Reading a SERP Before You Commit to the Keyword

Most keyword selection happens in a spreadsheet. Volume in one column, difficulty in another, a filter applied, and a target list produced. The step that gets skipped is the one that would have prevented most of the wasted effort: actually looking at the results page for the query.

A difficulty score is a model of link strength. It cannot tell you what kind of content the query returns, whether the page format you intend to publish appears at all, or whether the clicks that exist are already being absorbed before anyone reaches the organic results. All of that is visible in thirty seconds of looking.

First Question: What Kind of Page Ranks?

Look at the top ten and classify them. Are they product pages, editorial guides, comparison tables, forum threads, videos, or tools? This is the query's revealed search intent, and it is not negotiable. If the entire first page is user discussion and you plan to publish a vendor guide, you are not competing on quality. You are proposing a format the query does not reward.

This single check invalidates a large share of typical target lists. A commercial term where all ten results are editorial roundups is not a page you can win with a product page, no matter how good the product page is.

Second Question: Who Actually Ranks?

Domain-level metrics mislead here. What matters is whether the ranking pages resemble anything you could plausibly produce.

Three patterns are worth naming. If the page is entirely dominated by very large publishers and marketplaces, the query is usually not winnable for a smaller site regardless of content quality, because the ranking is driven by entity strength rather than page merit. If the page contains a mix of large and small sites, there is room. If it contains forum threads and thin pages ranking well, the query is under-served and represents genuine opportunity, because the bar is low and the intent is real.

That third pattern is the most valuable and the one difficulty scores flag least reliably. A high domain authority average on a page full of mediocre content is a soft target wearing a hard number.

Third Question: How Much Click Is Left?

Count what sits above the first organic result. An featured snippet, a People Also Ask block, a local pack, a shopping unit, a video carousel, and four ads can push the first organic listing well below the fold. Ranking first on that page is worth a fraction of ranking first on a clean one.

This is where volume figures become actively misleading. A query with ten thousand monthly searches and a saturated results page can deliver fewer visits than one with eight hundred and nothing above the organic results. Compare queries on estimated available clicks rather than raw volume, even if the estimate is rough.

Features You Can Take

The features are not purely a cost. If a featured snippet is present, it is winnable, and winning it is often easier than moving from position four to position one. If the People Also Ask questions reveal sub-questions you can answer better, they are a content brief. Read the features as instructions, not just as obstacles.

Read the Titles, Not Just the Domains

The titles on the results page are a compressed statement of what the query rewards. If eight of ten include a year, currency is part of the expectation. If most are framed as questions, the query is informational regardless of how commercial the keyword sounds. If several promise a specific number of items, the format expectation is a list of roughly that size.

These are cheap signals and they are usually more reliable than any tool output, because they reflect what actually won rather than what someone predicted would win. Skim them before writing a brief and the brief tends to write itself.

Fourth Question: How Old Is What Ranks?

Look at the publish dates. If the top results are all recent, the query rewards freshness and any content you publish will need maintenance to hold position. If the top results are years old and still ranking, the query is stable and a good piece will earn a long return. If the top results are old and visibly outdated, that is an opening, and it is one of the few situations where simply being current is a competitive advantage.

A Worked Example

Take a query like "best project management software for agencies." The spreadsheet says decent volume and high difficulty, and most teams either chase it or discard it on the number alone.

Reading the page tells you more. If the top ten are all editorial roundups from software review sites, the format is a listicle and a product page cannot win it, regardless of how good the product is. If those roundups are two years old, freshness is an opening. If there is a featured snippet pulled from a comparison table, that table is winnable independently of the ranking. If four ads and a shopping unit sit above everything, the available organic click is much smaller than the volume implies.

The conclusion might be: do not target this query with a product page, do target it with a genuinely current comparison piece, expect the realistic ceiling to be position four or five, and treat the snippet as the actual prize. That is a completely different plan from "high difficulty, skip it," and it came from looking rather than filtering.

Record What You Saw

Write the read down next to the keyword: dominant format, who ranks, features present, freshness. Six months later when the page underperforms, that note is the difference between diagnosing the problem and guessing at it. It also stops the same query being re-evaluated from scratch every planning cycle, and it makes disagreements about targeting concrete rather than a contest of opinions about difficulty scores.

Putting It Together

A workable process is to shortlist in the spreadsheet and decide on the results page. For every candidate that survives the volume and difficulty filter, spend a minute answering four questions: what format ranks, whether anyone like me ranks, how much click is left after the features, and how fresh the winners are. Record the answers next to the keyword.

Doing this well takes about a minute per keyword, which sounds expensive until you compare it against the cost of producing a piece of content that had no chance. One afternoon spent reading results pages routinely saves a quarter of misdirected production.

Most lists shrink by half. That is the point. The remaining queries are the ones where a good piece of content has a real path, and the effort saved on the discarded half funds doing the survivors properly. Publishing against a query you cannot win is not a small loss; it consumes the same production budget as one you can.