How to Choose Keywords for SEO: Expert Selection Framework

Why Keyword Selection Matters More Than Ever

If you picked keywords in 2020 the way you pick them now, you would need to rethink your entire approach. The search landscape has shifted under everyone's feet, and three forces are driving the change.

First, AI Overviews are absorbing clicks. Ahrefs tracked 300,000 keywords and found that when Google shows an AI Overview, the top organic result loses up to 58% of its clicks. For informational queries, position-1 CTR collapsed from 7.3% to 1.6% between December 2023 and December 2025. That is not a minor dip. That is a category of keywords that stopped paying rent.

Second, zero-click searches keep climbing. SparkToro data shows that for every 1,000 US Google searches, only 360 clicks reach the open web. On mobile, 77.2% of searches end without a click at all. If you target keywords where Google can answer the question inside the SERP, you are building on a shrinking foundation.

Third, ranking takes longer and requires more authority. The average page in position one is now five years old. Only 0.3% of new pages rank in the top 10 for high-volume keywords within a year. Topical authority is no longer optional. Kevin Indig's research found that pages with strong topical authority gain traffic 57% faster than pages without it.

These numbers tell a clear story: the margin for error in keyword selection is smaller than it has ever been. Every keyword you target is a bet on your time, budget, and content resources. This framework exists to make those bets smarter.

Step 1: Match Search Intent Before You Touch a Tool

The single most common reason content fails to rank is not poor writing or weak backlinks. It is targeting the wrong intent. Eric Siu put it directly: the most common reason for content failure is creating the right content for the wrong intent.

Search intent is what the person typing a query actually wants to accomplish. Google categorizes this into four buckets in its Search Quality Evaluator Guidelines: Know (informational), Do (transactional), Website (navigational), and Visit-in-person (local). SEO tools typically use four labels: informational, navigational, commercial investigation, and transactional.

The split is lopsided. Over 80% of all queries are informational. Roughly 10% are navigational, and about 10% are transactional. But that 10% transactional slice generates most of the revenue for commercial websites, which is why understanding conversion-ready intent matters more than chasing volume.

The 3 Cs Framework for Intent Diagnosis

The most practical method for diagnosing intent comes from Ahrefs. For any keyword you are evaluating, open an incognito browser and scan the top 10 results for three things:

• Content Type: Are the top results blog posts, product pages, category pages, or tools? If Google ranks product pages and you plan to publish a blog post, you will not rank. Period.
• Content Format: Are the ranking pages how-to guides, listicles, comparisons, reviews, or tutorials? Match the dominant format.
• Content Angle: What hook do the top results use? "For beginners," "in 2026," "free," "step by step." The angle tells you what Google's users respond to.

Ahrefs has a concrete example from their own site. Their "backlink checker" page sat below position 8 for months when it was structured as a marketing landing page. Once they rebuilt it as an actual free tool, it climbed to position 1 and monthly traffic grew from around 150,000 to over 600,000 visits. The content quality did not change. The intent match did.

Fractured Intent and Mixed SERPs

Not every keyword has a clean, single intent. Some queries produce mixed SERPs where Google shows a combination of blog posts, product pages, videos, and knowledge panels. This is called fractured intent.

When a SERP is fractured, you have a decision to make. You can target the dominant interpretation (the content type that occupies the most positions in the top 10) or find a wedge in a secondary interpretation that is underserved. Ahrefs' Keywords Explorer now includes an intent breakdown showing what percentage of users want each type of result.

The practical takeaway: spend five minutes analyzing the actual SERP before committing resources to any keyword. Check what types of content rank, note which SERP features appear (featured snippets, People Also Ask, AI Overviews, video carousels), and confirm that your planned content type matches what Google rewards for that query.

Step 2: Evaluate Keyword Difficulty (Without Getting Fooled)

Keyword difficulty scores are the most misunderstood numbers in SEO. Every major tool calculates them differently, the scores are not comparable across tools, and they all miss factors that determine whether you can actually rank.

How KD Scores Work Across Tools

Here is the problem in real numbers: the keyword "bike tire pump" scores 13 in Ahrefs, 39 in KWFinder, 45 in Moz, and 56 in Semrush. Same keyword, four different difficulty assessments. Pick one tool and stay with it. Cross-tool comparisons will waste your time.

