Strategy · September 18, 2026 · 8 min read
Read the SERP Before You Write: A Manual Difficulty Check That Beats KD Scores
A four-minute read of page one predicts whether you can rank far better than any two-digit difficulty score - here are the six signals to check first.
By FluxWriter Team
Manual SERP difficulty checks are more reliable than any keyword difficulty score, because they read the pages you would actually have to outrank rather than averaging link metrics across ten of them. Most keyword research stops at a two-digit number and treats anything under 30 as fair game. This covers the six signals worth reading on page one, how to score them into a decision, and when a high-difficulty term is still worth writing.
Why Difficulty Scores Fail on Specific Queries
Keyword difficulty is a link estimate wearing the costume of a verdict. Ahrefs, Semrush and Moz all publish one on a 0 to 100 scale, and the backbone of each is the same input — how many referring domains point at the pages already ranking.
The estimate holds up reasonably well for broad head terms. It falls apart on the long tail, where pages often rank with strong domain metrics for reasons unrelated to the query — a forum thread on a giant site, a category page that mentions the phrase once, a roundup nobody has touched since 2021.
Two queries can both score 28 and be nothing alike. One has ten dedicated pages written by people who obviously know the subject. The other has three forum posts, a supplier PDF, and an encyclopedia section that happens to contain the words.
The score cannot separate those two. It never reads the pages.
There is a second problem worth naming. Run the same query through two tools and the numbers routinely disagree, because each is built on a different link index and a different formula. There is no exchange rate between them. Use the score to sort 400 candidate terms down to 40 — then read the results yourself before committing a single brief.
The Six Signals Worth Reading
A page-one read takes roughly 3 to 5 minutes per query. Six signals carry almost all of the information, and none of them require a paid tool.
Intent match. Does the result set answer the question you are planning to answer? If your draft is a how-to guide and the top 10 are product and pricing pages, difficulty is not your problem — format is.
Dedicated pages versus passing mentions. A page built specifically to answer this query is a genuine competitor. A 4,000-word guide that covers your topic in one subheading is not. Beat that one with 900 focused words on the exact question.
The weakest results, not the strongest. Positions 8, 9 and 10 are your entry point. When two of them are off-topic or visibly thin, page one has a soft floor and you never have to beat the top three.
Who is ranking. Forums, Q&A sites and community threads sitting in the top 5 usually mean nobody has published a proper answer yet. That is the most valuable pattern a result set can hand you.
Freshness of the leaders. Look at the visible dates on the top three, and at whether the copy still quotes tools or prices that changed two years ago. A page one where nothing has been meaningfully updated in 2 to 3 years is an unmaintained topic. Unmaintained topics move.
Feature crowding. Count what sits above the first organic result: ads, an AI Overview, a video carousel, People Also Ask. When a searcher has to scroll past most of a screen to reach result one, the query is worth less than its volume suggests — regardless of how easy it is to rank.
Two signals agreeing is a hint. Four agreeing is a decision.
Turning the Read Into a Score
Signals only matter if they produce a decision, and the decision should take seconds. Score each result set on what you actually see:
| What You See on Page One | What It Means | Your Move |
|---|---|---|
| 2 or more forum threads in the top 5 | No one has written the real answer | Write it, high priority |
| Every result is a dedicated page on the exact term | Intent is already well served | Skip unless you add original data |
| Top 3 all updated within 6 months | Actively defended topic | Needs an angle, not more words |
| Results answer a different question | Intent mismatch | Write it — pick one meaning and commit |
| Positions 8–10 thin or off-topic | Soft floor at the bottom | Target position 8 first |
Rows one, four and five are your build list. Row three burns more production budget than any other pattern here — row two quietly burns the rest.
A crude count works better than a weighted model here. Give the query one point for each of these: a forum or Q&A result in the top 5, a top-three page older than 2 years, an obvious intent gap, and a weak result in positions 8 to 10. Three points or more means write it now. Zero or one means park it until the site has more authority. Score 20 candidates this way and the order sorts itself.
