Customer Support prompts

Generates a polished first-reply email to any incoming support ticket, saving agents time on routine responses.

You are a friendly, professional customer support agent for [COMPANY_NAME], a [INDUSTRY] company. A customer has submitted the following support ticket:

---
[PASTE_TICKET_HERE]
---

Write a first-response email that:
1. Acknowledges the customer's issue with empathy.
2. Confirms what you understand their problem to be.
3. Provides a clear next step or resolution timeline.
4. Closes warmly with your name ([AGENT_NAME]) and team name ([TEAM_NAME]).

Keep the tone [TONE: e.g. warm and casual / formal and professional] and the response under 150 words.

How to use: Paste the raw ticket text and adjust the tone placeholder to match your brand voice.

Turns a rough article outline into a publication-ready help center doc suitable for a self-service knowledge base.

Write a clear, friendly help center article for [PRODUCT_NAME] customers on the topic: '[ARTICLE_TOPIC]'.

Use this outline as your structure:
[PASTE_OUTLINE_HERE]

Guidelines:
- Use second-person ('you') throughout.
- Break each section into short paragraphs of 2–3 sentences.
- Use numbered steps for any process.
- Add a 'Tip:' callout box wherever a common mistake could occur.
- End with a 'Still need help?' section pointing customers to [SUPPORT_CHANNEL].

Target reader: [TARGET_AUDIENCE, e.g. non-technical small business owners].

How to use: Supply a bullet-point outline and specify your audience's technical level for appropriately scoped language.

Crafts a de-escalating, legally cautious reply to an angry customer complaint while offering a concrete remedy.

You are a senior customer support specialist. A customer has sent the following angry complaint:

---
[PASTE_COMPLAINT_HERE]
---

Write a de-escalating reply that:
1. Opens by sincerely apologizing without admitting legal liability.
2. Validates the customer's frustration in one sentence.
3. Explains what went wrong (use only the facts provided — do not invent details).
4. States the concrete remedy you are offering: [REMEDY, e.g. full refund / replacement / account credit].
5. Provides a direct contact for follow-up: [CONTACT_DETAILS].

Tone: calm, empathetic, solution-focused. Maximum 200 words.

How to use: Fill in the actual remedy being offered before sending — never leave the placeholder blank in a live response.

Extracts recurring questions from raw ticket data and converts them into a structured, publishable FAQ section.

Below are [NUMBER] recent customer support tickets for [PRODUCT_OR_SERVICE]:

---
[PASTE_TICKETS_HERE]
---

Analyze these tickets and:
1. Identify the top recurring questions or issues.
2. Group them into [NUMBER_OF_CATEGORIES] logical categories.
3. Write a clean FAQ section with a bold Question and a concise 2–4 sentence Answer for each.
4. Order questions from most to least common based on the data provided.

Format the output as ready-to-paste help center content.

How to use: Paste at least 10–20 tickets for meaningful pattern detection; adjust category count to fit your help center structure.

Humanizes overly formal or template-sounding support replies while preserving all factual content.

The following customer support reply sounds stiff and robotic:

---
[PASTE_ORIGINAL_REPLY_HERE]
---

Rewrite it so it:
- Sounds warm, natural, and human — like a real person wrote it.
- Keeps all the factual information intact.
- Removes jargon, passive voice, and corporate filler phrases.
- Matches the brand voice: [BRAND_VOICE_DESCRIPTION, e.g. friendly and straightforward, like a knowledgeable friend].
- Stays within 10% of the original word count.

Return only the rewritten reply, no commentary.

How to use: Describe your brand voice specifically — vague descriptors yield generic results.

Helps agents decline an out-of-policy refund request while preserving the customer relationship through empathy and an alternative offer.

A customer is requesting a refund outside your standard policy window. Here is their message:

---
[PASTE_CUSTOMER_MESSAGE_HERE]
---

Your refund policy states: [PASTE_POLICY_HERE]

Write a reply that:
1. Acknowledges the customer's situation with genuine empathy.
2. Clearly but kindly explains why the standard policy applies in this case.
3. Offers at least one goodwill alternative (e.g., [ALTERNATIVE_OFFER, e.g. store credit, extended support, a discount on next purchase]).
4. Leaves the door open for the customer to respond without encouraging policy escalation.

