The conventions behind McKinsey-style slide decks — action titles, the pyramid principle, horizontal logic, ghost decks — and a practical way to build them yourself.
Anyone who has taken a deck into a partner review knows the feeling: the numbers are right, the analysis is sound, and the pages still come back covered in ink. The problem is usually not the content. McKinsey slide decks — and the wider MBB house style shared by Bain and BCG — follow a set of conventions that make an argument land in seconds, and most decks quietly ignore all of them.
None of these conventions are secret. They come from a handful of ideas that have been written down and taught for decades — most famously in Barbara Minto's The Pyramid Principle, which she developed while at McKinsey. This piece breaks down what actually makes those decks distinctive, then walks through how to build one in the same style yourself.
What makes McKinsey slide decks distinctive
Strip away the logo and the fonts, and five things separate an MBB-style page from an ordinary corporate one.
Action titles. Every slide title is a full-sentence conclusion, not a topic label. Not "Revenue by segment" but "Enterprise now drives 60% of revenue but is growing slower than SMB." The title states the takeaway; the body of the slide is only there to prove it. Read the titles of a good deck in sequence and you have the whole argument without opening a single chart.
The pyramid principle. The deck leads with the answer, then supports it. At the top sits one governing thought. Beneath it sit a small number of mutually exclusive, collectively exhaustive (MECE) supporting arguments. Beneath those sits the data. You never make the reader assemble the conclusion themselves — you state it first and spend the rest of the deck defending it. Executives read this way because they may only see the first three pages before a decision.
Horizontal and vertical logic. Two tests run at once. Vertical logic: everything on a slide supports that slide's title and nothing else. Horizontal logic: the action titles, read top to bottom across the deck, form a coherent story on their own. If the titles alone don't argue the case, the deck doesn't hold together — no chart will save it.
One idea per slide. A McKinsey-style slide makes a single point, usually with a single exhibit. The page can be dense — issue trees, waterfalls, 2×2 matrices, Marimekko charts — but it is never cluttered. Density and clutter are different things: density is a lot of information in service of one point; clutter is several points fighting for attention.
Restraint. No gradients, no 3D icons, no stock photography, no "AI look." One accent colour, a consistent grid, and a source line on every page that carries data. The quiet look is deliberate. It signals that the thinking, not the decoration, is doing the work.
Start with the storyline, not the slides — the ghost deck
The single biggest difference in how these decks get made is that the design comes last. Before anyone builds a real slide, the team writes a ghost deck (also called a storyboard, skeleton, or wireframe): a full set of empty pages where each one has only its action title and a rough sketch of the exhibit that will go on it — a scribbled bar chart, a box labelled "issue tree", a note that says "table of three options."
The ghost deck forces the argument to exist before any effort goes into polishing it. You can read the titles straight through and ask the only question that matters: does this actually make the case? Rearranging a fifteen-page storyline is five minutes of work. Rearranging fifteen finished slides is an afternoon. Building the argument first, in title form, is what keeps MBB decks tight — and it's the step most people skip.
A worked example: horizontal logic
Say you are advising a software company on pricing. Built McKinsey-style, the action titles alone might read:
- "Our list prices sit below three of our four closest competitors."
- "Discounting adds a further margin gap, concentrated in deals closed at quarter-end."
- "Customers in two segments show low price sensitivity in win/loss data."
- "A tiered increase in those segments recovers margin with limited churn risk."
- "We recommend phasing the change over two quarters, starting with new logos."
Notice that you understood the recommendation without seeing a single chart. Each title is a claim; each slide behind it exists only to prove that claim; and the five together move from problem to evidence to recommendation to plan. That is horizontal logic working. The numbers above are illustrative — the point is the shape of the argument, not the figures.
Build one yourself: a working sequence
You do not need to have worked at an MBB firm to apply the method. A practical order of operations:
- Write the governing thought first. In one sentence, what is the single answer the deck delivers? If you cannot write it, you are not ready to build slides.
- Frame the setup with SCQA. Situation (what everyone agrees on), Complication (what changed), Question (what that forces us to decide), Answer (your governing thought). This becomes the executive summary.
- Draft the storyline as action titles. Write eight to fifteen full-sentence titles that argue from problem to recommendation. This is your ghost deck. Do not open a chart yet.
- Pressure-test horizontal logic. Read the titles alone. Do they flow? Are the supporting points MECE — no overlaps, no gaps? Rewrite here, where it is cheap.
- Make each slide vertically consistent. One idea, one exhibit, and everything on the page supports its own title. Cut anything that supports a different point; it belongs on a different slide.
- Design last, and quietly. One accent colour, a consistent grid, a "so what" callout on each chart, and a source line on every data page.
Do this in that order and a deck built in PowerPoint by one person can carry the same logical spine as one from a studio.
Where AutoPresent fits
The method above is judgment work — no tool writes your governing thought for you. What a tool can do is remove the mechanical friction so you spend your time on the storyline instead of aligning boxes.
AutoPresent generates fully editable PowerPoint (not a web-only export you screenshot), so the action-title-and-exhibit structure lands as real, movable slides you can restructure the moment the storyline changes. You can turn a diligence dump, a Word report, or a financial model into a first draft of slides, then rewrite the titles into proper conclusions and reorder pages as your ghost deck evolves. Its smart templates cover the dense, structured layouts MBB decks lean on — driver trees, waterfalls, staged recommendations — so density stays clean instead of turning into clutter, and the theme switcher rebrands a whole deck to a client's colours in one step.
For work that has to genuinely pass as MBB-grade, AutoPresent also offers Expert Deliverables, where former MBB consultants build or review the deck on demand. That is the honest limit of any AI: it can draft structure and layout quickly, but the horizontal logic and the "so what" still benefit from someone who has defended decks in front of a partner. Having that human option is the difference between a deck that looks the part and one that survives the review.
The takeaway
A McKinsey-style deck is not a visual style you buy — it is an argument you build. Action titles carry the story, the pyramid puts the answer first, horizontal logic holds the whole thing together, and the ghost deck forces all of that to happen before you spend an hour on any single slide. Get the storyline right in title form and the design is the easy part. Whatever tool you use, that is the order the good decks are made in.