Caption Integrity Checker: tags that died
The caption you typed is not always the caption that got saved
Caption boxes on these platforms are not plain text fields. They are rich editors with autocomplete attached to # and @. Every time you type #, a suggestion panel opens and the field re-renders. If text arrives faster than the panel settles — which is what happens when text is pasted in pieces, entered programmatically, or typed on a slow connection — characters get dropped mid-tag.
The result is the worst kind of bug: it looks fine. The post publishes, the caption reads almost normally, but #weekendtravel became #weekndtravel — a tag nobody follows and nobody searches. The entire tag-driven distribution path for that post is silently zero, and you blame the algorithm.
What it checks
- Character-level tag integrity — every hashtag, intended versus saved. A single dropped character routes the tag nowhere, and a visual skim will not catch it.
- Truncation — each platform cuts at a different length, and not always where you think.
- Stripped formatting — line breaks and blank lines are handled differently per platform; four readable lines can arrive as one wall of text.
- Tags that resolve to something else — saved correctly but too generic, ambiguous across languages, or dominated by an unrelated topic. Flagged separately from character damage.
- Encoding damage — curly quotes, em dashes, joined emoji, non-Latin scripts.
- The link that is not a link — on several surfaces a caption URL is plain text, not tappable. If that is your only path to a destination, you have no path.
What you get
1. caption-diff.html — one self-contained file. No external stylesheets, scripts, fonts, or images; no network requests. Character-level diff with insertions and deletions marked inline, a hashtag table with a verdict per tag, the truncation line drawn at the platform's cut point, and an encoding report.
2. The corruption verdict per tag — intact, corrupted, dropped, or semantic risk. Corrupted and dropped are reported first because they are absolute failures, not matters of degree.
3. A paste-safe caption block — tags moved after the body text so autocomplete fires only once the sentence is committed, formatted for a single paste rather than character-by-character entry, within the platform's limit, with risky characters substituted, plus the read-back instruction that catches a bad save before it publishes.
4. A per-platform pre-publish checklist — where the caption gets cut, whether line breaks survive, whether a URL is tappable, and the one read-back step that catches everything above.
Honest limits
- Works only from text you paste. It does not post for you and does not read your accounts. No login, no API key, no automation, no scheduling.
- It does not look up live tag volumes and will not invent one. Where volume matters it tells you what to check and how.
- Promises nothing about views, reach, or revenue.
- Declines, with the reason: buying engagement, bot or multi-account automation, tag stuffing into unrelated topics, evading platform enforcement, impersonation, and fabricating analytics.
Use it on an old post first
This defect is retroactive. If it has been happening, it has been happening for months — and confirming it on a post that already underperformed costs you nothing.
Related
Retention Check (Free) tells you whether a video's watch-through is healthy — worth running first, because if watch-through is also weak, tags were never the main problem. Retention Rebuild Kit does the full teardown. Both are on my publisher page.
