TL;DR: A quarterly business review (QBR) is a 30 to 45 minute meeting every three months where you walk a voice AI client through what their agent did, what it was worth to them, what you fixed, and what you plan to change next quarter. It is the single best retention habit I know for a voice AI agency, because most clients who cancel do not think the agent is bad. They just stop being sure it is doing anything. A good QBR uses the same numbers the client can already see in their portal, adds three or four real call examples, and ends with one clear recommendation, which is often an upgrade or a new use case.
I build production voice agents for US clients on Retell AI, n8n, GoHighLevel and Twilio, and we run VoiceDash, the white-label client portal for voice AI agencies. Monthly reports keep clients informed. QBRs are what keep them renewing. Here is the structure I use, the slides I build, the mistakes I made early on, and how to prepare one in under an hour.
What a QBR is for (and what it is not)
A QBR is not a status update. Your client already gets status in their dashboard and in any monthly summary you send. If the QBR just reads the dashboard out loud, it is a waste of their time.
A QBR has three jobs:
- Connect the agent to money. Calls answered is a nice number. Appointments booked, leads captured after hours, and staff hours saved are numbers an owner actually cares about.
- Show that you are actively managing the agent. Clients pay a retainer for ongoing work. If they never see that work, they start to wonder what they are paying for.
- Set up the next quarter. Every QBR should end with a decision: keep going as is, fix something, expand to a new call type, or change the plan.
If you only take one thing from this article: the QBR is a renewal conversation that happens long before the renewal date.
Who should be in the room
On the client side, you want the person who signs the check, not only the office manager who handles day to day questions. The office manager knows if the agent is annoying callers. The owner decides if it stays. If the owner never sees results, you are one budget review away from being cut.
On your side, it is usually just you for a small agency. If you have someone handling support for that client, bring them for the "what we fixed" section.
Keep it to 30 to 45 minutes on a video call. Shorter is fine for a small single-location client. Longer usually means you are explaining too much.
The QBR structure I use
I use the same six sections for every client. Having a fixed template means preparation is mostly pulling numbers, not deciding what to say.
1. The quarter in one sentence
Open with the headline. Something like: "Your agent answered every after hours call this quarter, and most of the bookings it made came in when the office was closed." Only say what the data supports. If the quarter was flat, say it was flat, then explain why and what you will change.
2. Volume and outcomes
This is the core slide. For the quarter, and compared to the previous quarter:
- Total calls handled by the agent
- Calls by outcome: booked, lead captured, question answered, transferred to staff, spam or wrong number
- After hours and weekend calls as a share of total
- Minutes used against minutes included in their plan
The outcome split is what matters. A client does not care that the agent took 900 calls. They care that a meaningful share of those turned into appointments or qualified leads. If you are not tagging outcomes yet, set up post call analysis first. I walk through it in Retell AI post call analysis.
3. What it was worth
Turn outcomes into a value estimate the client agrees with. I do not invent a number. I ask the client, ideally in the first QBR, what a booked appointment or a new lead is worth to them on average, and I use their number every quarter after that.
Then the math is simple: bookings times their value, plus an estimate of staff time saved on routine calls. Show the formula on the slide so the client can check it. If they push back on the value per booking, adjust it right there. A number they believe is worth more than a big number they doubt. My full method is in how to calculate voice AI ROI.
4. Real calls
Pick three or four calls and play short clips or show the transcript excerpts:
- One great call, ideally a booking that happened after hours
- One tricky call the agent handled well, like an upset caller or an unusual request
- One call that went wrong, and what you changed because of it
That third one feels risky. It is actually the most trust building part of the whole meeting. It proves you are reading transcripts, not just billing. I explain how I pick calls to review in voice AI call transcript review.
5. What we changed this quarter
A short list of every change you made to the agent: new FAQs, prompt fixes, updated hours, a new transfer rule, calendar changes. Include changes the client requested and changes you made on your own after spotting issues.
This is the slide that justifies the retainer. Most clients have no idea how much small maintenance a live agent needs. When they see a list of eight or ten fixes, the monthly fee stops looking like rent and starts looking like a service.
