Key Takeaways
- Filter for KD under 30 in Ahrefs, Semrush, or Ubersuggest, then manually review the top 10 results for weak competitors like forums, thin content, or outdated pages.
- Long-tail keywords (3+ words) convert at 2.5x the rate of broad head terms and account for over 70% of all web searches.
- KD scores lie. Use the SERP Weakness Score (a 5-point manual check) to evaluate how beatable the pages actually ranking are.
- AI tools are great for ideation, terrible for data. Use ChatGPT or Claude to brainstorm keyword angles, then validate volume and difficulty in a real SEO tool.
- Speed wins. Our 48-hour pipeline from keyword discovery to published page captures rankings before competitors finish their editorial calendar.
In This Article
- Keyword Research Tool Comparison (2026)
- Can AI Tools Find Low-Competition Keywords?
- Building AI-Powered Keyword Research Workflows
- The SERP Weakness Score: How to Evaluate Competition Beyond KD
- From Keyword to Published Page in 48 Hours
- Keyword Cannibalization: The Hidden Problem Nobody Talks About
- Downloadable Keyword Research Template
- Frequently Asked Questions
To find low-competition keywords, filter for terms with a keyword difficulty (KD) score under 30 in tools like Ahrefs, Semrush, or Ubersuggest, then validate by manually reviewing the top 10 search results for weak competitors such as forums, thin content, or outdated pages. The best low-competition keywords combine low KD with clear search intent and enough monthly volume (100-1,000 searches) to drive meaningful traffic. Long-tail variations of three or more words are where most of these opportunities live, since they account for over 70% of all web searches and convert at 2.5x the rate of broad head terms.
96.55%
of web pages get zero organic traffic from Google (Ahrefs)
~1,000
other keywords the average #1 page also ranks for in the top 10 (Ahrefs)
48%
of e-commerce keywords are low-difficulty; fewer than 2% are highly competitive (Semrush)
93:1
ratio of easy to difficult keyword opportunities in local services (Semrush)
Keyword Research Tool Comparison (2026)
KD Scores Are Not Universal
A keyword showing KD 25 in Ahrefs might show KD 45 in Semrush. Each tool uses a different formula. Ahrefs weights backlinks heavily. Semrush factors in 10+ parameters including authority scores and SERP features. Never compare raw KD numbers between platforms. Pick one tool and stay consistent.
Can AI Tools Find Low-Competition Keywords?
Short answer: sort of. But not the way you'd hope.
We've tested ChatGPT, Claude, and Semrush's AI features for keyword research across multiple client accounts. The results are a mixed bag that's worth understanding before you build a workflow around any of them.
What AI does well is ideation. Ask ChatGPT or Claude to brainstorm keyword angles for a niche and you'll get dozens of ideas you wouldn't have considered. They're particularly good at generating long-tail variations and question-based queries that mirror how real people search. We've pulled usable seed lists from AI conversations that would have taken an hour of manual brainstorming.
What AI can't do is give you reliable search volume or difficulty data. ChatGPT and Claude don't have access to live search volume databases. They'll sometimes hallucinate numbers if you push them, confidently telling you a keyword gets "approximately 2,400 monthly searches" with zero basis for that figure. Any metric they provide needs verification in an actual SEO tool.
Pro Tip: Semrush Personal Keyword Difficulty
Semrush's Copilot and Personal Keyword Difficulty (PKD) scoring sit on top of real data. PKD adjusts difficulty estimates based on your specific domain's authority. Instead of asking "is this keyword easy to rank for generally?" you're asking "is this keyword easy to rank for us?" That distinction matters when you're choosing between 50 keyword candidates.
The practical workflow we've landed on: use ChatGPT or Claude to generate a broad list of keyword angles and topic clusters. Export that list into Semrush or Ahrefs. Filter by KD and volume there. The AI handles creative expansion, the traditional tool handles validation.
Consistency Warning
AI recommendations are inconsistent. Research from 2025 showed there's less than a 1-in-100 chance that ChatGPT will produce the same list of recommendations if you ask it the identical question twice. That's fine for brainstorming. It's a problem if you're trying to build a repeatable process.