Personal Keyword Difficulty: The 2025 Breakthrough

Both Ahrefs and Semrush launched personal keyword difficulty features in 2025, and this changes the math for keyword targeting significantly.

Ahrefs' Personal KD formula weights four factors: 40% base KD, 30% topical authority, 20% Domain Rating, and 10% URL Rating. Tim Soulo's analogy is useful here. Think of the KD score like the speed limit on a highway. The sign says 70 MPH, but your personal difficulty depends on the car you are driving. A new site with DR 15 and no topical footprint has a very different "personal difficulty" than an established site with DR 60 and 200 published pages in the same topic area.

If your tool supports Personal KD, use it. If not, here is a manual workaround: look at the keywords you already rank in the top 20 for in Google Search Console. Note their KD scores. That range is your baseline. Target keywords 10-15 points below that baseline for quick wins, at baseline for stretch targets, and 10-25 points above for aspirational content that requires deliberate link building.

Topical Authority vs. Domain Authority

This is the biggest ongoing debate in SEO. Which matters more for ranking: your site's overall domain strength or your depth of coverage on a specific topic?

Kevin Indig's position is clear. He wrote in Growth Memo that topical authority matters more now than ever, alongside brand authority. His research found that sites with deep topical coverage gain traffic 57% faster. His measurement method: export matching keywords for a head term, re-upload to identify traffic share by domain. That share equals your topical authority. For the topic "ecommerce," Shopify held 11%, BigCommerce 10%, and NerdWallet 3%.

Patrick Stox at Ahrefs offers the counterpoint. Google still relies on link-based signals like PageRank, and LLMs are increasingly using link graph data too. Assessing link authority is arguably only growing in importance.

The practical resolution is that both matter, but in different ways. Domain authority through backlinks gets you into the conversation. Topical authority from deep, interconnected content coverage is what keeps you there. If you are deciding between a high-KD keyword where you have strong topical authority and a lower-KD keyword in a topic you have never written about, the high-KD keyword in your wheelhouse is often the better bet.

Step 3: Run a Keyword Gap Analysis

A keyword gap analysis compares your organic keyword profile against your competitors' to find terms they rank for but you do not. It is one of the fastest ways to identify proven demand you are missing.

There are three distinct types of gaps to look for. A keyword gap is a specific query your competitors rank for that you do not have content targeting. A content gap is broader: missing topics or themes across your customer journey. And a SERP gap exists when competitors appear in SERP features (featured snippets, People Also Ask boxes, image packs) that your content does not.

The Practical Gap Workflow

Here is the step-by-step process using Ahrefs, adapted from their 2025 guide:

1. Open Ahrefs Competitive Analysis and enter your domain plus one to three competitors.
2. Click "Show keyword opportunities" and toggle "main positions only" to exclude SERP features you are unlikely to win.
3. Apply filters: exclude competitor brand names (partial match), exclude irrelevant topics, require at least two competitors ranking in the top 10.
4. Set minimum volume to 20 per month and KD under 30 to surface quick wins first.
5. Separate domain-level gaps (topics you lack entirely) from page-level gaps (both of you have a page, but your competitor ranks for more keywords on that page).
6. Use the site: operator to check whether you already have content that could target these keywords before creating new pages.

One important caveat from Ahrefs' own team: just because a keyword appears in the gap report does not mean it is a good fit for new content. Ahrefs does not know your business context. It only knows which sites rank for similar keywords. You still need to filter through business potential and intent alignment before committing.

If you use Semrush instead, their Keyword Gap tool categorizes results into six buckets: Shared (both rank), Missing (all competitors rank, you do not), Weak (you rank lower), Strong (you rank higher), Untapped (at least one competitor ranks, you do not), and Unique (only you rank). Prioritize in this order: Missing, then Weak, then Untapped. Semrush supports up to five domains per comparison and includes paid/PLA data that Ahrefs does not.

Identifying the right competitors matters here. Search competitors are not always your business competitors. Run your target keywords through Ahrefs' Organic Competitors report to find sites that overlap with you in search results, regardless of whether they compete for the same customers. A content strategy that only benchmarks against direct business competitors will miss the publishers and resource sites that dominate informational SERPs.

Step 4: Score and Prioritize Keywords

At this point in the process you have a list of keyword candidates that pass intent matching, difficulty evaluation, and gap analysis. The list is probably too long to act on all at once. You need a scoring system to decide what gets built first, what gets queued, and what gets cut.

