FairDesk Tools · $0.010/use · No signup needed

QA Every Support Reply Before You Hit Send

Paste your drafted response, get a quality score across 5 dimensions, specific improvement suggestions, and an optimized rewrite.

Example

Input

Draft: "Hey, so about your problem — we looked into it and couldn't figure it out. Maybe try clearing your cache or something? If that doesn't work just submit another ticket lol. Sorry bout that."
Context: "Customer reported data loss in their account after the latest update. They've been a paying customer for 2 years."

Output

{
  "overall_score": 2.7,
  "dimensions": {
    "accuracy": { "score": 3, "reasoning": "Claims investigation happened ('looked into it') but provides zero details. No mention of the data loss — the actual issue." },
    "tone": { "score": 1, "reasoning": "'lol' in a data loss response is unprofessional. 'Sorry bout that' minimizes a serious issue. Casual tone is inappropriate for a 2-year customer reporting data loss." },
    "completeness": { "score": 2, "reasoning": "Doesn't address the data loss at all. Suggests cache clearing, which is irrelevant to data loss. No timeline, no escalation, no accountability." },
    "clarity": { "score": 4, "reasoning": "Message is understandable but vague. 'Your problem' — which problem? 'Couldn't figure it out' — what was investigated?" },
    "actionability": { "score": 3, "reasoning": "'Submit another ticket' puts the burden back on the customer. No clear next step from the support team." }
  },
  "critical_issues": [
    { "issue": "Data loss not acknowledged or addressed", "severity": "blocking", "recommendation": "The customer reported data loss. The response must acknowledge this, explain what's being done to recover data, and provide a timeline." },
    { "issue": "'lol' in a serious support context", "severity": "major", "recommendation": "Remove immediately. This signals the agent doesn't take the customer's problem seriously." }
  ],
  "improvements": [
    { "original_phrase": "Hey, so about your problem", "suggested_phrase": "Hi [Name], thank you for reporting the data issue in your account", "dimension": "tone", "impact": "critical", "explanation": "Specific acknowledgment replaces vague, casual opening" },
    { "original_phrase": "we looked into it and couldn't figure it out", "suggested_phrase": "I've escalated this to our engineering team as a priority investigation", "dimension": "actionability", "impact": "critical", "explanation": "Shows concrete action instead of admitting defeat" }
  ],
  "rewritten_version": "Hi [Name],\n\nThank you for letting us know about the data issue in your account following the recent update. I take this seriously — data integrity is our top priority, especially for a valued customer of two years.\n\nHere's what's happening:\n\n1. I've escalated this directly to our engineering team as a priority investigation\n2. They're looking into whether the data can be recovered from our backup systems\n3. I'll update you within 4 hours with their findings\n\nIn the meantime, please don't make any changes to your account — we want to preserve the current state for recovery.\n\nI'll personally follow up today. You have my word this won't require another ticket."
}

Frequently asked

What does the quality score measure?: Five dimensions: Accuracy (is information correct?), Tone (appropriate for the situation?), Completeness (addresses all parts of the issue?), Clarity (easy to understand?), and Actionability (customer knows what happens next?). The overall score is a weighted average.
Will this catch factual errors?: Yes. If your draft mentions a wrong amount, promises an impossible timeline, or makes claims inconsistent with the ticket context, the tool flags it as a critical issue. This is the #1 reason teams use QA tooling.
How is the rewritten version different from the Tone Rewriter?: The Tone Rewriter changes the voice of a message. QA Response reviews the substance — accuracy, completeness, actionability — and rewrites to fix content problems, not just tone problems. Use QA for draft review, Tone Rewriter for voice adjustment.
Can I use this without ticket context?: Yes. The context field is optional. Without it, the tool evaluates the draft on its own merits. With context, it can check whether the response actually addresses the customer's issue and flag mismatches.
What score means "ready to send"?: Scores 8+ generally indicate a strong response. Scores 5-7 have specific improvements to make. Below 5 usually means the response needs a significant rewrite. No critical issues is the real gate — a score of 7 with zero critical issues is better than a 9 with one.