Sales Chat Audit Checklist: How to Review a WhatsApp Sales Conversation

Decide what the audit is for

Before you start, ask one question: what decision will we make based on the result? Coaching a rep, changing a sales script and assessing a new hire’s probation period are different goals, and each needs a different depth of review.

The second point: sales and support conversations should not be judged by the same criteria. In a sales chat, the main question is whether the customer moved to a next step. In a support chat, it is whether the problem was solved. Score a price enquiry and a warranty complaint with the same sheet and the result will be unfair to both.

• Sales chat: need, matching offer, objection, next step.
• Support chat: understanding the problem, resolution, customer confirmation.
• Mixed chat (a question after purchase): score each part with its own criteria.

Audit criteria and scoring rules

Use three scores and one special status for each criterion: 0 — missing, 1 — partial, 2 — complete. If the chat does not show enough to judge, mark it “unknown” and leave it out of the total. A score means nothing on its own. It always needs the message that supports it.

1. Need and matching offerDid the rep ask what the customer is buying for, for whom and in what situation? Does the offer follow from the answer? 2: at least one clarifying question was asked and the offer is based on it. 0: a price was sent without any question.
2. Answer accuracyWere all of the customer’s questions answered? Are the numbers, timings and terms correct? 2: every question has a concrete answer. 1: some questions were left open. To check accuracy, you need the current price list and terms at hand.
3. Objection handlingIf the customer raised a doubt or objection, did the rep find out what was behind it? 2: the reason was explored and the reply addresses it. 0: the objection was ignored or a discount was offered straight away. If there was no objection, mark it “not applicable”.
4. Next stepDid the conversation end with a concrete agreement: a visit, payment, measurement, call or decision date? 2: a step and a date. 1: a step without a date. 0: it ended with “OK, let us know”.

Scoring one conversation together

The dialogue below is an illustrative example. A customer writes to an online appliance store: “Hi, do you have an 8 kg washing machine? How much is it?” Two different reps answer the same question.

• Weak version. Rep: “Yes, 450.” Customer: “That’s a bit expensive.” Rep: “No discounts, sorry.” The customer does not reply.
• Strong version. Rep: “Yes, we have two models. How many people are in your household, and where will it go? Depth matters.” Customer: “Four of us, and the bathroom is small.” Rep: “Then the 45 cm deep model fits. It’s 450 with installation included.” Customer: “That’s a bit expensive.” Rep: “I understand. What price did you have in mind? There is a 7 kg option that costs less, but for four people you would run it more often.” Customer: “OK, I’ll take the 8 kg.” Rep: “Shall we deliver tomorrow between 2 and 6 pm?”

1. Need and matching offerWeak: 0. Evidence: “Yes, 450” — no question asked. Strong: 2. Evidence: “How many people are in your household, and where will it go?” — the offer follows the answer.
2. Answer accuracyWeak: 2. Evidence: both questions were answered. Strong: 2. Evidence: price and installation stated. Note that the weak version does not fall behind here. The problem lies elsewhere.
3. Objection handlingWeak: 0. Evidence: “No discounts, sorry” — the reason was never explored. Strong: 2. Evidence: “What price did you have in mind?” plus an honest explanation of the alternative.
4. Next stepWeak: 0. Evidence: the chat ended with silence. Strong: 2. Evidence: “Shall we deliver tomorrow between 2 and 6 pm?” — a specific time.

Result: the weak version scores 2 out of 8, the strong one 8 out of 8. More useful than the number is seeing where the gap comes from: accuracy is the same, and the difference is in asking questions and proposing a next step. Those are the two points to show the rep.

Turning the result into a coaching task

An audit is only worth as much as the change it produces in the rep’s work tomorrow. Instead of a long list of mistakes, pick one concrete fix.

1. Pick one weak criterionChoose the criterion with the lowest score that repeats most often. In the example above, that is “need”: the rep sends a price without asking anything.
2. Describe a specific behaviourNot “communicate better with customers” but “ask at least one clarifying question before sending a price”. The task must be measurable.
3. Re-checkA week later, review 5–10 new chats from the same rep against that one criterion. If the score has improved, move to the next criterion. If not, discuss why with the rep. Maybe customers ask for the price very abruptly and the script needs to change.

Auditing one conversation and comparing a whole team are different jobs: for the second, see WhatsApp sales team metrics. In support chats the key criterion is resolution, so checking whether a reply resolved the problem is a separate method. If a chat changed hands, also check what was lost in the customer handover.

How to check your own chats with Chatdue

Manual audits work for a handful of chats a day. When there are hundreds, even choosing which conversations deserve a review gets hard. Chatdue connects to your company’s WhatsApp number and produces a summary, sales stage, objections, sentiment and an agent score for each chat. Your company chooses the analytics fields and can create new ones for its own criteria.

Team members are added with roles, and each person sees only what they are allowed to. You can export filtered chats with all AI fields to Excel and work in your own audit sheet. The agent score is useful for picking chats to review, but give the final score after reading the actual messages.

Frequently asked questions

1. Can every conversation be audited with the same criteria?Sales conversations, yes. Support and complaint chats need their own criteria. Also adjust each criterion to the type of sale: a “next step” for a one-off purchase is a delivery slot, while in B2B it is a meeting or an approval date.
2. How should I score when information is missing?Never guess a score. If the conversation moved to a call or part of it happened on another channel, mark that criterion “unknown” and leave it out of the total. Otherwise the rep gets penalised for work you simply could not see.