There are several frameworks worth knowing. The right one depends on your team size and decision-making style.

Business Potential Scoring (Ahrefs)

Tim Soulo's Business Potential score is the simplest and often the most effective filter. Rate every keyword on a 0-3 scale based on how naturally your product or service fits into the content:

• 3: Your product is an irreplaceable solution to the searcher's problem.
• 2: Your product helps significantly, but alternatives exist.
• 1: You can mention your product, but it is not the main focus.
• 0: There is no natural way to connect this keyword to what you sell.

Soulo's principle is worth internalizing: keyword research is not the process of finding "easy to rank for" keywords. It is the process of finding the keywords that make the most sense to your business. Ahrefs reports that 77% of their own blog posts score a 2 or 3 on this scale. The ones scoring 0 or 1 are informational plays for brand awareness, not conversion-focused content.

The THRICE Framework (Eli Schwartz, 2026)

Eli Schwartz published THRICE in April 2026 as an extension of the RICE framework adapted specifically for SEO prioritization. It scores six factors on a 1-10 scale, then sums the totals:

Schwartz's reasoning for Reach over search volume is worth noting: keyword research data is notoriously inaccurate. TAM tells you how many people actually have the problem described by that phrase. Schwartz's own assessment is blunt. Most SEO teams read this framework and think they already prioritize well. They don't. They prioritize loudly, where everything is "mission-critical" and then nothing gets done.

KOB Analysis (Siege Media)

Ross Hudgens' Keyword Opposition to Benefit framework takes a more quantitative approach. The classic formula is: KOB = (Traffic Value) / Keyword Difficulty, where Traffic Value equals estimated monthly traffic multiplied by the keyword's CPC. Keywords with high traffic value and low difficulty bubble to the top.

The insight behind KOB is that you are not just scoring keywords. You are scoring the return on effort. A keyword with 500 monthly searches, $8 CPC, and KD 15 is a better investment than a keyword with 10,000 monthly searches, $0.50 CPC, and KD 75. Siege Media typically analyzes 100-200 topics per vertical in their month-one KOB sheets, scoring and sorting 150+ keywords before a single piece of content gets planned.

Grow & Convert adds a complementary philosophy: prioritize bottom-of-funnel keywords first, then move up. Their data shows BOFU pages convert at 0.3-4.3% while top-of-funnel content converts at 0.03-0.19%. Their question is direct: why would you produce top of funnel content before owning all of the keywords with product buying intent?

Step 5: Build Keyword Clusters, Not Keyword Lists

Individual keywords do not win rankings anymore. Topic clusters do. The question is no longer "which keyword should this page target?" but "which cluster of related queries can this page comprehensively serve?"

Ahrefs data makes this tangible: the average page ranking in position one also ranks in the top 10 for roughly 1,000 other keywords. The median is around 400. You are not targeting one keyword per page. You are building a resource that answers an entire cluster of related questions.

The practical method for clustering keywords by SERP overlap is straightforward. For smaller lists (under 20 keywords), Google each keyword and note which URLs appear in multiple top 10 results. If three or more keywords share the same ranking URLs, they belong on one page. For larger lists, tools like Keyword Insights and Ahrefs' parent topic clustering automate this with a default 30% URL overlap threshold.

The topic cluster model has been around since HubSpot introduced it in 2017, but it works differently now. A pillar page (typically 2,000+ words covering a broad topic) links to 10-15 cluster pages covering specific subtopics. Each cluster page links back to the pillar and to related cluster pages. Crawl data from 84 ecommerce sites showed that pages one click from the homepage get 2.3 crawls per day compared to 0.4 for pages five clicks away. Internal linking structure directly affects how often Google discovers and re-evaluates your content.

Kevin Indig's extension of this model is the "Keyword Universe": instead of periodic keyword research sprints, maintain a continuously updated keyword database organized by four research streams refreshed on different cadences. Audience research quarterly, product-tied keywords per launch, competitor analysis once (then updated on changes), and location-based terms per new market entry.

Advanced Tactics for 2026

Featured Snippet Targeting

Featured snippets still drive traffic, and the data on optimal formatting is specific. Paragraph snippets should land between 40 and 50 words (roughly 300 characters). List snippets with 8 or more items trigger a "More items" link that drives additional clicks. Tables are underused and consistently high-leverage. All featured snippet URLs already rank in the top 10, meaning Google picks the best answer from existing top results, not the strongest backlink profile.