Tone: firm but compassionate. Under 180 words.

How to use: Always include the exact policy text so the AI can reference it accurately rather than inventing terms.

Produces a safe, escalating troubleshooting guide for any product issue, formatted for a specific audience technical level.

Write a numbered, step-by-step troubleshooting guide for the following issue: '[ISSUE_DESCRIPTION]' affecting [PRODUCT_NAME].

The guide should:
- Be written for a [TECHNICAL_LEVEL: beginner / intermediate / advanced] user.
- Start with the quickest/easiest fix and escalate to more complex steps.
- Include a 'What to check before you start' prerequisites section.
- Flag any step that could cause data loss with a ⚠️ warning.
- End with a 'If none of these steps worked' section directing users to [ESCALATION_PATH].

Do not invent product-specific menus or settings — use only the information I have provided.

How to use: Provide as much context about the issue as possible; the instruction to avoid invented details prevents hallucinated steps.

Condenses long, multi-turn support threads into a structured briefing and next-action plan for agents picking up a conversation.

Below is a full customer support email thread:

---
[PASTE_THREAD_HERE]
---

Please produce:
1. A 3-sentence summary of the issue and its current status.
2. A bulleted list of all open action items, each with a responsible party (Customer / Agent / Engineering / etc.).
3. Any commitments or deadlines that were made to the customer, listed separately.
4. A recommended next reply (2–3 sentences) to send to the customer right now.

Format with clear headings for each section.

How to use: Ideal for handoff situations — paste the thread when a new agent is taking over a complex ticket.

Drafts a clear, honest customer-facing incident or outage notification that builds trust rather than eroding it.

You need to proactively contact customers affected by the following incident:

Incident summary: [DESCRIBE_THE_INCIDENT]
Affected customers: [DESCRIBE_AFFECTED_SEGMENT]
Current status: [INVESTIGATING / IDENTIFIED / MONITORING / RESOLVED]
Estimated resolution time (if known): [ETA_OR_UNKNOWN]

Write a short, transparent outreach message (email or in-app notification — specify: [CHANNEL]) that:
1. States what is happening plainly in the first sentence.
2. Acknowledges the impact on the customer.
3. Shares what your team is doing.
4. Sets realistic expectations about updates.
5. Does NOT over-promise a fix time unless you have confirmed it.

Max 120 words. Avoid jargon.

How to use: Update the status field as the incident progresses and regenerate for each new communication.

Generates a ready-to-import set of canned replies with variable placeholders for a support team's response library.

Create a library of [NUMBER] canned responses for [COMPANY_NAME]'s support team covering the following common situations:

[LIST_SITUATIONS_HERE, e.g.
- Password reset request
- Billing question escalation
- Feature not yet available
- Shipping delay notification]

For each canned response:
- Give it a short internal title (for the agent's menu).
- Write the customer-facing text using [BRAND_VOICE: e.g. warm, concise, professional].
- Include [PLACEHOLDER_TAGS] in curly braces for variable info (e.g. {customer_name}, {order_number}).
- Keep each response under 100 words.

Format as a numbered list with the internal title bolded.

How to use: List every situation you want covered; the more specific the scenarios, the more tailored the outputs.

Crafts a dignified, retention-focused reply to a churn-risk customer that respects their decision while presenting value.

A customer has sent the following message indicating they intend to cancel their account:

---
[PASTE_MESSAGE_HERE]
---

Context about this customer (fill in what you know):
- Plan/tier: [PLAN]
- Time as a customer: [DURATION]
- Known issues they've experienced: [ISSUES_IF_ANY]
- Retention offer you are authorized to make: [OFFER_OR_NONE]

Write a retention-focused reply that:
1. Takes their frustration seriously without being sycophantic.
2. Acknowledges the specific pain point they raised.
3. Presents the retention offer (if any) naturally, not as a bribe.
4. Respects their autonomy — do not beg or guilt-trip.
5. Makes cancelling easy if they still want to, to preserve goodwill.

Tone: genuine, respectful, low-pressure.

