What is clickbait (and how does Seeklyn detect it)?

2 patterns count as clickbait markers in Seeklyn's honesty check. First, an ALL-CAPS alphabetic word of 4 or more letters — for example "SHOCKING", "FREE", or "BEST" written fully capitalised. Short initialisms like "AI" or "UK" are 2–3 letters and do not trigger the rule. Second, an ellipsis — either the Unicode character "…" or the 3-dot sequence "..." — which signals a curiosity gap and withholds the answer rather than stating it.

Superlatives, power words, and listicle framing ("best", "ultimate", "top 10") are deliberately not treated as clickbait. Research and Seeklyn's own scoring data show that these formats are cited more by answer engines, not less, so penalising them would steer writers away from effective headlines. Seeklyn's honesty check targets only signals that deceive — shouting and teasing — not signals that emphasise.

The honesty component starts at 20 points inside the AEO sub-score, and each marker costs 10 points, floored at 0. A headline with 0 markers scores the full 20 honesty points. 1 marker (either ALL-CAPS or ellipsis) scores 10. 2 markers (both present) score 0. Both honesty points feed directly into the AEO sub-score, which combines answerability (0–30), entity front-load (0–25), specificity (0–25), and honesty (0–20) into a total of up to 100. The overall headline score is then 0.5 × CTR + 0.5 × AEO.

Seeklyn's honesty moat is built on precision: fire only on patterns that reliably signal deception rather than catching legitimate emphasis. An ALL-CAPS shout word creates an aggressive tone without adding information; an ellipsis withholds the answer to manufacture suspense. Both patterns correlate with content that underdelivers on its headline promise. By limiting detection to 2 well-defined markers, Seeklyn avoids false positives that would punish legitimate stylistic choices — a policy consistent with its broader principle that applying a fix must never lower a comparable score.