How to Monitor Your Team’s WhatsApp Calls

Here is a question worth sitting with for a minute: how do you know what is said on your agents' WhatsApp calls?

Not what you hope is said. Not what agents tell you was said. What was actually said, in the calls where your customers decided to buy, to wait, or to leave.

If you manage a sales or support team that works on WhatsApp, the honest answer for many teams is some version of "I don't, not really." That is not a failure of management. It is a side effect of how WhatsApp-first teams are built: every agent has their own number, every call happens on their own account, and nothing is recorded by default.

This article walks through the ways managers usually try to answer that question, why each one falls short, and what a real answer looks like. It also shows how Qualitify approaches it, because this is the exact problem it was built for.

Why the question matters more every year

WhatsApp is not a niche channel. Meta's CEO said in April 2025 that WhatsApp has more than 3 billion monthly active users (Newsquawk). The WhatsApp Business app passed 200 million monthly active users in June 2023, up from 50 million in 2020 (TechCrunch).

Customers are also asking for it. In a Kantar survey commissioned by Meta (11,056 adults in 22 markets, fieldwork April to September 2025), 73.3% said they prefer messaging when contacting a business, and 69% said waiting on hold is a waste of time (WhatsApp Business). One caveat: that survey asked consumers, not companies, and Meta commissioned it.

So more of your revenue conversations happen on WhatsApp, and customers want it that way. The question is whether you can see them.

The five usual answers, and why they fall short

Most managers answer the question in one of five ways. Each has something going for it. None is enough on its own.

1. "I ask my agents."

It is quick and it builds a relationship, and it is worth doing. But it relies entirely on the agent's own account of their own performance. Even honest agents remember the good calls better than the bad ones, and no agent can tell you what they would have done differently if they could not see the mistake in the first place.

2. "I check the CRM notes."

Notes tell you what the agent chose to write down. They rarely capture tone, hesitation, a promise made in passing, or the moment a lead lost interest. They are a summary written by the person being evaluated, and they are only as complete as that person's habits.

3. "I wait for complaints."

This one is not a method, but it is how many teams actually find out. By the time a customer complains, the damage is done, and you have learned about one bad call out of hundreds. You also learn nothing about the agents who are doing fine, or the ones quietly losing deals without anyone complaining.

4. "I get agents to screen-record their calls."

This is the common workaround. It sometimes works. But screen recordings often capture only one side of the conversation, or no audio at all, depending on the device and settings. They depend on the agent remembering to start the recording every time, and you cannot tell whether a missing recording means the call was not recorded or the agent chose not to record it. Files end up scattered across personal phones with no list, no order, and no way to review them fairly.

5. "I listen live, or ask for call-backs."

Shadowing gives real information, but it does not scale, and agents behave differently when the boss is listening.

The pattern is the same across all five: they depend on someone choosing what to report, and none gives you a consistent, fair view of the whole team.

What a real answer needs

If you wanted to answer the question properly, you would need five things. Call centers have had all five for decades. WhatsApp teams usually have none.

  1. A record. The calls have to exist somewhere you can reach, captured without relying on agents to remember.
  2. A sample. Nobody can listen to every call, so you need a fair way to choose which ones to review, one that does not only land on the easy ones.
  3. A standard. You need to define what a good call looks like for your business, so a score means something.
  4. Specific feedback. Comments have to point to a moment in a call, not a general impression.
  5. A way to check it worked. Coaching without measurement is just conversation. You need to see whether scores improved afterwards, and how many reviews that conclusion rests on.

That list is the core of quality assurance in a call center, and it is exactly what our first post describes missing when our founder moved from a call center to a team that worked entirely on WhatsApp.

How Qualitify answers the question

Qualitify gives WhatsApp teams those five things, without asking agents to change how they work. Each agent keeps their own number and their own customers. The how it works page covers the five stages, and here is the short version.

A record, captured automatically. A desktop app on the agent's Windows computer records WhatsApp calls in the background and uploads the audio when the call ends. Every call appears in one list: who took it, how long it ran, whether it was incoming or outgoing, whether the audio is ready, and where it stands in review. If audio is missing, the list says so, instead of leaving you to guess.

A fair sample. Qualitify picks calls for review so that coverage spreads across your team. It also shows review coverage per person, so you can see who is under-reviewed. The reasoning on the site is blunt: a team that only reviews the easy calls learns nothing.

Your own standard. You set the criteria, the weights, and the mistakes that fail a call outright, such as skipping a required disclosure or promising something operations cannot deliver. If you are starting from scratch, the free call QA scorecard builder has templates for WhatsApp sales, WhatsApp support, forex, collections and general calls, and it needs no account.