How to use: Only include a retention offer if you are actually authorized to make it — do not fabricate offers.

Converts internal engineering release notes into a customer-friendly product update email with a clear benefit focus.

Transform the following internal release notes into a customer-friendly update email:

---
[PASTE_INTERNAL_RELEASE_NOTES_HERE]
---

Guidelines:
- Lead with customer benefit, not technical detail.
- Use plain language suitable for [AUDIENCE, e.g. non-technical users].
- Highlight the top [NUMBER] improvements with short, punchy bullet points.
- Translate any technical terms using the following glossary: [GLOSSARY_OR_NONE].
- End with a clear call-to-action: [CTA, e.g. 'Try it now', 'Read the full docs', 'Watch the walkthrough'].
- Subject line: suggest 3 options.

Do not add features or fixes that are not in the notes I provided.

How to use: Include a glossary for technical terms if your audience is non-technical to prevent oversimplification errors.

Creates an interactive training simulation where the AI plays a difficult customer and then debriefs the agent afterward.

You are going to play the role of a difficult customer for customer support training purposes. Do not break character.

Scenario: [DESCRIBE_SCENARIO, e.g. A customer who received a damaged product and has already contacted support twice without resolution. They are frustrated and skeptical.]

Customer persona:
- Communication style: [e.g. passive-aggressive, very direct, emotional]
- Core complaint: [CORE_COMPLAINT]
- What they actually need to feel resolved: [UNDERLYING_NEED]

Begin the conversation as the customer sending their opening message. When the trainee (support agent) responds, reply in character based on their reply. Continue until told to stop.

After the trainee types 'END ROLEPLAY', step out of character and give them feedback on: empathy, clarity, resolution effectiveness, and tone.

How to use: Use this with new hires — define the persona carefully so the difficulty level matches the training goal.

Performs a structured editorial and usability audit of a help article, returning actionable line-level feedback.

Please audit the following help center article:

---
[PASTE_ARTICLE_HERE]
---

Evaluate it against these criteria and return a structured report:

1. **Clarity** – Are instructions easy to follow for a [TARGET_AUDIENCE] user? Flag any jargon or ambiguous steps.
2. **Completeness** – Are there obvious gaps a user would get stuck on?
3. **Tone** – Does it match [BRAND_VOICE]? Highlight any off-brand sentences.
4. **Structure** – Is the heading hierarchy logical? Are steps numbered where they should be?
5. **Accuracy risks** – Flag any claims that depend on product details I should verify (do not invent corrections).

For each issue found, quote the problematic text and suggest a specific fix.
End with an overall score out of 10 and a priority list of edits.

How to use: Run this before publishing or during quarterly content reviews; flag accuracy risks for SME verification.

Drafts a public, brand-positive response to a negative review that demonstrates accountability and moves toward resolution.

A customer has left the following [STAR_RATING]-star public review on [PLATFORM, e.g. Google / Trustpilot / App Store]:

---
[PASTE_REVIEW_HERE]
---

Write a public response on behalf of [COMPANY_NAME] that:
1. Thanks the reviewer by [FIRST_NAME_IF_KNOWN / 'name' placeholder] for their feedback.
2. Addresses their specific complaint directly — do not give a generic reply.
3. Does NOT make excuses or get defensive.
4. Offers a concrete next step (e.g., 'Please reach out to us at [CONTACT]') to move the conversation private.
5. Is visible to future readers as proof of responsive service.

Max 100 words. Tone: [TONE, e.g. professional and warm]. Do not promise outcomes you cannot guarantee.

How to use: Keep responses short for public platforms — length reads as defensiveness to onlookers.

Produces a clear, professional internal escalation email that gives a manager everything they need to act quickly.

Write an internal escalation email from a support agent to their manager regarding the following customer situation:

Customer name: [CUSTOMER_NAME]
Ticket ID: [TICKET_ID]
Issue summary: [DESCRIBE_ISSUE]
What has already been tried: [STEPS_ALREADY_TAKEN]
Why it needs escalation: [REASON, e.g. requires refund approval above agent limit / technical issue needs engineering / legal risk]
Urgency level: [LOW / MEDIUM / HIGH / CRITICAL]

The email should:
- Be concise and factual (no venting).
- Clearly state the decision or action needed from the manager.
- Attach relevant context without requiring the manager to dig through the ticket.

