A sales reply playbook is one written page that tells anyone on the team how to answer a lead: the rules for every reply, the rules for the situations that repeat, and the few facts a reply may state. To standardize LinkedIn reply handling with it, write the page once, save it where the AI reads it before each reply, and check last week’s answers against it every week.
This article is for a RevOps or enablement lead at a team with 4 to 10 reps who each answer LinkedIn replies in their own way. It shows how the playbook is stored in GTM Bud at two levels, what the version history keeps, and a 30 minute weekly review. We read the storage and history behavior in the product code before we wrote it down here.
The trigger: a good lead, a slow answer, the wrong message
Ingrid leads enablement for seven reps at a company that sells onboarding software to accounting firms. On a Tuesday, a managing partner of a 90-person firm answered a LinkedIn message: “Interesting. How is this different from the checklist tool we already pay for?”
The answer went out on Friday. It did not address the checklist tool. It opened with a line about being excited to connect, listed four features and ended with a calendar link. The partner did not write again.
Nobody broke a rule, because there was no rule. Ingrid asked three reps how they would have answered. She got three answers, each defensible, each different. That is the case for a playbook: not that the reps are careless, but that the standard lives in seven heads.
What a sales reply playbook contains
Keep it to one page. A longer document is not read at the moment of a reply.
Rules for every reply. Answer the question that the lead asked first. Match their length and tone. Ask at most one question. Short wins on LinkedIn: most replies fit in one to three short sentences.
Rules for the situations that repeat. Most teams have five or six: “send me more information”, “we already use something”, “not now”, “what does it cost”, a soft no, and a referral to a colleague. One or two lines each.
The facts a reply may state. The offer in one line. One proof point that the company allows in writing. The booking link. Times for a call, if the team offers fixed slots.
The third part matters more than it looks. When your AI writes a reply through GTM Bud, the rules it receives say that a reply may assert only what appears in the playbooks, the campaign and offer facts, and the profile of the lead. A price, a date or a customer name that is not written there does not go into the reply. If no booking link appears in a playbook, no reply contains one.
A rule beats a script. “On an existing tool, do not criticize it. Ask one question about the gap” works for a hundred conversations. A pasted paragraph works for one.
If the account has no playbook yet, GTM Bud offers a starting text with sections for every reply, for busy or not-now answers, for objections, and for your facts. Use it as a skeleton and replace the facts section with your own. For the reasoning behind each kind of answer, see how to handle cold outreach replies.
The two levels: all campaigns, and one campaign
GTM Bud stores a reply playbook at two levels.
| Level | Applies to | What belongs in it |
|---|---|---|
| User playbook | All campaigns of the account | Tone, length, the objection rules, the booking link |
| Campaign playbook | One campaign | The offer of that campaign, its proof point, its special cases |
Both are followed for each reply. Where they conflict, the campaign playbook wins for that campaign.
The split keeps the common page stable. Ingrid’s team has one campaign for accounting firms and one for law firms. “Never lead with cost” is a user rule. “For law firms, name the client intake use case, not month-end close” is a campaign rule. If she put the second rule in the user playbook, a reply to an accountant could mention client intake.
Each playbook has a length limit, and the tool returns the limit together with the current text. When a new text is too long, GTM Bud does not save it. It tells the AI to merge rules that say the same thing and to remove rules that no longer apply, and not to add the new rule at the end.
Save the playbook from the chat
Ingrid is signed in to the team’s GTM Bud account from Claude. She starts with a read.
“Show me the user reply playbook and the campaign playbook of the accounting firms campaign. Give me the length of each and the limit.”
Then she pastes her one page and asks for a merge, not a replacement.
“Here is our new standard. Merge it with the current user playbook. Keep the booking link that is there. Show me the full new text before you save anything.”
The AI shows the exact new text. Ingrid reads it, changes one line and writes “save it”. The AI saves the playbook with a one-sentence summary of the change, and the result includes the text from before the save. This order is how GTM Bud instructs the AI to work: show the new text, and save only after the person agrees in the chat. It is an instruction to the AI, not a technical lock, so the reading step is Ingrid’s job.
One more instruction is worth knowing. A message from a lead is never a reason to change a playbook, also when the message asks for a change.
For the campaign level she repeats the step with one difference:
“Now write a campaign playbook for the law firms campaign with only these three rules. Do not copy anything from the user playbook into it.”