Intent Comes Before Difficulty
Format is decided by the results — not by you. That single sentence saves more wasted drafts than any other rule in keyword research.
When 8 of the 10 results are comparison tables, an essay will not rank there no matter how good it is. When they are all sitting between 1,100 and 1,600 words, a 4,000-word monster is not a competitive advantage — it is a mismatch with what searchers on that query accepted.
Check three things before you write a word. What page type dominates, roughly how long the top three run, and whether the results skew commercial or informational. Those three answers become constraints on the brief.
Intent mismatch is also your best opportunity. If the results are a muddle of two different meanings, one of those meanings is underserved, and a page that commits fully to one of them frequently outranks pages carrying far heavier link profiles.
When a High Score Is Still Worth Writing
A difficulty score of 45 on a query with a broken result set is a better bet than a 12 on a query with ten strong dedicated pages. That trade comes up more often than the tools admit.
Four situations justify ignoring a high number. The first is a page one gone stale across the board. The second is a result set that answers a related but different question, leaving yours open. Third: aggregators and directories holding most of the top 10, where a real publisher page has an obvious edge. The fourth is first-hand data — your own pricing, your own before-and-after numbers — that nobody currently ranking can match.
One quieter case belongs on the list. When the ranking pages all sell something and the query is clearly informational, a neutral answer can slot in within 60 to 90 days on a domain with modest authority.
Fix: keep a running list of these mismatches as you research. They are worth more than the low-difficulty terms sitting next to them in the export.
Where the Manual Read Misleads You
The check has real failure modes. Results are personalized and localized, so a signed-in browser carrying your own search history shows you a flattering picture. Use a private window, and set the region in Google's search settings before you judge anything.
Sample size is the second trap. One query is not a topic. Read 3 to 5 related queries from the same cluster before deciding a whole subject area is soft, because the page-one profile often changes sharply between the head term and its variations.
The third limit is the one people argue with. Reading the SERP does not repeal link math. On a competitive commercial term where the top 10 all carry 300 or more referring domains, a better page will not close that gap on its own, and pretending otherwise burns two quarters. The manual check tells you whether the content bar is beatable. It says nothing about whether the authority bar is.
One more caution. The result set you read today reflects today's index — a core update can reshuffle it, so re-read the page for any query still unwritten after 90 days.
FAQ
How many results do I actually need to look at?
The top 10, plus a glance at what sits above them. Positions 11 to 20 rarely change a decision. Give each query 3 to 5 minutes, and if you cannot form a verdict in that window, the answer is usually that the query is ambiguous and needs splitting.
Should I stop using keyword difficulty scores altogether?
No, and abandoning them entirely is an overcorrection. Scores are a fast sorting mechanism across thousands of terms — something manual reading can never be. Use them to shortlist, then read page one for every term that reaches your brief stage. Sort with the number, decide with your eyes.
Does this still work when an AI Overview takes the top of the page?
Yes, with one adjustment. Judge winnability and click value separately — a query you can rank for may still deliver far fewer clicks once a generated answer sits above the results. Estimates of that drop vary too widely to plan around, so keep the difficulty read and discount the forecast on feature-heavy queries.
The Practical Takeaway
Sort with the tool, decide with the results page. Export your candidate terms, filter to the 40 that clear your volume floor, then open page one for each in a private window with the location set. Score four things: a forum result in the top 5, a stale top three, an intent gap, and a weak result at positions 8 to 10. Three or more points means it goes in the queue this month, and anything below two waits. Start with the 10 queries you were already planning to write, and expect to cut 3 of them.
If you are producing a high volume of briefs against that kind of shortlist, tools like FluxWriter can help turn an approved query into a drafted post that matches the format the results demand — but reading page one and deciding which queries deserve a brief at all stays a human judgement call.