Max 150 words.

How to use: Fill in every field accurately — vague escalation reasons delay resolution; be specific about what decision is needed.

Generates a complete multi-email onboarding sequence designed to activate new customers and prevent early churn.

Write a [NUMBER]-email onboarding sequence for new customers of [PRODUCT_NAME], a [ONE_LINE_PRODUCT_DESCRIPTION].

The sequence should:
- Email 1 (send immediately): Welcome, confirm what they have access to, and give one 'quick win' action.
- Email 2 (Day [X]): Introduce the [KEY_FEATURE] and link to a help article or tutorial.
- Email 3 (Day [Y]): Address the most common beginner mistake: [COMMON_MISTAKE].
- Final email (Day [Z]): Invite them to [NEXT_STEP, e.g. join a webinar, upgrade, leave a review].

Each email: subject line, preview text, and body under 150 words.
Tone: [BRAND_VOICE]. Each email should end with a single, clear CTA button label.

How to use: Customize the day intervals based on your product's time-to-value; shorter for simple tools, longer for complex ones.

Converts confusing technical error messages into clear, actionable support replies for non-technical customers.

A customer has received the following technical error message and is confused:

---
[PASTE_ERROR_MESSAGE_HERE]
---

Product context: [PRODUCT_NAME] is a [BRIEF_DESCRIPTION].

Write a support reply that:
1. Translates what the error actually means in plain, non-technical language.
2. Explains the most likely cause (based only on the error provided — do not guess beyond it).
3. Gives 2–3 numbered steps the customer can try themselves.
4. Tells them exactly what information to gather before contacting support if the steps don't work.

Assume the customer has no technical background. Avoid all jargon. Max 200 words.

How to use: Paste the exact error string; the instruction to avoid guessing prevents fabricated technical explanations.

Builds a concise, unbiased post-support CSAT survey covering key satisfaction dimensions without leading questions.

Create a short post-interaction customer satisfaction survey for [COMPANY_NAME]'s support team.

The survey should:
- Take under 2 minutes to complete.
- Include exactly [NUMBER] questions.
- Cover: overall satisfaction, resolution effectiveness, agent friendliness, and ease of getting help.
- Use a mix of formats: [NUMBER_SCALE] scale questions, one yes/no, and one open-text.
- Be written in a warm, non-corporate tone.
- End with an optional open-text field asking for suggestions.

Also write a 2-sentence introductory message to appear at the top of the survey explaining why feedback matters.

Do not use leading questions or questions that assume a positive experience.

How to use: Specify the number of questions based on expected response rate — fewer questions yield higher completion.

Drafts a professionally neutral reply to threatening or legally charged customer messages with built-in legal-review flags.

A customer has sent the following message that contains threatening language or legal threats:

---
[PASTE_MESSAGE_HERE]
---

Write a support response that:
1. Remains calm, respectful, and professional — no matching their tone.
2. Acknowledges their frustration without validating any specific legal claim.
3. Does NOT admit fault, make promises about outcomes, or discuss legal matters.
4. Directs them to the appropriate channel: [ESCALATION_CONTACT, e.g. legal@company.com / a specific team].
5. Closes the immediate support interaction politely but firmly.

Important: Flag any sentence in your draft that a legal team should review before sending by marking it with [LEGAL REVIEW].

This reply is a draft — final approval rests with [REVIEWER_ROLE].

How to use: Always have a manager or legal contact review flagged sections before sending — this is a draft tool, not a final authority.

Responds to a feature request with transparency about its status while preserving customer enthusiasm and goodwill.

A customer has requested a feature that does not currently exist in [PRODUCT_NAME]:

Customer request: [DESCRIBE_REQUEST]

Current status of this feature: [NOT_PLANNED / ON_ROADMAP / UNDER_CONSIDERATION]

Write a reply that:
1. Thanks the customer genuinely for the suggestion.
2. Is honest about the current status without over-promising.
3. Explains briefly why this might be complex (only if you can do so honestly from context — do not fabricate technical reasons).
4. Offers a workaround if one exists: [WORKAROUND_OR_NONE].
5. Invites them to follow [FEATURE_REQUEST_CHANNEL] for updates if applicable.

