Competitor Page Analyzer

Writing content without checking what's already ranking is taking an exam without reading the syllabus. The top 3 results for any keyword are Google's revealed preference for that query — the length, the structure, the topics, the depth that has already been judged good enough to rank. This skill reverse-engineers that pattern and turns it into a numeric target you can write against.

Step zero is always SERP intent. If the top results are commercial listicles and you planned a how-to guide, the blueprint is moot — you'd be writing the wrong format. The skill classifies the SERP into informational, commercial, answer-first, or mixed first, and if it disagrees with what you said you wanted to write, it surfaces the conflict before anything else. The "avg + 10%" word-count target is the default for in-depth informational SERPs only — an answer-first SERP wants speed-to-answer, not length, and padding actively hurts there.

The output is a numbers-first blueprint: a per-competitor table with word count, H2/H3 count, image count, FAQ presence, internal links, plus a topic coverage map showing which H2-level topics all three cover (must-include), which the majority cover (recommended), and which are unique angles. Then the content gaps — what none of the top three cover — because that's where your post differentiates. Every average shows its arithmetic so the target is verifiable, not asserted. When the values are skewed, the skill reports the range and recommends the cluster the majority of winners sit in instead of a misleading mean.

A few things it deliberately won't do: it won't fabricate a blueprint from pages you haven't supplied — if URLs are missing, it asks for them and stops. It won't accept "just guess the metrics" — invented data poisons the averages and the recommendation. And it won't tell you to copy the winners. The goal is to clear the bar set by the SERP, then differentiate with the gaps.