How to build a McKinsey-style presentation (storyline first)

Start with the question and a hypothesis

The style begins before any slide. A team first agrees on the exact question the deck must answer, written as one sentence the client would sign off: “Should we consolidate our two European distribution centers, and if so, where and when?” A vague question (“Review our European logistics”) produces a vague deck.

The team then writes down its best guess at the answer on day one. This is the hypothesis-driven approach: rather than gathering all the data and seeing what emerges, you state what you think is true and design the work to prove or disprove it. The hypothesis is usually broken into an issue tree of the sub-questions that would have to be answered “yes” for it to hold. Those sub-questions become the key line of the storyline, structured by the rules in our guide to the Pyramid Principle.

A hypothesis is not a conclusion. Its job is to focus the analysis, and it is expected to change.

Build the ghost deck as a workplan

Next comes the ghost deck: every page of the eventual presentation, each with a draft action title and a sketch of the evidence it needs. Our guide to action titles covers how to write those headlines. In a consulting-style process the ghost deck does a second job: it becomes the team's workplan.

• For each ghost page, note the chart or table that would prove the title, the data source, who will produce it and by when.
• Mark the pages whose titles are the riskiest assumptions, and do that analysis first.
• Review the ghost deck with the client or sponsor early. Disagreement on a headline is far cheaper on day five than in the final meeting.
• As analysis comes in, rewrite titles to match what the data shows, and cut pages whose claims did not survive.

By the end, the ghost deck and the final deck are the same document. Nobody builds a chart that has no page to go on.

Write the executive summary in dot-dash form

The executive summary is the deck in one page, and it is often written and rewritten more than any other slide. A common format for it is the dot-dash structure: each “dot” is a bolded key line statement in a full sentence, and beneath it sit two or three “dashes”, shorter sub-points that give the evidence. Read the dots alone and you have the argument; read the dashes and you have the proof.

• Consolidating both distribution centers into Venlo saves $6.8M a year and pays back in 16 months
• – Running one site removes $4.1M of duplicated overhead and $2.7M of cross-border transfers
• – One-off costs are $9M, mostly severance and racking
• Service levels can be held through the move
• – Venlo reaches 92% of customers within 24 hours, the same as today
• – Eight weeks of dual running covers the cutover
• The main risks are manageable if we act on them now
• – Early works council consultation avoids the main delay
• – A retention bonus keeps key staff through the transition

Keep the summary to one page, three or four dots, and no charts. Each dot should match, word for word or nearly so, the title of the section divider it leads to.

Navigation: trackers, dividers and sticker labels

Long decks need wayfinding, and the consulting style handles it with a few small, consistent devices.

• Section dividers that restate the dot from the executive summary, so the reader knows which claim the next pages prove.
• A tracker on every content slide: a small label or a miniature agenda in the top corner with the current section highlighted.
• Sticker labels: small boxed tags, usually in the top right corner, that qualify the status of a slide. Common ones are “Preliminary”, “Draft for discussion”, “Illustrative”, “Example”, “Estimate” and “Not exhaustive”. They warn a reader before they quote a number out of context.
• Slide numbers on every page, so a question can be pointed at “page 14” in a meeting or a call.

Sources, footnotes and chart conventions

The bottom of a consulting slide carries its credibility. A typical order is numbered footnotes for specific figures, then a “Note:” line for definitions and assumptions, then a “Source:” line naming the data and the year. “Source: Eurostat 2025; company shipment data FY2024–2025; team analysis” is enough.

• Chart titles state the measure and the unit: “Annual transfer costs, $M, 2021–2025”.
• Sample sizes appear wherever survey data does: “n = 412”.
• Numbers are rounded consistently across the deck, and totals add up.
• One accent color marks the client, the recommended option or the item the title talks about; everything else is gray.
• Titles use sentence case and run to two lines or fewer; body text is left-aligned.
• No animation, decorative images or 3D charts. The deck must work printed or as a PDF.

One slide, one message

The discipline that holds the style together is that each slide makes exactly one point, stated in its title and proved by its body. Two quick tests apply to every page.

• The vertical test: does everything on the slide support the title, and is everything in the title supported on the slide?
• The horizontal test: read the titles in sequence. Does each one follow from the last and lead to the next?

When a slide fails the vertical test, it usually holds two messages and should be split, or holds an extra chart that belongs in the appendix. When it fails the horizontal test, the storyline has a gap or a detour.

The recommendation slide

The recommendation slide is where the deck asks for a decision, and it has a recognizable anatomy. A strong one fits on a single page and contains five parts.

1. Recommendation: the action, stated as the title. “Consolidate both European distribution centers into Venlo by Q3 2027.”
2. Rationale: the two or three reasons, echoing the dots of the executive summary.
3. Impact: the quantified result: $6.8M a year in savings, $9M one-off cost, payback in 16 months, service levels unchanged.
4. Risks and mitigations: the two or three that matter, each with its response. Service disruption during cutover, mitigated by eight weeks of dual running; labor consultation delays, mitigated by starting the works council process in January.
5. Next steps: the decision needed today, then three to five actions with an owner and a date. “Approve consolidation today; site lease signed by March (Operations); works council consultation opens in January (HR); cutover plan to the steering committee in April (Program office).”

Lay it out as a left column for the recommendation and rationale and a right column for impact and risks, with next steps across the bottom. The title should be the decision itself, not “Recommendation”.

Common mistakes

• Copying the visual template without the process. Gray charts and sentence titles do not make an argument.
• Treating the hypothesis as the answer and ignoring the data that contradicts it.
• An executive summary that lists findings instead of making a case.
• Dots in the summary that do not match the section dividers.
• Missing sticker labels on preliminary numbers, which then get quoted as final.
• A recommendation slide with no impact numbers, no risks or no owners.

McKinsey-style deck checklist

• Is the question written as one sentence the sponsor agrees with?
• Did a ghost deck exist before the analysis, and does the final deck match it?
• Does the executive summary make the case in three or four dots, each with evidence?
• Does every section divider repeat its dot?
• Does every content slide have a tracker, a source line and, where needed, a sticker?
• Does every slide pass the vertical and horizontal tests?
• Does the recommendation slide show the action, rationale, impact, risks and next steps?