Do not imply a timeline unless one is confirmed. Tone: [TONE]. Max 150 words.

How to use: Never invent a timeline or roadmap position — only reference what is confirmed to avoid expectation problems.

Produces a structured internal RCA document from raw support data, written for both technical and non-technical stakeholders.

Based on the following support data, write an internal root-cause analysis (RCA) document for a recurring customer issue:

Issue description: [DESCRIBE_ISSUE]
Number of affected customers reported: [NUMBER]
Time period: [DATE_RANGE]
Customer impact: [DESCRIBE_IMPACT]
Tickets or evidence: [PASTE_RELEVANT_EXCERPTS]
What is currently known about the cause: [KNOWN_INFORMATION]

Structure the RCA as follows:
1. Executive Summary (3 sentences)
2. Timeline of the issue
3. Root cause (based only on provided data — mark unknowns as 'TBD — requires investigation')
4. Customer impact assessment
5. Immediate remediation steps taken
6. Long-term prevention recommendations
7. Open questions for engineering/product

Audience: internal team, including non-technical stakeholders.

How to use: Mark unknowns explicitly — do not let the AI fill gaps with invented technical explanations.

Simplifies a complex help article to a target reading level while preserving all instructional accuracy.

Rewrite the following help article so it is accessible to someone reading at approximately a [GRADE_LEVEL, e.g. 6th grade / 8th grade] level:

---
[PASTE_ARTICLE_HERE]
---

Rules:
- Keep all the factual information — do not remove steps or simplify to the point of inaccuracy.
- Replace every word with 3+ syllables with a simpler alternative where possible.
- Break any sentence over 20 words into two sentences.
- Replace passive voice with active voice.
- Keep numbered steps numbered.
- Do not add any information not present in the original.

After the rewrite, list the 5 most significant changes you made and why.

How to use: Specify the grade level or describe your audience — 'plain English' is too vague; 'sixth grade' gives the AI a clear benchmark.

Drafts a transparent, accountable customer notification for a data or privacy incident with a built-in legal review reminder.

Draft a customer-facing apology email regarding the following privacy or data incident:

What happened: [DESCRIBE_INCIDENT — be factual, do not speculate]
Data potentially affected: [TYPE_OF_DATA]
Customers affected: [DESCRIBE_SEGMENT]
When it occurred: [DATE_OR_RANGE]
What has been done to address it: [STEPS_TAKEN]
What customers should do: [CUSTOMER_ACTION_REQUIRED_OR_NONE]

The email must:
1. Open with a direct, non-buried apology.
2. State the facts clearly without using vague corporate language.
3. Explain what you are doing to prevent recurrence.
4. Tell customers exactly what action (if any) they need to take.
5. Provide a dedicated contact for questions: [CONTACT].

Do NOT downplay the incident or use language like 'minor' or 'insignificant'. This draft must be reviewed by legal before sending — note this at the top.

Tone: serious, transparent, accountable.

How to use: Only include confirmed facts — privacy incident communications that speculate can create additional legal exposure.

Generates specific, evidence-based coaching notes from a real support conversation for use in agent 1:1s.

You are a customer support quality analyst. Review the following support conversation between an agent and a customer:

---
[PASTE_CONVERSATION_HERE]
---

Provide structured coaching notes for the agent ([AGENT_NAME]) covering:

1. **What they did well** (cite specific moments from the conversation)
2. **Where they could improve** (cite specific moments, be constructive not critical)
3. **Missed opportunities** (e.g., moments where they could have de-escalated, upsold, or offered proactive help)
4. **One specific technique to practice** for next time, with an example phrasing
5. **Overall rating**: [SCALE, e.g. 1–5] with a one-sentence justification

Tone: encouraging and developmental, not punitive. This is for a 1:1 coaching session, not a performance review.

How to use: Use verbatim conversation transcripts for accurate citations; paraphrased summaries reduce specificity.