6. Recommendation for next quarter
End with one recommendation, not a menu. Common ones I make:
- Upgrade the plan because they are close to their minute bucket every month
- Add a new call type, like outbound appointment reminders or missed call follow up by SMS
- Expand coverage, for example routing daytime overflow calls to the agent, not only after hours
- Fix a process on their side, like a calendar that is always out of date or a transfer number nobody answers
If the recommendation includes more spend, connect it to the value slide. My approach to these conversations is in how to upsell voice AI clients.
How to prepare a QBR in under an hour
The first QBR for a client takes longer because you are building the template and agreeing on value numbers. After that, preparation should be quick if your data lives in one place.
My checklist:
- Pull the quarter's call volume, outcomes and minutes from the client's portal
- Pull the previous quarter's numbers for comparison
- Apply the client's agreed value per outcome
- Search transcripts for three or four example calls
- Copy the change log from your notes or ticket system
- Write one recommendation and the reason for it
- Send the deck or summary a day early so the owner can skim it
Step 1 is where most agencies lose time. If your client data is spread across a raw Retell account, a spreadsheet and a few screenshots, assembling it every quarter for every client becomes a full day. This is exactly why we built VoiceDash: each client has a branded portal with their call volume, outcomes, durations, recordings and searchable transcripts, scoped to only their agents. I prepare QBRs from the same view the client already uses, so the numbers on the slide always match what they see when they log in.
Mistakes I made with early QBRs
Too many numbers
My first QBRs had latency charts, token counts and per agent breakdowns. Owners glazed over. Technical metrics belong in your own monitoring, not in a client meeting. I cover what to keep where in voice AI agent monitoring.
Waiting for the client to ask
I used to schedule QBRs only when a client seemed unhappy. That made the meeting feel like damage control. Now the QBR is in the contract and on the calendar from onboarding, so it is routine, not a sign that something is wrong.
Hiding the bad quarter
When call volume drops, it is tempting to skip the meeting. Do not. A drop usually has a cause you can find: seasonality, a paused ad campaign, or call forwarding that got turned off on their phone system. Explaining it shows you are paying attention. Going quiet looks like you did not notice.
No decision at the end
A QBR that ends with "great, talk next quarter" wastes the best moment you have with the owner. Always leave with an action, even a small one.
QBRs vs monthly reports
You need both, and they do different jobs.
| Monthly report | Quarterly business review | |
|---|---|---|
| Format | Email summary or portal view | Live call with slides or a doc |
| Audience | Office manager, owner if they read it | Owner or decision maker |
| Focus | What happened | What it was worth and what is next |
| Time to prepare | Minutes, mostly automated | Under an hour per client |
| Outcome | Client stays informed | Client renews, expands, or flags problems early |
The monthly layer should be as automatic as possible. A live client portal does most of the work, so the client can check results any day without waiting for you. I go deeper on that layer in voice AI client reporting. The QBR is where you add the judgment and the story the dashboard cannot tell on its own.
Should small clients get a QBR?
Yes, but lighter. For a single location client on a lower plan, I send a one page written review each quarter with the same six sections and offer a 15 minute call. Most take the call the first time and skip it later, which is fine. The written review still lands in the owner's inbox and reminds them the agent is working.
For bigger clients, multi location businesses, or anyone on a higher retainer, the live meeting is not optional. Those are the accounts where losing one hurts most, and where expansion opportunities are biggest. If you are working on keeping clients longer overall, start with voice AI client retention.
The bottom line
The agents you build are only half the product. The other half is the client's confidence that the agent is working and worth the money. A QBR every three months, built from real outcomes, real calls and a clear next step, is the cheapest way I know to protect that confidence and grow the account.
If you want every client to have a branded portal with their calls, outcomes, recordings and transcripts in one place, so QBR prep is pulling up a screen instead of building a spreadsheet, start free on VoiceDash or book a demo and I will show you how a client portal goes live in under 10 minutes. VoiceDash works with Retell AI today, with VAPI and Bland support coming soon.