There has never been more content, and it has never been easier to tell which of it a machine wrote. "Slop" is the word that stuck: the flood of fluent, confident, ungrounded text that reads fine and says nothing you can verify. Readers skim past it. And increasingly, so do the AI answer engines that decide which pages to quote.
That second part is the shift most sites haven't caught up to. Ranking on Google was the whole game. Now a large share of answers come from ChatGPT, Perplexity, Gemini, and Google's own AI Overviews, and they don't cite whatever ranks. Analysis from Brandlight found the overlap between top Google results and the sources AI actually cites has fallen from roughly 70% to under 20%. Being ready to be cited has become its own job, separate from ranking.
What "slop" actually is
Slop isn't just "AI wrote it." Plenty of AI drafts are useful. Slop is the set of tells that signal no one grounded this or stood behind it:
- Stock openers. "In today's fast-paced world…", "When it comes to…", or the "straightforward in theory, frustrating in practice" construction. They fill space and commit to nothing.
- Em-dashes as a default connector. One of the most reliable AI-writing signals. A comma, a period, or a rephrase almost always reads more human.
- Title-Case Headings. Capitalising Every Word reads as dated marketing copy; sentence case reads like a person wrote it.
- Unsourced claims. A statistic or a strong assertion with no named, linked source. This is the single biggest reason an answer engine won't trust a page.
- A redundant TL;DR that just restates the intro right below it. Readers, who consume only about 20 to 28% of a page on average (per Nielsen Norman Group), bounce off the padding.
- Reflexive comparisons and unfair knocks. The endless "A vs B" filler, or an unverifiable dig at a competitor. Both read as biased, and bias fails trust.
Why citability is measurable, not a vibe
Here's the part that makes this fixable rather than mystical. A peer-reviewed study from Princeton and Georgia Tech (Aggarwal et al., presented at KDD 2024) tested what actually changes how often AI answers cite a page. Three moves drove the biggest gains: citing sources, adding statistics, and adding quotations, lifting a page's visibility in AI answers by up to 40%.
Read that back as a checklist and it's almost the opposite of slop. Sourced, specific, quotable content gets cited. Generic, unsourced, hedge-everything content gets skipped. So "citability" is something you can measure and improve on purpose, not a mood.
The six dimensions we grade
CommandSEO's enrichment editor scores a document, and every section inside it, across six dimensions, each with a plain-English reason so a low number always points at something you can fix:
- Worth citing — would an engine actually quote this?
- Backed by sources — are the claims named and linked, or bare?
- Human, not slop — the tells above, caught and quoted.
- Concrete and specific — real numbers, names, dates, prices.
- Fair and fact-correct — verifiable claims, honest competitor treatment.
- Answer-ready — a clear question, answer, and structure an engine can lift.
The score isn't a black box. Every low number cites the exact span that drove it, so you can audit the tool the same way it audits your content.
How enrichment de-slops a draft
Scoring is only half of it. The anti-slop chat reads the document and quotes the exact fix: the stock opener to cut, the em-dash to replace, the claim that needs a source, the section that repeats another one. Those fixes land as word-level tracked changes right in the document, red for what goes, green for what replaces it, and you accept them one at a time or all at once. You see the projected score before you commit anything.
One detail matters more than it sounds: the tool fact-checks its own flags. If it can't find the slop it's about to call out verbatim in the text, it drops the flag. An anti-slop tool that invents problems is just a different kind of slop, so it holds itself to the bar it holds your content to.
When the draft is grounded and de-slopped, you push it, with its SEO title, meta description, and schema, straight to WordPress. The version that earns citations is the one that goes live.
It works on any content, not just ours
Enrichment sits at the opposite end of the pipeline from an AI writer. A writer produces a first draft; enrichment makes that draft worth citing, whether it came from CommandSEO's own writer, another tool, or a person. (Our writer is already built to this bar, so its output needs less de-slopping, but the editor doesn't care where the text came from.)
If you've been publishing AI drafts and wondering why they don't get picked up, this is usually why: they read as slop, and the engines that now sit between you and your audience can tell. Fix the grounding, cut the tells, and you stop being skipped.
Learn the concepts behind this in the glossary — citability and anti-slop content — or see the feature on the anti-slop SEO page. It builds on the same foundation as AEO and GEO: content an engine can read, trust, and quote.