How to Prepare a Quarterly Business Review Using AI

What should a client QBR actually prove?

Before you build a quarterly business review using AI, be clear what it has to prove: that you kept the promises you made last quarter, that those results were worth something to the client’s business, and that both sides agree on what happens next. Anything that doesn’t serve one of those three belongs in an appendix or an email.

A quarterly business review (QBR) is a structured meeting, held every three months, where a supplier or partner shows a client the outcomes delivered, the value those outcomes created, and the plan for the next quarter. The audience is the client’s sponsor, not your own team.

The Kept-Promise Review in brief

The Kept-Promise Review approach we teach at PromptPeakAI builds every QBR section from one goal you agreed with the client last quarter, traced through four columns:

1. Promised: the goal, exactly as you both agreed it.
2. Delivered: the verified result, calculated from source data, never estimated in prose.
3. Worth: what that result means in the client’s own business terms.
4. Next: the promise for next quarter, with an owner on each side.

The loop is the point. When this quarter’s Next becomes next quarter’s Promised, the review builds on itself, and preparing the following QBR starts from a list that already exists.

QBR, MBR and renewal meetings

People blur these three meetings together, and the deck suffers for it. If your review is internal and variance-heavy, you want a monthly business review instead. The QBR faces outward.

Gather the quarter’s evidence before you prompt

The quality of a quarterly business review using AI is capped by the evidence you hand it. Spend 20 minutes building an evidence pack before you open any AI tool, and the rest of the process gets faster and safer.

The five inputs that matter

Most of this already exists if you send regular updates. A quarter of weekly client updates is a ready-made wins and incidents log, and you can turn call transcripts into minutes to capture what the client said, in their words.

Strip client identifiers first

A quarterly business review using AI means uploading client data, so decide what goes in before you upload anything. Your client contract matters as much as your AI plan’s settings.

A find-and-replace pass that turns the client’s name into [Client] takes two minutes, and you swap the real names back in the final deck.

How to prepare a quarterly business review using AI

To prepare a quarterly business review using AI, start from the goals you agreed last quarter, compute every result against target in a spreadsheet or code, and only then ask the AI to write each section in the client’s terms. Draft the executive summary last, then check every figure against the source before the deck goes out.

Open last quarter’s deck and list every goal, commitment and action item you agreed. This is the Promised column. Skipping it is the most common reason a quarterly business review using AI feels like a highlight reel: the wins get a slide each, and the missed commitments quietly vanish. Client sponsors notice.

You are helping me prepare a client quarterly business review.

Attached is the QBR deck (or plan) I presented to [Client] last quarter.

List every goal, commitment and action item we agreed, one per line, in this format:
Promise | Who owned it (us or client) | Target or due date | Exact wording from the source

Rules:
- Only include items that appear in the attached file. Do not add or infer any.
- If a promise has no measurable target, write "no target set" in that column.
- After the list, flag any promise that is vague enough to argue about.

Upload the results export itself, never a screenshot of it. OpenAI’s help page on data analysis warns that ChatGPT may not reliably extract exact values from image-based tables or scanned files. Ask the tool to run the calculations in code and show you the table, so you can analyse the results export line by line before any prose exists.

Attached is our results export for [Client] for [quarter]. It has one row per metric with columns: Metric, Target, Actual, Previous quarter.

Using code, not estimation:
1. Calculate the gap to target and the change from last quarter for every metric.
2. Mark each metric as Met, Missed, or No target.
3. Show the result as a table with the columns: Metric | Target | Actual | Gap | Change vs last quarter | Status.

Do not round any figure beyond one decimal place. Do not write any commentary yet. If any value is missing or looks inconsistent, list it separately instead of guessing.

Now the AI writes, one promise at a time, from the verified table only. This is where a quarterly business review using AI most often goes wrong, because a model asked to write freely fills gaps with plausible numbers. Drafting section by section keeps each claim traceable to a row. Paste the computed table and your wins and incidents log into the same chat.

You are writing one section of a client quarterly business review for [Client]'s sponsor, [sponsor role]. Their business goals this year are: [goal 1], [goal 2].

Use ONLY the figures in the table below and the notes below. Do not calculate, round or introduce any number that is not in the table.

Write this section in four short parts:
Promised: the goal as agreed, in one sentence.
Delivered: the result against target, stated plainly, including if we missed it.
Worth: what this result means for the client's business goals, in their terms. If the notes don't support a business impact, write "impact not yet evidenced" instead of inventing one.
Next: one proposed promise for next quarter, with a suggested owner on our side and on theirs.

Table:
[paste the computed row]

Notes:
[paste the relevant wins, incidents and client feedback]

Once every section exists, ask for a four-sentence summary: the headline result, the one miss that matters, the value in one line, and the single decision you need from the client. Writing it first tempts the AI to set a tone the evidence then has to match.

Turn outcomes into value the client recognises

Value is the Delivered result restated in terms of the client’s own goals: cost, revenue, risk, time or customer experience. If you can’t connect a result to one of those, it is activity, and activity belongs in an appendix.

Tie every result to their goal

This is the part of a quarterly business review using AI where your judgment matters most. Before the meeting, ask the sponsor, or check last quarter’s notes, for the two or three business goals they’re measured on this year. Put those goals into every Worth prompt. If your metrics don’t map to any of them, that’s a sign to agree the right KPIs in the Next column, not a sign to stretch the story.