Specific feedback. Reviewers scrub through a waveform, change the playback speed, and pin notes to the exact second they refer to. Drafts save automatically, so a long review never costs anyone their progress.

Proof that coaching works. A coaching plan names the area, the supervisor who owns it, and the follow-up date. Later, the dashboard compares reviews before and after the plan, side by side, with the number of reviews behind each figure.

Honest numbers: the detail that makes the answer trustworthy

Two design choices in Qualitify keep the answer honest.

First, an unreviewed call is never counted as a zero against an agent. It is a gap in coverage, and it is reported as one. Calls with no audio, such as unanswered ones, skip the review queue but still count in the figures.

Second, every average shows how many reviews it came from. A score built from five reviews is not presented as if it were as reliable as one built from forty. This protects agents from being judged on a thin sample, and it protects you from making decisions on one.

If you have ever wondered how many calls you actually need to review to trust a score, the free call review sample size calculator works it out. You enter calls per agent per month, number of agents, confidence level, margin of error and minutes per review, and it returns reviews per agent, total reviews and reviewer hours per month.

An example: what changes in one month

This is a fictional scenario for illustration, not a real customer.

Imagine a company with ten sales agents, each with their own WhatsApp number. The manager has always known the monthly numbers: deals closed, revenue, follow-ups logged. What she has not known is how those deals were won.

In the first week, she sets up a scorecard with five criteria: opening and rapport, understanding the customer's need, accuracy of information, handling objections, and closing with a clear next step. She gives accuracy double weight and marks one mistake as a critical failure.

By the end of the month, a few things are visible that were not before:

  • Two agents score well on rapport but lose points on accuracy, because they repeat an outdated price. The deals still closed, so nobody had noticed.
  • The agent with the strongest closing always asks about the customer's deadline before quoting. Her calls become the training material for the team.
  • One agent has far fewer reviews than the others. The coverage view shows it, so the manager fixes the sample rather than drawing conclusions from three calls.

She attaches coaching plans for the two agents. A month later, one has improved across a healthy number of reviews, and the other has too few to judge, so she schedules more instead of declaring success. None of this required listening to every call.

The objections worth taking seriously

"Will my agents hate being recorded?"

They might, if recording is introduced as surveillance. It works better when it is introduced as the way the company decides who gets coached, who gets recognized, and what good looks like. In Qualitify, agents see the plan written for them, a supervisor's private notes stay private, and agents can replay their own calls only if the owner turns that setting on. Honest scoring, with unreviewed calls never held against anyone, also helps agents trust the process.

"What about privacy?"

Audio is deleted automatically 30 days after the call, while scores, written feedback and notes are kept. Each company sees only its own data, enforced on the server, and recordings are served through short-lived signed links. Recordings are never used to train AI models.

"What about consent?"

This one stays with you. Asking the people you call for consent, and telling your agents that calls are recorded, is your responsibility, and rules vary by country and industry. Nothing in this article is legal advice. The FAQ covers consent and recording in more detail.

"Is there anything it does not do?"

Yes. The recording app currently runs on 64-bit Windows 10 and 11, so if your agents call only from phones, check that it fits how your team works before you start. And it is a quality assurance tool, not a CRM, so it does not replace the system where you keep customer contacts.

Five questions to ask your team today

Even before you choose any tool, you can test how well you can answer the opening question. Try these:

  1. Can I name three specific things my best agent does on calls that others do not?
  2. Can I name one thing each agent should improve, based on a call I actually heard?
  3. If a customer complained today, could I play back the call?
  4. Do I know how many calls each agent made yesterday, and in which direction?
  5. If my top agent left tomorrow, would their way of handling customers stay with the company?

If most answers are "no", you have found the gap. The good news is that it is a fixable gap, and the fix is old and well understood: record, review, coach, and measure.

Where to start

You do not need to commit to anything to see whether this fits your team:

  1. Look first. See the live demo, or book a demonstration and go through it on a shared screen.
  2. Define good. Use the free call QA scorecard builder to write down what a good call looks like for your business.
  3. Size your review program. Use the free call review sample size calculator to find out how many calls you need.
  4. Try it on real calls. Start the free trial: 14 days, 100 calls, no card required. See pricing for your team size.

So, back to the question: how do you know what is said on your agents' WhatsApp calls? The sooner you can answer it with evidence instead of hope, the sooner you can coach your team on what is really happening, and the sooner your best calls can become everyone's standard.

Sources

Qualitify is an independent product. It is not affiliated with, endorsed by or sponsored by WhatsApp or Meta Platforms.