Cyrus Shepard tested opting out of featured snippets with 95% statistical confidence. The result: a 12% traffic drop. If you hold a snippet, protect it.

AI Overview Strategy

AI Overviews now appear on roughly 15-25% of queries depending on the month, with prevalence peaking in certain verticals. Education sees AIO on 83% of queries, healthcare on 88%, while ecommerce sits at just 18.5% (Google protects its ad revenue in transactional searches).

The data on AIO citation eligibility is striking. Surfer's study of 173,902 URLs found that pages ranking for "fan-out queries" (the sub-questions an LLM generates when processing a broad prompt) are 161% more likely to be cited in AI Overviews. Even more interesting: 68% of AIO-cited pages are not in the top 10 organic results. Citation in AI responses operates on a different set of signals than traditional ranking.

Patrick Stox's analysis of 55.8 million AIOs found that branded web mentions had a 0.664 correlation with AIO inclusion, the strongest single predictor. Branded anchor text came second at 0.527. The implication for keyword selection: when evaluating whether a keyword is worth targeting, check whether the SERP includes an AI Overview and whether your brand has the recognition to get cited. If the answer to both is no, the keyword's effective traffic potential is lower than the tool shows.

Google Search Console Mining

"Striking distance" keywords (positions 5-15) are the fastest path to traffic gains. Filter GSC Performance data for positions between 5 and 20, sort by impressions descending, and you have a ready-made list of keywords where small improvements (title tag rewrites, internal links, a few external backlinks) can push you into top-5 positions where the click-through rates are meaningful.

Reddit and Forum Keyword Discovery

Reddit has 116 million daily active visitors and 4.5 billion monthly visits, making it the seventh most visited website globally. After Google's UGC algorithm changes, Reddit traffic grew 6x year-over-year. Running subreddits as domains in Ahrefs or Semrush reveals the questions people actually ask, sorted by organic traffic. This is keyword research based on real demand, not tool estimates.

Common Keyword Selection Mistakes to Avoid

Letting search volume drive every decision. Tim Soulo demonstrated this directly: one page got 5x more traffic than another despite identical search volume because it ranked for 406 related keywords versus 55. Always check traffic potential (the estimated traffic to the number one ranking page), not raw search volume.

Ignoring intent. If the top 10 is all product pages and you publish a blog post, you will not rank. We covered this in Step 1, but it bears repeating because it accounts for more failed content investments than any other mistake.

Chasing vanity volume. If you sell luxury hotel stays, targeting "cheap hotels" might drive traffic, but none of it will convert. Rand Fishkin's 2025 position is direct: site traffic is a vanity metric.

Targeting too-competitive keywords with a new site. Soulo advises that if your site is new, you should not waste time on keywords with 10,000+ monthly searches because they are almost certainly too competitive. Build authority with lower-difficulty keywords first, then graduate to harder terms as your domain strengthens.

Creating keyword cannibalization. When two pages target the same keyword, they compete against each other in Google's results. Backlinko documented a case where consolidating two cannibalized articles via a 301 redirect produced a 466% increase in clicks year-over-year. Before creating new content, check whether you already have a page targeting that keyword using the site: operator.

Skipping SERP analysis before committing. Difficulty scores, search volume, and even intent labels from tools are estimates. The SERP is the ground truth. Five minutes of manual SERP review will save you from investing weeks in content that never had a realistic chance of ranking.

Dismissing zero-volume keywords. About 15% of all daily Google searches have never been searched before and will not appear in any tool. These queries often have strong commercial intent and virtually no competition. The flip side: validate quickly and cut what does not perform. After Google's Helpful Content updates, thin or low-performing content can dilute your site's overall quality signals.

Not accounting for SERP feature click theft. If an AI Overview, featured snippet, or knowledge panel answers the query directly, the organic results below it get fewer clicks than the volume number suggests. Factor this into your traffic projections, especially for informational queries where AI Overviews dominate.

Author

Michael Timi

Partner & Marketing Manager, eMac Media

Drives strategic partnerships and revenue growth through high-impact marketing initiatives, business development, and lead generation.

Editor

Princess Pitts

Director of Communications Strategy, eMac Media

Specializes in editorial strategy, content governance, and brand communications at scale.