Building AI-Powered Keyword Research Workflows
Most keyword research guides tell you to type a seed keyword into Ubersuggest and scroll through suggestions. That works if you have one website and unlimited free time. It does not work if you manage 20+ client accounts and need to find low-competition opportunities across all of them every single week.
We built a different system. At Jetfuel, we use Claude Code (Anthropic's CLI tool) to create custom keyword research skills that automate the repetitive parts of the process. These aren't chatbot prompts. They're structured workflows that connect to real APIs and produce real output.
How We Do It at Jetfuel
One skill takes a seed keyword, hits the Google Ads API for search volume and CPC data, cross-references with Ahrefs or Semrush API for keyword difficulty scores, pulls People Also Ask questions via SerpAPI, and outputs a prioritized spreadsheet. The whole thing runs in under 60 seconds. Doing that manually across five seed keywords would take an hour.
Another skill monitors Google Search Console for "striking distance" keywords. These are queries where a client already ranks between positions 5 and 20. The skill pulls those keywords automatically, checks the current SERP landscape for each one, and generates a content brief that tells our writers exactly what to cover. No human has to dig through GSC filters to find these opportunities.
The full workflow looks like this:
Start With a Seed List
Client topics, competitor gaps, GSC queries.
AI Expands the List
Pulling related terms, questions, and long-tail variations.
Validate Against Real API Data
Each candidate gets checked for volume, CPC, and difficulty.
SERP Analysis on Top Candidates
Check what's actually ranking to confirm the opportunity is real.
Content Briefs Generated for Winners
Briefs are routed to the content team the same day.
The gap between "manual keyword research" and "automated keyword research" is the gap between finding one good keyword per hour and finding fifty. Most blogs won't tell you this because most blogs aren't running keyword research at agency scale. If you're trying to compete seriously, you need systems that move faster than manual browsing allows.
The SERP Weakness Score: How to Evaluate Competition Beyond KD
Keyword difficulty scores lie to you. Every tool calculates KD differently, and none of them actually look at the thing that matters most: how beatable are the pages currently ranking?
A keyword with a KD of 45 in Semrush might be trivially easy if the top results are all outdated forum posts from 2019. A keyword with a KD of 15 might be impossible if the first page is dominated by comprehensive guides from high-authority sites that were updated last month. The number alone tells you almost nothing.
We use a 5-point SERP Weakness Score to evaluate the real competition for any keyword. It takes about 60 seconds per keyword once you know what to look for.
Pro Tip: The 60-Second SERP Review
Open the top 5 results and check domain authority, word count, publish/update date, presence of schema markup, and backlink count. You're looking for signals of weakness. Old dates, thin pages, missing structured data, and low authority sites are all green lights.
We've seen this play out repeatedly. Keywords that tools scored as "hard" turned out to be easy wins because the actual ranking pages were outdated forum threads that nobody had bothered to compete against. The tool saw backlinks to the domain and assigned a high KD. The reality on the SERP told a completely different story.
Stop trusting the number. Start reading the SERP.
From Keyword to Published Page in 48 Hours: Our Workflow
Speed matters in SEO more than most people realize. When existing content on a topic is stale and search engines are hungry for fresh perspective, the first site to publish a comprehensive, up-to-date piece often captures the ranking. Waiting two weeks to go through a traditional editorial calendar means someone else gets there first.
We compress the entire process from keyword discovery to published page into 48 hours. Not every piece goes through this fast track, but when we spot a striking-distance keyword with weak competition, we move.
DAY 1, AM
Surface Striking-Distance Keywords
Our AI agent pulls the latest GSC data and surfaces 5 striking-distance keywords (positions 5-20) where the client has existing topical authority but no dedicated page. Each keyword comes with current ranking URL, impressions, click-through rate, and a SERP weakness score.
DAY 1, PM
Pick Winners and Generate Briefs
The team reviews the 5 candidates and picks the top 2 based on business relevance and competition level. For each winner, we generate a content brief using AI. The brief includes target word count, required subtopics, People Also Ask questions, internal linking opportunities, and schema recommendations.
