Inside the Assay: How We Score an Article in 14 Signals
Every experiment in AutoBlogLab ends the same way: three drafts, three scores, one winner. The score — we call it the Assay — is the part of the product customers ask about most, so this post opens the hood completely.
Why a score at all?
Content decisions inside teams are usually settled by seniority or exhaustion. Someone likes draft A, someone likes draft B, and the person with the strongest opinion or the latest meeting wins. A shared, inspectable score changes that dynamic: the debate moves from "which do you like?" to "which signal are we willing to trade away?" — a far more productive argument.
The four signal groups
Intent match asks whether the draft answers the question a searcher is actually asking. We check query alignment, answer completeness and topical depth. A beautifully written article that answers the wrong question scores low here, on purpose.
Structure covers heading hierarchy, scannability, introduction quality and internal-link opportunities. Structure is the cheapest signal to fix and the most commonly neglected, which is why it gets its own group instead of hiding inside readability.
Readability measures sentence variety, reading grade, transition quality and jargon density. We deliberately reward variety rather than simplicity alone — a wall of short sentences reads as robotic, and readers notice.
Originality is the group that keeps AI-assisted content honest: angle distinctiveness, example specificity and phrase-pattern variety. Drafts that could have been written about any topic in the niche score poorly; drafts with concrete, specific material score well.
How the groups combine
Each signal produces a 0–100 sub-score; groups are weighted by content type. A product description weights intent and readability heavily; an ultimate guide weights depth and structure. The final number is useful, but the breakdown is the point — it tells you what to improve, not just where you stand.
What the Assay is not
It is not a ranking guarantee, and we will never market it as one. Search performance depends on your domain, your competition and factors no tool can see. The Assay measures the qualities of the draft that are under your control — which is exactly where your effort belongs.