Skip to content
busyless
Book a call

Read the SERP before you write a word

The research pass that decides whether an article can rank, and how to make Claude actually run it.

8 min read

Level 0

Curious

0 XP
150 XP to ApprenticeNo streak yet0 badges0% done

You are not writing about a topic. You are writing against ten pages that already beat you for the phrase you want.

If Claude has not read those ten pages, it is guessing what the article should cover. It guesses reasonably, which is the problem, because reasonable and competitive are different things.

The pass, step by step

claude
>Research "inventory forecasting" for a guide.
WebSearch(inventory forecasting)
⎿15 results with position, page type, and URL
Six are vendor product pages, nine are editorial. Keeping the nine plus one strong vendor guide.
Sorting by page type matters. A product page tells you what Google thinks the intent is, but it is not a competitor for a guide.
WebFetch(10 URLs)
⎿Full text + heading map for each
Read(references/research-library.md)
⎿This stat was used in 4 articles already. Finding a different one.
Every competitor covers safety stock. None covers what to do when the forecast is wrong. That is the gap.
This sentence is the actual product of the research pass. Everything before it was gathering.

What you are looking for

Not "what do they say". You want three specific things:

  1. 1.The shared spine. Sections that appear in eight of ten pages. Skip one of these and the article reads as incomplete.
  2. 2.The gap. Something readers clearly want that nobody covers well. This is where the article earns its position.
  3. 3.The proof they use. Which studies, which numbers. So you can find better ones rather than repeating theirs.

Ask for the heading map, not a summary

A summary of ten pages tells you nothing useful. A list of every H2 across all ten tells you the spine and the gap in about thirty seconds of reading.

Two kinds of article, two depths of research

Article typeResearch depthWhy
Guide, how-to, comparisonFull SERP, scrape the 10 bestYou are competing on coverage and depth. You need the spine.
Roundup of toolsLighter SERP, top 3 to 5The work is verifying each tool, not out-structuring competitors.
Matching depth to type is worth doing. A full scrape for a roundup wastes an hour on pages you will not use.

Do this now

  1. 1

    Pick a keyword you actually want to rank for. Ask for the heading map.

    > Search "<your keyword>". Fetch the top 10 editorial results.
> Give me every H2 from all ten, grouped by how many pages cover it.
  2. 2

    Read the groups. The ones with 8+ pages are your spine.

    You now know what the article must cover before you have written anything. That is the whole point of the pass.

  3. 3

    Ask the question that finds the gap.

    > What does a reader searching this obviously want
> that none of these ten pages actually answers?

    Ask it directly. Left alone, it will summarize instead of finding a gap, because summarizing is safer.

Finished reading?

Mark it done to bank the XP and keep your streak alive. Any challenges on this page score separately, so you can come back for a perfect run later.