Most content is written before anyone looks closely at the search result page it is meant to compete on. The writer has a keyword, writes the best page they can, publishes, and then wonders why it sits at position fourteen.
The result page already tells you what Google thinks satisfies that query. Reading it properly takes about fifteen minutes and changes what you write. Here is the process.
Step 1: Search the Keyword Properly
Before anything else, make sure you are looking at an unbiased result page. Use an incognito window, set your location explicitly, and disable personalization where possible. If the keyword has local intent and you are targeting a different city than the one Google inferred, every conclusion you draw afterwards will be wrong.
Search the exact phrase your target reader would type, not the sanitized version in your spreadsheet.
Step 2: Identify the Dominant Format
Look at the top ten results and ask what kind of page each one is. In practice they cluster into a few types:
- Listicles — "10 best X", common for commercial and tool queries
- Step-by-step guides — common for how-to and process queries
- Comparisons — "X vs Y", common when the searcher is choosing
- Definitions or explainers — common for informational queries
- Product or category pages — common when the intent is transactional
If eight of the top ten are listicles, Google has decided that query wants a list. Writing a narrative essay is not a stylistic choice at that point — it is a decision not to compete. This single check prevents more wasted articles than any other.
Step 3: Read the Title Patterns
Titles tell you what angle is winning. Note what repeats: numbers, years, "free", specific outcomes, brand names, comparisons. Patterns in titles reflect what has historically earned clicks on that query.
Do not copy them blindly, but do not ignore them either. If every ranking title includes a year, omitting yours signals staleness. If they all lead with a number, a title without one will not stand out in a scan.
Step 4: Gauge Depth Realistically
Open the top three and look at length, structure, and coverage. You are not trying to match word count — word count is not a ranking factor — but you are trying to understand how much ground the query expects covered.
Ask what subtopics all three include. Those are table stakes. Then ask what none of them covers well. That list is your angle.
Step 5: Check What Features Are on the Page
Before you write, note what else occupies the result page: AI overviews, featured snippets, People Also Ask, local packs, ads. These change what you are writing for. Each one is broken down in SERP features and what they take from you.
If an AI overview or a featured snippet is present, parts of your page need to be directly extractable — a clean definition, numbered steps, or a comparison table placed under a heading that mirrors the query. If a People Also Ask block is present, the questions in it are free subtopic research.
If ads dominate the top of the page, your realistic organic ceiling is lower than the position number suggests, which is worth knowing before you invest in the piece.
Step 6: Name the Gap
Write down, in one sentence, what your page will do that the current top three do not. If you cannot write that sentence, the honest conclusion is that you do not have an angle yet, and publishing anyway is how pages end up at position fourteen.
Good gaps are concrete: more recent data, a genuinely different audience, coverage of a subtopic the others skip, a clearer comparison, hands-on testing the others did not do. "Better quality" is not a gap.
Step 7: Decide Whether the Keyword Is Worth It
Sometimes the right output of this analysis is a decision not to write. Signals to walk away: the top ten are all entrenched brands with enormous authority, the intent is transactional and the page is all ads and shopping results, or you have no genuine gap to fill.
Fifteen minutes of analysis that kills a bad idea is worth more than a week of writing that produces a page nobody sees.
Doing This at Scale
The process above is manual and takes about fifteen minutes per keyword. That is fine for ten keywords and impossible for two hundred. At scale you need the same inputs delivered as data: the ranking list, the titles, the SERP features present, and how they change over time.
That is what a SERP API or a SERP analysis tool gives you — not a shortcut around thinking, but the raw material for this same analysis across a whole keyword list.
Where SerpCraft Fits
SerpCraft is built around this exact sequence. The SERP analysis step returns the ranking pages, the feature layout, and the structure of what is currently winning. The content generation step produces a draft aligned with that format rather than a generic article. Rank tracking then tells you whether the published page actually moved. The free tier covers a handful of analyses per month, which is enough to run this process on your most important keywords before paying anything.