How to run a QBR in the age of AI

What a QBR is actually for

A quarterly business review is a scheduled conversation between you and a customer about whether they are getting what they paid for. That is the whole job. It is not a status report, it is not a relationship-building exercise, and it is emphatically not an upsell meeting wearing a lanyard.

You can tell a QBR worked by a simple test: did anything change as a result? A decision, an escalation, a commitment with a name and a date against it. If the only artefact is a deck sitting in someone's downloads folder, you held a webinar for one.

Most QBRs fail the test for a structural reason rather than a human one. The meeting is built around a document that was finished before the meeting started.

The problem with the deck

Think about what a QBR deck actually is. It is a snapshot of your data taken at the moment the account manager had time to build it, usually a week or two before the call. Between the export and the meeting, the numbers moved. Nobody mentions this.

Then the customer asks the obvious follow-up. What does that look like for just the EMEA team? What if we added the twenty seats we discussed? Which of those users are actually the ones who stopped logging in?

And the answer, every time, is a version of: let me pull that and come back to you.

That sentence is where the meeting quietly dies. You had the customer's attention and a live question, which is the rarest combination in account management, and you spent it scheduling homework. The follow-up email lands four days later, by which point the customer has moved on and the moment has gone.

Play with the second tab for a moment. Nothing there is technically impressive, and that is rather the point. It is the same four numbers. The difference is that a question from the customer produces an answer instead of an action item.

Why this is suddenly worth fixing

QBR decks exist because they used to be the cheapest option. Pulling usage data, reconciling it against the contract, building charts, and writing the narrative was genuinely a week of somebody's life. Once you have spent that week, you are certainly not rebuilding it live in front of a customer.

That constraint is gone. The assembly work, which is to say the export, the joins, the formatting, and the first draft of the commentary, is now close to free. What remains scarce is judgment: knowing which number matters for this account, and which question to ask about it.

So the honest summary of what AI changed about QBRs is not that it writes your slides. It is that the reason your QBR was static no longer applies, and most teams have not updated the ritual to match.

The 45-minute agenda

The most common failure is spending forty minutes on the past and five on the future, then running out of time exactly when the useful part begins. Invert it.

Note what is missing. There is no company update, no roadmap tour, and no slide about your funding round. The customer did not take an hour out of their quarter to hear about you.

The five things worth showing

1. Adoption against what they bought. Not raw logins. Seats used against seats paid for, and which teams are missing. This is the number that predicts renewal and the number most decks bury.
2. Outcomes in their metric, not yours. If they bought your tool to cut onboarding time, show onboarding time. Tickets closed and time in app are your metrics, and nobody renews on them.
3. Open issues with an owner and a date. Including yours. Especially yours.
4. The forward plan. Two or three specific things, each with a named human. A plan with no names is a wish.
5. The renewal position, said out loud. Date, amount, what would have to be true. If you cannot say it in the QBR, you have a problem that is not going to improve with silence.

What "live" actually requires

Live does not mean a business intelligence rollout or a new platform. In practice it is one screen, pulling from the systems you already have, that answers the four or five questions customers reliably ask. Usage by team. Value in their unit. Open issues. What changes if the contract changes.

The reason this rarely exists is not technical difficulty. It is that the tool is too specific to your accounts to buy off the shelf and, historically, too small to justify an engineering quarter. That gap is exactly where custom GTM tooling has become sensible, and it is the same build-versus-buy calculation we walk through in build vs buy.

One practical warning. A live view is only an asset if the underlying numbers are trustworthy. Putting a real-time dashboard in front of a customer while your health score is guesswork will expose that faster than any deck would. Get the definitions right first, then make it live.

Start with one account

Do not roll this out. Take the next QBR on your calendar, build the one screen that answers the questions that account always asks, and run the meeting from it. You will know inside forty-five minutes whether it changed the conversation, and you will have a much better brief for the second one.