How to find low competition keywords (a method that works on day one)

Most "low competition keyword" guides give you the same advice: filter your keyword tool by KD < 30, sort by volume, write the article. That advice is how thousands of sites end up writing for keywords that the tool says are easy but that are actually defended by a forum thread, a Reddit post with 800 upvotes, and three sites with 10x your domain authority.

The keyword difficulty score is a heuristic. The actual SERP is the truth. Looking at the SERP yourself for 60 seconds tells you more than any tool's KD column.

This is the workflow that produces low-competition keywords your new site can actually rank for — not just keywords that look low-competition in a dashboard.

What "low competition" actually means

A low competition keyword has all three of these properties:

1. Real demand — at least 30–50 searches per month, ideally more
2. Weak top-10 results — pages without strong topical authority, missing pieces of the answer, or from sites you can plausibly outrank
3. Clear intent — you know what content would satisfy the search

Tools approximate (1) with volume data and (2) with a difficulty score. They cannot approximate (3) — that's a judgment call. And tools often get (2) wrong because they aggregate signals (backlinks, domain authority) that don't map cleanly to whether a specific query is winnable.

Real low-competition queries usually have:

• Search volumes between 30 and 1,000/mo (above 1,000, expect more competition than tools suggest)
• Keyword difficulty under 30 in most tools, often under 15
• A SERP with at least 2–3 results that aren't from major brands
• A SERP where at least one top result is clearly thin or dated

The method

A six-step process. Steps 1–3 are about generating candidates; steps 4–6 are about validating them. Most teams over-invest in 1–3 and skip 4–6, which is why their keyword lists look full but produce no traffic.

1. Start with intent-driven seed queries

Don't start from a keyword tool. Start from your audience's actual problems.

Open a blank document. Write down 20 sentences of the form "[my buyer] is trying to [achieve outcome] but [obstacle]." Examples for a hypothetical CRM:

• A solo realtor is trying to follow up with leads but doesn't want to learn Salesforce
• A consultant is trying to track recurring clients but their CRM keeps adding fields they don't need
• A small agency is trying to migrate from Trello to a CRM without breaking client comms

Each sentence is a real problem. Each problem maps to long-tail search queries. This is how you find keywords competitors haven't found because they started from the keyword tool, not the user.

2. Expand each problem into 3–5 candidate queries

For each problem statement, write 3–5 ways a real user would phrase it as a Google search.

The realtor problem above maps to:

• "easy CRM for solo realtors"
• "alternative to Salesforce for individual agents"
• "real estate CRM for one person"
• "CRM that works on iPhone for realtors"

Write the queries the way you'd type them into Google, not the way a marketer would phrase them. Real user phrasing is messier and more specific than marketer phrasing — and that's exactly the signal that the query is low-competition.

You should now have 60–100 candidate queries.

3. Validate volume with one tool

Use Keywords Everywhere, Google Keyword Planner, or any tool you have access to. Look up volume for your candidate list.

Discard queries with zero volume — they may exist semantically but don't get searched enough to matter. Keep queries with volume of 30/mo or more.

You should now have 20–50 candidates. The next steps are about filtering these.

4. Manually check each SERP

This is the step that matters. For each candidate, search the query in Google (incognito, in the location your buyers are in) and look at the top 10 results.

What to look for:

Weakness signals (good — query is winnable):

• Forum threads (Reddit, Quora) in the top 10 — Google is showing forums because no editorial content fits
• Pages older than 2–3 years — content that hasn't been updated is beatable
• Generic listicles with surface-level coverage
• Top results from low-authority sites (small blogs, niche forums)
• Featured snippet pulled from a sub-section that doesn't really answer the query

Strength signals (bad — query is not winnable yet):

• Top 3 dominated by major brands (Wikipedia, Forbes, Investopedia, HubSpot)
• Recent (last 6 months) comprehensive guides from established sites
• Featured snippet from a clearly authoritative page
• All top 10 results from sites with high topical authority on the query

If a SERP has more weakness signals than strength signals, mark the query as a candidate. If it's the opposite, drop it.

This step takes 30–60 seconds per query. For 50 queries, that's an hour. It's the highest-leverage hour of keyword research you'll spend.

5. Confirm intent and feasibility

For each surviving candidate, ask yourself two questions:

1. Do I know exactly what content would satisfy this query? If you have to guess, the intent isn't clear and you'll likely write the wrong article.

2. Could I produce content that's clearly better than the current top 3? If yes, the query is yours to win. If no — if you don't have the expertise, the data, or the perspective — drop it. Even a low-competition query requires you to be better than what's already there.

These two questions kill about 30% of surviving candidates. That's fine. The remaining queries are ones where you have a clear path to ranking.

