Read the trigger before the draft
A draft comes with the reason it exists: the card that raised the work and the purpose of the step it was written for. Reading the draft first anchors you to its wording instead of to whether it should exist.
Review a drafted customer message against the trigger that produced it, know what to edit versus discard, and send the approved version yourself.
A customer response reaches you in one of two ways. A card in the queue, such as a renewal follow-up, carries the trigger (why the action was raised) and opens a blank panel when you choose to write; nothing is drafted there. Where the contact is enrolled in a messaging sequence, AI Message on the contact generates the draft for the current step, written against the sequence’s purpose, its referenced artifacts and the contact’s context. The review panel holds it with three controls: Save changes, Approve draft and Discard draft.
Nothing sends until you send it. Approving records that the content is right; it does not deliver or schedule anything. You copy the approved message, send it from your own email, and mark the step completed. That is worth restating because it defines what kind of task this is: not quality control on something already in flight, but the actual decision about whether this message exists. The order you read things in is the entire skill of this lesson.

The trigger is the reason the work exists: the card’s why-this-action line, and for a sequence draft the purpose of the step. Read those first. It is a small discipline with a real effect: opening the draft first anchors you on the wording, and you end up editing sentences in a message that should not have been drafted at all.
The trigger also tells you whether the system's reading of the situation matches yours. When it does not (the response is pitched at a concern the account resolved last month, or it treats a routine question as an escalation) the wording is irrelevant and the useful feedback is about the trigger.
Reviewing well is mostly a matter of asking things in the right sequence, because each question makes the next one cheaper. Should we be responding to this at all? Then: is this the right thing to say? Only then: is this the right way to say it?
Most reviewers start at the third question, since wording is what is in front of them and it is the easiest thing to have an opinion about. That is how a well-written response to the wrong situation gets approved, polished, on-brand, and answering something nobody asked.
Once you know what is wrong, the choice between editing and discarding is a question of what recurs. Tone, phrasing and a detail specific to this customer are faster to fix than to explain, so fix them, save the changes and approve. A wrong argument, a wrong trigger, or a claim that should not have been made is a defect that will show up again on the next account, and fixing it where it comes from is what stops it.
Discard properly when you do. Discard draft records the decision but carries no reason, so the reason has to travel another way: a note to whoever owns the sequence, or a change to the step’s purpose or the artifact library if that is yours to make. On a queue card, Ignore asks for one of five reasons, and choosing the right one is the same discipline. This is the campaign course's pattern-versus-instance distinction, arriving in a different queue.
A health card flags an account that has gone quiet, and the CSM generates the next step of a re-engagement sequence for the main contact. The draft is good: warm, specific, references the last QBR, asks a clear question. Reviewed on wording alone it sails through.
The card says it was raised because engagement dropped below a threshold. Reading that first changes the question. Engagement dropped because the champion has been on parental leave for six weeks, which is recorded on the contact and was not part of what the threshold looks at. The account has not gone quiet; one person has.
The right outcome here is not an edit. Editing the wording and sending it would put a well-phrased message in front of a covering stakeholder implying their team has gone silent, which is both wrong and slightly insulting. The right outcome is to discard it and pass the reason to whoever configures the agents, because the same threshold will fire again on the next account whose main contact is on leave, and the reason is the only thing that can prevent that.
This is the case that makes the trigger-first habit worth the two seconds. Read in the other order, nothing about the draft invites suspicion, and a fluent message papers over a wrong premise.
You should be able to answer each of these from memory before opening it. Recalling the answer is what makes it stick; recognizing it when you read it does not.
Should this response exist, is the substance right, is the wording right. Most reviewers start at the third because wording is what is in front of them, which is how a well-written response to the wrong situation gets approved.
Edit what is local: tone, phrasing, a detail specific to this customer. Discard what will recur: a wrong argument, a wrong trigger, a claim that should not have been made, and make sure the reason reaches whoever owns the framework, because Discard draft does not carry one.
A draft comes with the reason it exists: the card that raised the work and the purpose of the step it was written for. Reading the draft first anchors you to its wording instead of to whether it should exist.
A wording change is faster to make than to explain. A wrong argument is a defect worth reporting to whoever owns the framework, because it will recur.
Approval is a decision about the content. RevTech does not send or schedule the message; you send it from your own email.
Explore how agents prepare work across your revenue teams, and how people review it, guide it and decide what happens next. Fictional data. Real product.
Explore the demo