Short answer: To find profitable keywords with low competition, pair buyer-intent modifiers with your customers' problems, pull candidates from a keyword tool, then keep the ones whose results page is weak. Look for a CPC that shows buyers pay for the click, a low difficulty score, and page-one results that are forums, thin pages or old posts. Estimate what a visit is worth and start with small clusters you can publish soon.
This is our main keyword research guide: how to judge competition, read the results page, and score keywords for value. Our other keyword guides go deeper on single tasks, such as keyword difficulty, competitor keywords, long-tail keywords and keyword intent; they are listed by task under “All our keyword research guides” near the end.
A low-competition keyword is a query where page one shows few strong pages that match the searcher's intent. Typical signs are weaker domains, forum results, and thin or outdated pages.
KD compresses a messy results page into a single number. A low KD can sit on results locked by big brands with heavy backlinks, while a higher KD can be open when the lower positions hold old posts with few links. Use KD to shortlist and the live results to decide, and judge volume against your niche.
De-risk AI Overviews and zero-click loss. Search the query in an incognito window and note if an AI Overview appears consistently.
If an AI Overview appears: bias toward answer-first pages with a crisp Short Answer block the Overview can cite. Google says there are no additional requirements or special optimizations to appear in AI Overviews beyond normal SEO best practices. Avoid queries where the Overview fully resolves the task without sources.
If no AI Overview: favor long-tail clusters (3–6 posts) and add internal links from related how-tos and comparisons.
Prioritize by time-to-publish and cost per indexed page. Fast beats perfect.
Keep your cluster small enough to ship inside two weeks (3–6 URLs). Target one H2-format per article to standardize generation.
Track cost per indexed page (content + editing + design + PM). If it exceeds your 90-day LTV per visit, cut or reframe the topic.
For deeper context, see Startup SEO ROI Cost Per Indexed Page Demo.
Check intent and format first. Scan titles for the intent they serve (how-to, versus, pricing, definition) and note the content type that wins: guide, template, tool or comparison. If the top results solve a different job than your draft, change the angle or pick a tighter query. A featured snippet with weak formatting, or none, is an opening for a page with a clear direct answer.
Compare 'invoice email template' with 'how to write an invoice email'. Planning tools can show similar volume for both, yet the template query returns large libraries with heavy link profiles and ads, while the how-to query may return a forum thread and a small agency blog. The how-to query is the faster target for a practical walkthrough.
When the same URLs rank for several keywords, treat them as one topic and give them a single page. Separate pages would compete with each other.
Pick an approach based on your constraint: time, links, or budget: and know the tradeoff. The table below compares common sourcing strategies and where they win.
Standardize your checks so a junior can run the process in hours. Use a three-column scorecard: Revenue Fit (0–3), SERP Weakness (0–3), Execution Speed (0–3). Ship anything at 7+. The cutoffs below are our starting points, not industry standards; tune them to your niche and your tools.
Revenue Fit: CPC over $3 (3), $1–3 (2), under $1 (1). Modifier matches monetizable intent (pricing, alternatives, vs) add +1.
SERP Weakness: allintitle under 10 (1), 2+ forums (1), 2+ pages older than 24 months (1). DR of top 3 under 60 adds +1.
Execution Speed: You already have assets (charts, screenshots, templates) (2). You can publish 3–6 URLs in 14 days (1). If you need integrations or net-new research, subtract 1.
On-page must out-execute, not just exist. Build pages that are easy to cite and easy to top.
Lead with a 45–60 word Short Answer block. Add a scannable H2 structure that mirrors the SERP. Include a concise comparison table where intent expects it.
Add a short FAQ and internal links from adjacent posts. Cite two or three sourced quotes or stats with context to give other sites a reason to link.
You can run this system by hand, or use Mergeflo for the research and drafting. Mergeflo finds search gaps and buyer questions, drafts briefs and articles in your brand voice for review, and publishes approved articles to Webflow. It also tracks your Google positions and, on paid plans, checks whether ChatGPT, Gemini and Claude mention your brand.
Each guide below covers one keyword research task in more depth.
Tight answers to the follow-up questions.
Start with one or two small clusters, each with a few URLs you can publish and interlink properly. Add more once Search Console shows the first ones earning impressions.
Yes, but treat it as a directional signal. It is noisy for fresh news or brand-led queries, so combine it with age of top results and visible content depth.
No. KD blends links across many URLs and can mask weakness. If the top results are thin, old, or off-topic, you can win with better structure and intent coverage even at a higher difficulty score. The tradeoff is you may need stronger internal links and sharper on-page execution.
Test the query. If AI Overviews appear and cite sources, build answer-first pages with extractable 45–60 word answer blocks the Overview can cite. If the Overview solves the task without sources, downrank that keyword and focus on workflows, comparisons, or templates that require clicks.
Forum-heavy results often leave room for a page that answers the thread's core question directly. Lead with the answer and use clear headers. Google's systems choose featured snippets, and no markup can request one.
Often, when the results page is weak, the format matches, and the page has strong on-page signals and internal links from related posts. Links still help, but a weak results page lowers how many you need.
You trade ceiling for speed: and that’s fine when runway is short. Low competition keywords can cap volume, but they can bring earlier wins and cash-flow learning. Forum-heavy SERPs flip as publishers pile in; expect decay and plan refreshes. CPC as a revenue proxy can mislead in niches with ad arbitrage; validate with GSC post-launch.
Final move: pick 10 candidate queries today, run the SERP weakness checks, and commit to two clusters you can publish in the next few weeks. Ship, measure, then scale what ranks.