6. Prioritize and write

You should now have 10–25 high-quality low-competition keywords. Prioritize them by:

1. Conversion intent — queries that imply the user is closer to buying
2. Search volume — within reason, higher is better
3. Strategic value — queries that fit a broader topic cluster you want to build

Pick your top 5. Write those first.

A worked example

Say you're an indie hacker building a calendar app for freelancers. You start with the problem statement: "freelancers want a calendar that doesn't show their personal events to clients but doesn't require maintaining two separate calendars."

Candidate queries:

• "calendar that hides personal events from clients"
• "share calendar without personal events"
• "calendly alternative that hides private events"
• "freelancer calendar privacy"
• "how to share calendar availability without details"

Validate with a keyword tool. Search volumes might be 70, 320, 50, 20, 880 respectively.

Drop "freelancer calendar privacy" (too low volume). Run the others.

Manually check SERPs. Likely findings:

• "share calendar without personal events" — featured snippet from Google's own help docs, then a mix of forum threads and blog posts. Winnable.
• "calendly alternative that hides private events" — top 3 are listicles of Calendly alternatives, none specifically address the privacy angle. Winnable.
• "calendar that hides personal events from clients" — Reddit thread in top 5, the rest are dated blog posts. Highly winnable.
• "how to share calendar availability without details" — Google has dedicated help docs ranking #1 with featured snippet. Hard to win, possible to win.

You'd target the first three, possibly the fourth if you have something genuinely better than Google's help docs.

Notice what didn't happen: you didn't filter by KD score, you didn't trust the tool's "easy" tag, and you didn't generate keywords by typing seed terms into a keyword expander. You started from your buyer's actual problem and worked outward.

Tools that help

You don't need expensive tools, but a few help.

• Keywords Everywhere ($10 for credits) — volume data inline on Google searches. Highest value-per-dollar tool in this space.
• Google Keyword Planner (free with a Google Ads account) — volume data, less granular than paid tools but accurate.
• AnswerThePublic (free tier) — surfaces "people also ask" patterns and question-based queries you might miss.
• Reddit and niche forums — free, unstructured, and the source of the most authentic long-tail phrasings you'll find anywhere.
• Google itself — incognito search, location-set, manually scanning the SERP. Free and irreplaceable.

You don't need Ahrefs or Semrush for this. They're better for established sites that need to track competitor content and backlinks. For finding low-competition keywords on a new site, the free tools plus 30 minutes of SERP-staring outperform expensive tools used lazily.

Comparison: tool-based vs. SERP-based keyword filtering

Frequently Asked Questions

What's a good keyword difficulty score for low competition keywords?

For a brand-new site with no backlinks, target keywords with difficulty scores under 15 in tools like Ahrefs or Semrush, or queries the tools mark as "easy." For an established site with some authority, you can stretch up to KD 30. But treat these scores as filters for your candidate list, not as the final answer — manual SERP analysis often contradicts the difficulty score in both directions.

How many low competition keywords do I need to target?

For a topic cluster, 10–20 well-chosen low-competition keywords give you enough depth to rank for the cluster's hub page eventually. For a content site overall, plan for 50–100 low-competition targets in your first 12 months. More than that and quality usually drops faster than the SEO benefit grows.

Are zero-volume keywords worth targeting?

Sometimes. Zero-volume keywords with clear commercial intent (like a specific feature, integration, or comparison your buyers would search) can produce high-converting traffic even with low volume. They're also often genuinely uncontested. The trap is targeting zero-volume keywords without clear intent — those produce neither traffic nor conversions.

How long until low competition keywords start ranking?

For a new site, expect 4–8 weeks before pages get indexed and start showing impressions, and 2–4 months before they rank on page 1 for genuinely low-competition queries. For an established site with topical authority, the timeline compresses to weeks. The biggest variable is actually whether your content is meaningfully better than the current top 3 — if it is, ranking happens; if it isn't, no amount of SEO patience will fix it.

Can I use ChatGPT to find low competition keywords?

For generating candidate keywords, yes — ChatGPT is excellent at producing semantically related variants of a seed keyword. For quantifying competition, no — ChatGPT cannot reliably estimate search volume, difficulty, or current SERP composition, and will fabricate plausible-looking numbers if asked. Use ChatGPT to expand your candidate list. Use a real keyword tool plus a manual SERP check to filter it.

The summary

Low competition keywords aren't an output of a tool. They're an output of a workflow: real audience problems → candidate queries → volume validation → manual SERP check → intent and feasibility check → write.

The SERP check is the part most guides skip. It's also the only step that's genuinely accurate. Spend the hour. The keywords you pick afterward will rank.