and into 2026, but have not reached saturation. There are still commercial queries where the SERP is dominated by comparison articles, review sites, and product pages, not AI summaries. Those are your targets.
The Informational Keyword Tradeoff
Informational keywords are lower competition, which makes them attractive for a low-authority domain. They are also increasingly zero-click. The honest tradeoff: targeting informational keywords is still worth it for building topical authority and earning internal linking opportunities, but you should not expect them to drive significant click traffic if an AI Overview appears for the query.
The exception is citation value. If you can earn a citation inside the AI Overview itself, you capture a small but qualified stream of clicks. According to Semrush’s AI SEO data, being featured as an AI Overview source increases CTR from 0.6% to 1.08%. That is not a lot, but it is not zero, and the visitors who do click are pre-qualified by the AI summary.
Answer Engines as a Parallel Discovery Channel
ChatGPT, Perplexity, and Google Gemini are not search engines, but they function as discovery channels. When a user asks ChatGPT for a recommendation or Perplexity for an explanation, these engines cite sources. Studies of AI answer engine outputs suggest that Perplexity typically cites more sources per answer than ChatGPT, giving well-cited pages a meaningful visibility advantage. Research suggests the overlap between pages cited by answer engines and pages ranking in Google’s top 10 is limited, which means ranking well in Google does not automatically get you cited by answer engines.
For keyword research, this means you should check whether your target keywords also surface in answer engine responses. If a keyword triggers a Perplexity or ChatGPT response that cites sources, earning that citation is a parallel traffic channel that does not depend on your Google ranking position.
Common Keyword Research Mistakes That Waste Your Budget
These are the four mistakes that waste the most time and money. Each one includes a diagnostic question to catch it and a correction to fix it.
Targeting Head Terms With 60+ KD on a DR 15 Domain
You found a keyword with 12,000 monthly searches. The difficulty score is 62. Your domain rating is 15. You publish a 2,000-word post targeting it. Six months later, you are at position 38 and the post has generated 4 clicks. The diagnostic question: is the KD score more than 20 points above your domain rating? If yes, you cannot rank for this keyword with your current authority. The correction: filter your keyword list to KD scores within your ceiling. For DR 15, that means KD 0 to 18 for your first six months.
Ignoring SERP Feature Overlap
You target a keyword without checking what the SERP actually looks like. The top of the page has an AI Overview, a People Also Ask box, a video carousel, and then the first organic result at position 4. You ranked at position 6, which means you are below four SERP features and five organic competitors. Your click share is near zero. The diagnostic question: what is above the first organic result for this keyword? If the answer is more than two SERP features, your effective position is much worse than your reported position. The correction: search the keyword in Google before you write. Count the SERP features above the first organic result. If there are three or more, either target a different keyword or structure your content to earn a featured snippet or AI Overview citation. For CTR optimization after you have ranked, see our guide on optimizing blog post titles with CTR data.
Skipping Intent Verification
You found a keyword with good volume and manageable difficulty. You wrote a 1,500-word tutorial. The top 5 results are all product pages. You ranked a blog post for a transactional query, and Google shows product pages above you because the intent is to buy, not to learn. The diagnostic question: did you Google the keyword and read the top 5 results before writing? If no, you are guessing at intent. The correction: for every keyword you plan to target, spend 5 minutes searching it in Google and noting the format of the top 5 results. If they are all tutorials, write a tutorial. If they are all comparison tables, build a comparison table. If they are all product pages, you need a product page, not a blog post.
Confusing Search Volume With Traffic Potential
You found a keyword with 8,100 monthly searches. You assumed that ranking at position 5 would send you roughly 800 clicks per month, based on the old CTR curve. But the SERP has an AI Overview, a featured snippet, and a People Also Ask box above position 1. Your position 5 ranking sends you 40 clicks in 30 days. The diagnostic question: have you subtracted zero-click rate and SERP feature displacement from the search volume to estimate actual traffic? If no, your traffic projection is based on 2023 assumptions that no longer hold. The correction: search volume is the ceiling, not the floor. Subtract the AI Overview click reduction (roughly 58% if present), subtract the SERP features above your position, and then apply the CTR curve to what remains. That is your realistic traffic estimate.
Start Publishing What You’ve Researched
Keyword research is 5% of the work. The other 95% is turning that research into published, SEO-optimized WordPress posts. You can have the best winnability matrix in the world, but if the posts do not get published, the research produces zero traffic. This is where most solo founders and small marketing teams stall: the research is done, the spreadsheet is built, and then the writing and publishing pipeline collapses under manual labor.
This is the gap we built ClearPost to close. ClearPost runs the full pipeline inside WordPress: keyword data from your Search Console, AI-generated drafts with SEO structure, automatic internal linking, meta tags pre-filled, and scheduled publishing through the WordPress REST API. The only manual step is the approval gate, where you read the draft and click approve. Nothing publishes without your review. That is a non-negotiable safety principle. Any tool that auto-publishes AI content without a human approval step is a liability, not a feature.
Try ClearPost free for 7 days. AI does the heavy lifting, you approve every post before it goes live. No long onboarding, no agency overhead, cancel anytime. See what 30 SEO-optimized posts a month looks like compared to the 4 you are getting now.
Frequently Asked Questions
Are 2023 keyword research strategies still effective in 2026?
The core mechanics are still valid: find keywords with real demand, check difficulty, and match content to intent. What changed is which keywords are worth targeting. Informational keywords that trigger AI Overviews now produce fewer clicks, so the 2023 advice to chase high-volume informational terms needs adjustment. Add winnability scoring to filter out keywords your domain cannot rank for, and check for AI Overview presence before committing to a keyword.
What is a winnability score and how do I calculate it?
A winnability score combines three factors: rankability (can your domain rank for this keyword based on difficulty and competitor authority), clickability (will the keyword produce clicks given AI Overview presence), and convertability (will the clicks convert based on intent type). Each factor scores 0 or 1, for a total from 0 to 3. Score 0 means skip, 1 means target only for citation value, 2 means target now, and 3 means priority target.
How do AI Overviews change which keywords I should target?
AI Overviews appear on roughly 48% of tracked queries and reduce the click-through rate for the top-ranking page by 58% when present. This makes informational keywords that trigger AI Overviews less valuable for click traffic, while commercial intent keywords that AI Overviews do not satisfy become more valuable. Check for AI Overview presence before targeting a keyword.
Should I target informational keywords that trigger AI Overviews?
Only if you can earn a citation inside the AI Overview itself. Being cited as a source increases CTR from 0.6% to 1.08%, which is small but not zero. If you cannot earn the citation, the keyword will generate impressions but few clicks. Target these keywords for topical authority and internal linking value, not for direct traffic.
What keyword difficulty should I target for a new website in 2026?
For a domain rating under 15, target KD 0 to 12 for your first three months. Raise the ceiling to 0 to 18 by months four through six as your domain rating climbs. Always verify by checking the domain ratings of the pages currently ranking in the top 5. If all five have DR 50 or higher, skip the keyword regardless of its KD score.