DAY 2, AM
Write the Draft
A writer produces the draft using a structured writing workflow that ensures the piece matches the client's voice and is optimized for both traditional search and AI overview inclusion. The draft covers every subtopic from the brief and includes original perspective.
DAY 2, PM
Review, Polish, Publish
Editorial review, final polish, and publish. Once live, submit through IndexNow for near-instant Bing indexing and request indexing through Google Search Console.
WEEK 2
Monitor and Optimize
Monitor GSC daily for position changes. If the page enters the top 10, look for quick optimization wins (title tag tweaks, FAQ section, internal links). If it stalls, assess whether it needs additional depth or backlink support.
Key Insight: Why Speed Compounds
Google and AI models both reward fresh, comprehensive content on topics where existing results are stale. A 48-hour turnaround means we're publishing while competitors are still putting the keyword on next month's content calendar. Early ranking signals feed more impressions, which feed more clicks, which feed stronger rankings.
Keyword Cannibalization: The Hidden Problem Nobody Talks About
You published a great blog post targeting "email marketing for ecommerce." Six months later, you wrote another post covering "best ecommerce email strategies." Then your services page also targets "ecommerce email marketing." Now Google doesn't know which page to rank, so it rotates between all three and none of them rank well.
This is keyword cannibalization, and we see it in a significant percentage of client SEO audits. It's one of the most common reasons a site with good content still underperforms in search.
The diagnosis is simple. Go to Google Search Console, filter by a query you care about, and look at the Pages tab. If multiple URLs are getting impressions for the same query, you have cannibalization. The telltale sign is two or more pages splitting clicks, with neither one ranking as high as it should.
The fix depends on the situation:
Do Not Wait
Every day that two pages compete against each other is a day neither one ranks at its potential. If you haven't audited your site for cannibalization in the last 6 months, do it this week. The fix is usually straightforward, and the ranking improvements can be significant.
Downloadable Keyword Research Template
We've put together a Google Sheets template with 4 tabs that mirrors the exact workflow we use internally. Here's what each tab includes:
Pro Tip: Priority Score Formula
=IF(KD<30, (Volume/100) * SERP_Weakness * IF(Intent="Transactional",2, IF(Intent="Commercial",1.5,1)), 0)
This auto-filters out anything above KD 30 and weights transactional intent keywords 2x higher. Conditional formatting colors KD: green (<30), yellow (30-50), red (>50).
Frequently Asked Questions
What KD score is considered "low competition"?
Most SEO practitioners treat a KD score under 30 as low competition, meaning a newer or lower-authority site has a realistic chance of ranking on page one. Scores between 30 and 50 are medium difficulty, and anything above 50 typically requires significant domain authority and backlinks. But KD alone doesn't tell the full story. A keyword with KD 20 where the top results are all major publications is harder than a KD 35 keyword where the top results are thin forum posts. Always check the actual SERP alongside the score.
How many keywords should you target per page?
One primary keyword and two to five closely related secondary keywords per page. An Ahrefs study of 3 million searches found that the average top-ranking page also ranks for nearly 1,000 other keywords, but those are semantically related terms that Google associates with the same topic. The goal isn't to stuff multiple unrelated keywords onto one page. It's to cover a single topic thoroughly enough that you naturally rank for the variations. If two keywords have clearly different search intents, they need separate pages.
Free vs. paid keyword tools: is there a real difference?
Yes, and it comes down to three things. First, volume accuracy: Google Keyword Planner's free tier gives you ranges like "1K-10K" instead of exact numbers. That's a 10x spread, which is useless for prioritization. Second, difficulty scoring: free tools either lack it entirely or use simplified versions. Semrush's paid Personal KD feature adjusts difficulty based on your specific domain's authority. Third, competitive analysis: seeing which keywords your competitors rank for (and you don't) requires a paid tool. For someone just starting out, Ubersuggest at $12/month is enough. For anyone managing more than a handful of pages, a paid tool pays for itself in time savings.
Need Help Finding Keywords That Actually Drive Traffic?
Our team runs AI-powered keyword research workflows that surface low-competition opportunities across every client account. Let's find the keywords your competitors are sleeping on.
