Keyword difficulty (KD) is an SEO tool's estimate of how difficult it may be to reach the leading organic results for a query. It is not a Google score or your website's probability of ranking. For beginners, there is no universally best KD number: start with relevant opportunities, use one tool's scale to organize them, then inspect the actual results and the work your site can deliver.
A filter such as “KD below 30” can shorten a list inside a particular tool. It cannot tell you whether the searcher wants your page, whether you have the evidence to answer the question, or whether that topic deserves the investment. The useful output is a page decision with reasons, not a collection of green scores.
What does KD measure in each tool?
The product name belongs beside the number. Even when tools use similar scales, they can analyze different inputs and weight them differently. These are the vendors' published definitions, checked in October 2026; they are not results from a controlled comparison of paid accounts.
| Metric | Published scope | How to interpret it |
|---|---|---|
| Ahrefs KD | A 0–100 estimate of entering the top 10, based on the referring domains of the current top-ranking pages | A backlink-based estimate, not a complete assessment of content, intent or your site |
| Semrush KD% | A 0–100 estimate of the SEO effort to reach Google's organic top 10; includes link, domain-authority and SERP-related inputs | A general estimate that does not initially specify which domain you are researching for |
| Moz Difficulty | A 1–100 score analyzing the strength of the top 10 organic blue-link results | Use Moz's own scale and method; do not assume its 40 equals another tool's 40 |
| Mangools KWFinder KD | A 0–100 difficulty metric using the strength of pages on the first results page | Keep this general measure separate from its URL-specific Relative Keyword Difficulty |
See the definitions from Ahrefs, Semrush, Moz and Mangools. Their level names and recommended bands belong to their products. “Easy” is not a promise that a new page will rank quickly.
Keep the underlying units straight too. A referring domain is a website linking to a page; that website can supply more than one backlink. A vendor's estimated referring-domain requirement is not an interchangeable count of individual links, a guaranteed purchase quantity or a published Google threshold. Ahrefs' explanation of keyword difficulty also discusses factors beyond the score. Read the definition for the displayed metric rather than turning a general blog example into an exact formula.
KD, PKD and RKD answer different questions
General difficulty asks about competition for the query without first selecting your own website. Semrush's Personal Keyword Difficulty (PKD) estimates the difficulty for a specified domain to reach the top 10. Its documented method considers topical relevance and metrics for your domain and competing domains.
Mangools describes Relative Keyword Difficulty (RKD) for a specified URL. A domain and an individual page are different objects, and these are different vendors' products. Do not treat RKD as a renamed Semrush PKD, subtract one from a general score to calculate a ranking advantage, or infer a success probability from the percentage sign.

When recording a personalized metric, save the exact domain or URL you entered. A report for the wrong property, an unrelated page or a different target market does not become applicable just because the keyword text is the same. A custom AI assessment of your content is another assessment again; it should not silently replace the vendor's original KD.
What is a good keyword difficulty for a beginner?
A useful starting opportunity combines a relevant audience, a page you can actually build and a credible reason for that page to help. A lower score within one consistent tool can make a keyword worth checking earlier. It does not remove the need to examine those conditions.
For example, a low-scoring query may be navigational: the user wants a particular brand's login page. A competing tutorial may not serve that task. A higher-scoring query may still belong in your long-term plan because it describes a core product or a question customers repeatedly ask. The decision depends on the purpose, resources and current results, not just the color of the number.
Use these questions when reviewing the shortlist:
- What is the searcher trying to do? Open representative results in the intended language and market. Identify buying, comparing, troubleshooting or finding a known destination; allow for mixed intent.
- What page would serve that task? Decide whether it needs a product collection, specification page, comparison, tutorial or tool. Check whether you already have an appropriate URL to improve.
- What useful evidence can you provide? Name the actual specifications, documented procedure, original test or relevant expertise the answer needs. An attractive outline does not establish that the evidence exists.
- What work and dependencies remain? Consider content, technical access, internal navigation, promotion and the site's link plan. A tool score cannot estimate the whole project budget by itself.
- Why does the topic belong on your site? It may support a purchase decision, customer education or a useful next step. Informational traffic can have value without producing an immediate enquiry.
A forum result is not automatically weak: it may answer an experience-based question well. An old publication date does not by itself make a page obsolete. Read the content and verify a specific gap before deciding that your page could add something. Google's people-first content guidance asks about originality, completeness and usefulness, not a word-count multiplier based on KD.