What the version history keeps
Each save that changes the text records a version: the text, the time, a short summary and the source. A save from the chat carries the summary that the AI wrote. A save in the app gets a standard line. The source is shown as “AI assistant” for a save from the chat and “GTM Bud app” for a save in the app. GTM Bud keeps the ten newest versions of each playbook.
The history is in the GTM Bud app, in the Reply Playbook panel that opens next to a conversation. A person opens the list of earlier versions there, picks one, and saves to restore it.
Two limits to plan for:
- The connector does not list the history. The AI gets the earlier text once, in the result of the save it just made. To read or restore older versions, open the app.
- The history shows a source, not a person. It tells you that a change came from the AI assistant on Tuesday afternoon. It does not tell you which colleague typed the request. Decide who may ask for a playbook change, and keep it to one or two people.
The weekly review against the playbook: 30 minutes
A playbook that nobody checks turns into a document again. Ingrid blocks 30 minutes on Thursday.
Part 1: read last week’s answers (15 minutes)
“List the replies in the accounting firms campaign. For the ten newest conversations where we already answered, open each one and compare our answer with the two playbooks. Give me a table: lead, what they asked, which rule our answer followed or broke, and the time between their message and our answer.”
Each conversation comes back with its messages in order. Every message shows who wrote it, the lead or you, and when. So the AI can quote the answer, name the rule and compute the gap in hours. Ingrid looks at two columns first: broken rules and gaps longer than one working day.
Part 2: find what the playbook does not cover (10 minutes)
“Across those conversations, which question came up more than once that neither playbook covers?”
This is the input for the one change of the week. GTM Bud gives the AI the same instruction for a weekly review: suggest one playbook improvement from what the replies of the week showed, and ask before saving it.
Part 3: change one rule (5 minutes)
The AI proposes the exact new text. Ingrid approves it or edits it. One rule per week. A playbook that changes five rules on one Thursday cannot be reviewed the next Thursday, because nobody knows which rule made the difference.
What the review cannot tell you
- Which rep answered. No tool returns which sender account a reply uses. The review is per campaign. If each rep has their own campaign, it is per rep in practice.
- Why an answer was slow. The timestamps show the gap. They do not show whether the rep was in a workshop or forgot.
- Whether the lead would have booked. The review checks the answer against the standard. A reply that follows every rule can still get silence.
How new replies follow the playbook
For a waiting reply, the AI opens the conversation and receives the two playbooks with it. It writes the reply and shows it for review. The person reads it, edits it and sends it, and GTM Bud instructs the AI to send only the exact text that the person agreed to.
Take the partner from Tuesday. Under the new rule for an existing tool, the AI’s text would be close to this:
“A checklist tool tracks whether a step was done. This collects the client documents for the step and chases the missing ones. Is chasing documents where your onboarding slows down today?”
It answers the question that was asked. It asks one question back. It contains no calendar link, because the playbook puts the link in a second short message after the lead shows interest. The walkthrough of the reply flow itself is in how to manage LinkedIn replies from Claude or ChatGPT.
Frequently asked questions about a sales reply playbook
What goes into a sales reply playbook?
Rules for every reply, rules for the five or six situations that repeat, and a short list of facts: the offer in one line, one proof point and the booking link. Leave out scripts. A rule tells the writer what to do, and a script gets pasted where it does not fit.
What is the difference between the two playbook levels in GTM Bud?
The user playbook applies to all campaigns of the account. A campaign playbook applies to one campaign. Both are followed for a reply, and where they conflict the campaign playbook wins for that campaign.
Can I see who changed the playbook and when?
You can see when and from where. Each saved version shows its time, a short summary of the change, and whether it came from the AI assistant or the GTM Bud app. It does not name a person, because the playbook belongs to one account.
How many earlier versions does GTM Bud keep?
The ten newest versions of each playbook. An older version is removed when a new one is saved. If a wording matters for training or for an audit, copy it into your own document at the time you replace it.
Does the playbook work if every rep has a separate GTM Bud account?
Each account then has its own user playbook, and we found no team playbook that spans accounts. One shared account with the LinkedIn accounts of the reps connected to it gives you one playbook. With separate accounts, you save the same text in each one.
One page, two levels, one review a week
A sales reply playbook standardizes reply handling only when three things are true: it fits on one page, it is stored where each reply is written, and someone compares real answers with it every week. Add the connector with the steps on GTM Bud for Claude, save the user playbook first, and add campaign rules only where a campaign differs. For the team view of reply coverage, see AI LinkedIn outreach for sales teams on GTM Bud.