In practice, most account managers find one or two results carry the whole meeting. Give those a slide each and group the rest into a single scorecard.

When you missed a target

Put the miss in the deck, near the top, in plain words. A sponsor who finds it on page nine stops trusting page one. State the gap, the cause you can evidence, and the fix you’re proposing, then let the AI tighten the wording without softening it. For the email that goes out before the meeting, the same approach you’d use to deliver bad news works well.

• Do: “CSAT finished at 83% against an 85% target. Most of the drop followed the returns policy change in August.”
• Don’t: “CSAT remained broadly stable in a challenging quarter.”

Build a next-quarter plan both sides own

The Next column is where most QBRs fall apart: a list of things your team will do, with no one on the client side accountable for anything. A plan both sides own names an owner at each company, a date, and how you’ll both know it worked.

Owners on both sides

Ask the AI to propose the plan, then edit the owners yourself: it can’t know who in the client’s organisation actually has the authority. The structure of a 90-day plan transfers well here, since a quarter is 90 days, and drafting it is where a quarterly business review using AI saves the most time.

Based on the Kept-Promise sections below, draft a next-quarter plan for [Client] as a table with these columns:
Next promise | Our owner (role) | Their owner (role) | Due | How we'll both know it worked

Rules:
- Every missed target from this quarter must appear as a promise with a fix.
- Every promise needs an owner on both sides. If the client-side owner isn't clear, write "confirm with sponsor".
- Keep to 3 to 5 promises. Do not include sales, upgrade or renewal items.

Sections:
[paste the drafted sections]

Keep the renewal pitch out

A QBR that ends with an upsell teaches the client that every review is a sales call, and attendance drops. If there’s a genuine commercial conversation to have, book it separately.

Worked example: one account, four AI tools

Here is a quarterly business review using AI on one illustrative account. A customer-support outsourcing team is preparing its Q3 review for a homeware retailer. The figures below are invented for this example, but the structure is exactly what you’d build.

The evidence pack is identical whichever tool you use. What changes is where the deck gets built.

ChatGPT and Claude routes

Both work well for steps 2 and 3. Run Prompts 2 and 3 in the same chat so the computed table stays in context, then ask for a slide-by-slide outline with speaker notes.

• ChatGPT: the free plan includes data analysis, but with limits, so a paid plan is smoother for a full quarter’s export. Where ChatGPT Work is available to you, @Presentations can build a Google Slides deck; PowerPoint wasn’t included at launch.
• Claude: turn on “Code execution and file creation” under Settings > Capabilities (it’s available on every plan, Free included), and Claude can hand you a .pptx you then restyle in your company template.

Copilot in PowerPoint route

If your company runs on Microsoft 365, write the four-part sections in a Word document first, using proper headings, then build the deck from it. Microsoft’s Copilot in PowerPoint FAQ confirms it can turn a Word file into a presentation, and work accounts can use PDFs as well. On the home version, you reference the Word document by typing / in the Copilot box. Start from your company template so the branding is right before Copilot adds a single slide.

Gemini in Google Slides route

In a blank Google Slides presentation, open Gemini, describe the QBR structure, and use “Add from Drive” to attach the Doc holding your drafted sections. Gemini proposes an outline you review before it builds the slides. Generating a full presentation is desktop-only and English-only for now. Our guide to Gemini in Google Slides covers the side panel in more detail.

Whichever route you take, the deck is a support act. Use AI to write speaker notes for the deck, then rehearse the Worth lines out loud: they’re what the sponsor will repeat to their boss.

Fix a QBR draft that falls flat

When a quarterly business review using AI falls flat, it usually fails in one of five predictable ways, and each has a quick fix that doesn’t mean starting over.

The five-minute number check

Before anything goes to the client, run this check yourself. It’s the step AI can’t do for you.

1. Tick every number in the deck against the computed table from step 2.
2. Tick every number in that table against the original export.
3. Confirm each missed target appears in both the deck and the next-quarter plan.
4. Search the deck for words like “approximately” and “around” and replace them with the real figure.
5. Read the executive summary aloud, as if you were the sponsor.

Compare the QBR draft below with the verified results table below.

List every number, percentage or claim in the draft that does not appear in the table, or that differs from it in any way, in this format:
Claim in draft | What the table says | Match / Mismatch / Not in table

Do not fix anything. Do not summarise. Only list the checks.

Draft:
[paste the deck text or outline]

Table:
[paste the computed table]

What AI can’t do in a client QBR

A quarterly business review using AI still needs a person in charge. AI can compute, draft and format; it can’t know your client, sense a relationship at risk, or decide what to say in the room when the sponsor pushes back.

Client data and your contract

Your AI plan’s data settings are only half the question. Many client contracts restrict where their data can be processed, and some ban third-party AI tools outright. Check the contract before your first quarterly business review using AI, not the night before the meeting. If you’re unsure how your own account handles uploads, start with whether ChatGPT is safe to use at work on your plan.

Should you tell the client?

If you prepared the quarterly business review using AI and it touched their data, the honest default is yes, briefly: “We use AI to draft our reviews; every figure is checked against source by our team.” Most sponsors care far more that the numbers are right than how the slides were made. For a fuller way to decide, see when to tell clients you use AI.

Run the Kept-Promise Review once and keep the files: last quarter’s Next becomes this quarter’s Promised, and your next quarterly business review using AI starts half-built.