Why tools disagree—and how to compare them sensibly
Two tools can return different scores for the same phrase because their models, data and result snapshots differ. That observation alone does not show either score is unstable. Instability over time would require comparable observations from the same metric under consistent conditions; cross-tool disagreement is a different question.
- Record the original label. Keep the vendor, product, metric name, value and collection date. Preserve a missing or unavailable value as missing rather than changing it to zero.
- Match the research conditions. Record the country/database, language and other available settings, plus the entered domain or URL for a personalized measure. Keep the result-refresh date when the tool exposes it.
- Read the definition. Check whether the metric assesses a query generally, a domain or a specific page. Confirm the scale and which inputs the vendor describes.
- Keep the scores side by side. Do not average them into a supposed true KD. Use one consistent metric to sort a working list, and use another source only when it helps resolve a real question.
- Inspect the important candidates. Read the relevant ranking pages, record what they answer and identify what you would do differently. Keep the original values even when your editorial decision goes against the initial score ordering.
If a tool provides sparse or stale information for your target language or country, document that limitation for the actual query. Do not assume all smaller-language markets have equally poor data, or that every low-volume query is automatically easy. Missing scores, low search-volume estimates and low KD are separate observations.
A worked example: choose a page, not the lowest score
Imagine a supplier that sells storage boxes. The following is a fictional planning exercise, not a client case, live keyword export or measured ranking opportunity. The supplier is considering “storage boxes,” “stackable storage boxes for a narrow cupboard,” and “how to measure a cupboard for storage boxes.” No KD or search-volume values have been assigned.
For the broad query, first inspect whether shoppers are being served product collections, comparisons or a mix. If a collection is appropriate, the supplier needs real inventory, dimensions and useful filters. A long blog article about the history of storage would not automatically satisfy the same task.
The narrow-cupboard query makes a more specific promise. A useful page might need outside dimensions, lid clearance, stacking constraints and a way to compare available models. Specific wording does not prove low competition or high sales potential. Verify the products and current results before selecting the format.
The measuring question can support customers before they choose a product. A guide could show which clearances matter and how to compare a measurement with the product's actual dimensions. The supplier must obtain or create accurate instructions and images; it should not invent a test result. A relevant link to the product collection can then help readers use what they learned.
Suppose a tool later assigns the lowest score to the broad query. That is a reason to inspect it, not a reason to publish an unrelated article. Conversely, an unavailable score for the measuring question does not mean the question is worthless. The recorded decision should name the page, the reader's task, the evidence still needed and the reason for proceeding or waiting.
Do not confuse KD with search volume or advertising competition
Difficulty asks about competition; volume estimates demand. Neither is a forecast of your page's organic visits. When sources disagree about demand, use the search-volume cross-checking method and keep geography, periods and variant handling visible instead of adding or averaging incompatible estimates.
Google Ads uses another meaning of “Competition.” Its Keyword Planner definitions describe the number of advertisers showing for a keyword relative to all keywords, under the selected location and Search Network. That is advertising competition, not organic KD. Top-of-page bids, monthly searches and projected ad clicks have their own definitions too.
A high bid can prompt a commercial investigation, but it does not establish your profit margin or prove SEO will be worthwhile. Likewise, a low-volume estimate does not establish that a specific customer question has no value. Keep the business decision tied to your actual audience and offer.
Turn the shortlist into a plan you can review
For a new site, the keyword-list and page-mapping workflow turns those judgments into records with sources, page destinations, evidence requirements and next actions.
Keep a research record containing the query, market, tool and metric, observation date, value or missing status, inspected result URLs, intended reader task, proposed page and required evidence. Add a clear next action: improve an existing page, gather product material, commission an answer or defer the topic. You can begin with the keyword-research tools with free access, provided their current limits cover the job.
After publication, record actual visibility and business outcomes separately from the original estimate. Compare the same queries and pages under comparable conditions. A score changing does not prove your content improved; a ranking change does not isolate the cause of that change. Revisit your assumptions when evidence accumulates rather than treating an arbitrary number of days as a pass-or-fail deadline.
The next practical checks are covered in the Google-ranking implementation checklist. For a site-wide plan, Guangsuan's Google SEO service (Chinese) combines technical work, content and link resources. Bring the website, target market and real product or service material to the discussion. The objective is to choose and deliver useful pages with a reasoned investment plan—not to buy a particular KD color.