تجاوز إلى المحتوى
جرّب المرجع الدورة المحفوظات
العربية
← كل playbooks دعم العملاء

playbook

Build a canned-response library you'll actually reuse

Turn the questions you answer most into warm, on-brand template replies with `[placeholders]` an agent can personalize in ten seconds — speed without sounding like a robot.

متوسّط ~45 min

متى تلجأ إلى هذا

Your team retypes the same answers all day — the refund explainer, the "where's my order" reply, the reset-your-password walkthrough — and every agent writes a slightly different, slightly worse version each time. Most canned-response libraries fix the speed and kill the warmth: customers can feel the form letter instantly. This system turns your highest-volume questions into templates that still sound human, with [placeholders] for the parts only a person should fill — so an agent grabs one, spends ten seconds personalizing it, and the reply lands like it was written for that customer.

جهّز هذا أولًا

  • Your top 8–12 repeat questions — pull them from a ticket export or just list them from memory, plus one good real reply you've sent for each.
  • Your voice in a few lines (or a voice.md if you have one): how you say sorry, how formal you are, phrases you'd never use.
  • The hard facts each macro needs — refund window, shipping times, what you can and can't offer — so the templates state policy correctly, not from a guess.

الـ workflow

  1. Confirm the list and the voice before drafting

    Start by having Claude read your example replies back as voice rules. Get the voice right on one before you mass-produce twelve in the wrong tone.

    أنت تطلب 
    Here are 6 real replies I'm proud of. In 4 bullets, describe our support voice — how we apologize, sentence length, formality, and 2 phrases we'd never use. Then list the 10 questions you'd turn into templates first, ranked by how often they likely come up. Don't write any templates yet.

    ما تحصل عليه A short voice read-back — "we apologize plainly in the first line, short sentences, never say 'per our policy'" — and a ranked shortlist. Fix the voice description now if it's off, before it shapes a dozen templates.

    This "read our voice back to me first" step is what keeps the whole library from sounding subtly wrong in twelve places at once.

  2. Draft one template, perfect it, then scale

    Nail the structure on a single high-volume macro — where the placeholders go, where the warmth lives — before generating the set, so every template inherits a good pattern instead of a mediocre one.

    أنت تطلب 
    Write ONE template for the 'where is my order' question in our voice. It must have a warm human opener, a clear middle, and a next step — with [name], [order number], and [ship date] as placeholders. Mark in a comment which lines an agent should personalize and which are safe to send as-is.

    ما تحصل عليه A single clean template — warm opener, the answer, a next step — with placeholders only where a human detail belongs, and a note on which line the agent should make their own.

  3. Generate the full set from that pattern

    Now batch the rest against the approved structure. Asking for them all in one pass keeps them consistent in tone instead of drifting template to template.

    أنت تطلب 
    Using that same structure and voice, write templates for the other 9 questions on the list. Each: warm opener, clear answer, next step, [placeholders] only where a personal detail or a fact I must confirm belongs. State the refund window as [refund window] rather than guessing a number. Keep each under 130 words.

    ما تحصل عليه A set of 9 more templates, consistent in tone and shape, every customer-specific or policy detail left as a labeled placeholder rather than invented.

  4. Assemble the library and a one-line index

    Pull it into one artifact an agent can actually search mid-ticket. A library nobody can find fast is a library nobody uses.

    أنت تطلب 
    Assemble all 10 into a single macros.md: each template under a clear heading, a one-line 'use this when…' note above each, and a list of every placeholder it contains. Add a short index at the top so an agent can find the right macro in 5 seconds.

    ما تحصل عليه A paste-ready macros.md — indexed, each macro labeled with when to use it and which placeholders to fill — the file your team opens beside the queue.

    Save this file. It becomes the starting point for the *Draft a reply that sounds like a person* playbook, and you refresh it as new repeat questions emerge.

اجعله ملكك

  • **Feed it from real data:** instead of listing questions from memory, run the *Find what the queue is really about* playbook first and build templates for the top clusters by volume — you'll cover the most tickets per template.
  • **Promote the winners:** when a macro answers a question that keeps coming back anyway, the real fix is the *Build a help center from the questions you actually get* playbook — a template treats the symptom, an article treats the cause.
  • **Make them one keystroke:** load macros.md as a reusable skill or a set of /macro custom commands (see the Playbook's *Features* tab) so the right template is always one command away instead of a copy-paste hunt.

انتبه إلى

  • A template is a starting line, not a send button — every reply still needs a human to fill the [placeholders] and confirm the offer; never let a macro go out with brackets still in it.
  • Never bake a customer's real name, account number, or a specific refund figure into a template — those stay [placeholders], both for privacy and so an agent can't ship a stale or wrong number.
  • Claude drafts the templates; a human owns whether the policy in them is current. Re-read the refund-window and shipping-time lines whenever a policy changes, because a confident outdated fact in a macro spreads to every customer who gets it.

ستحصل في النهاية على An indexed `macros.md` of warm, on-voice templates for your highest-volume questions — each personalized in ten seconds because a human still fills the part that should sound human.