McKinsey Slide Examples: What Their Public Decks Teach
The McKinsey slide examples worth studying are the ones the firm publishes itself. That corpus teaches exhibit craft extremely well and deck craft not at all, and the reason for the gap is the most useful thing here.
Every document this article points to is something McKinsey published on purpose. The firm runs a large open research operation and puts that work on the open web for anyone to read. That published library is the entire corpus we work from here. What you will not find on this page, or anywhere else outside the firm, is client work. The decks built for an actual engagement do not circulate, are not published later, and are not what these public reports are drawn from.
What is public here, and what never is
The materials linked below are published by McKinsey & Company on its own public website. Firms publish for reasons unconnected to client delivery: to make an argument in public, to attract people who want to work on those problems, and to contribute research meant to be read widely. That output is written for publication from the first draft, then reviewed, cleared, and stripped of anything client-specific long before it goes out. Even so, the terms are not uniform: some files a firm posts carry a restrictive notice of their own, so every link below points at a landing page rather than a download.
Engagement work is a different category entirely. Client data, the findings drawn from it, the recommendations, and the deck that carries them stay inside the client relationship permanently, under obligations that do not expire when the project ends. There is no public version, no redacted version, and no waiting period after which they appear. Anyone who offers you a real client deck is either misrepresenting what they have or has broken an obligation to get it.
This is a professional standard, not a legal footnote. Discretion is part of the job in consulting, banking and corporate strategy. Clients hand over their numbers on the understanding that the numbers stay put, and treating that seriously is what makes the relationship possible. So this article studies how public work is constructed. It does not reach behind that line, and neither should you.
Independent and unaffiliated. The skills, prompts, and frameworks referenced on this page are not created by, endorsed by, or affiliated with any consulting or financial-services firm whose methods may be referenced here. They are built on publicly available frameworks and ways of working, and are inspired by how such approaches are used to tackle complex business problems.
Oria One is not affiliated with, endorsed by, or a partner of McKinsey & Company. The firm is referenced here only to discuss research it has published publicly, and all such material remains the property of its author. We link to the publisher's own pages and do not host, mirror or reproduce any of it.
What McKinsey slide examples actually teach
Public McKinsey slide examples teach one half of the craft extremely well. Open any recent flagship and you can study exhibit construction in detail: a title that states a finding, one message per exhibit, units printed on the chart, a source line on the page rather than in an appendix. That layer is visible, repeated across years and industries, and free.
The other half is missing for a reason. You never see the storyline that orders pages into an argument, the build order, the speaker notes, or the cutting that turns a long analysis into a short deck. That layer only exists inside client work, which is confidential. The asymmetry is not a gap in the research. It is the shape of what is safe to publish.
So use the corpus narrowly: as a reference library for exhibits, while building the storyline layer from method rather than imitation. Our guide to the pyramid principle and the walkthrough on storylining a deck top down cover the part the public documents cannot show you.
Visible in public work
- Exhibit titles written as complete sentences that state a finding, not a subject.
- One message per exhibit, with the surrounding page kept deliberately thin.
- A source line, sample size and fielding window carried on the chart itself.
- Horizontal logic: read the exhibit left to right and you have read the point.
- The same argument shipped at two lengths, so you can see what survived the cut.
Never visible, anywhere
- The storyline: the governing thought, the decomposition beneath it, the order of the pages.
- Build order, which is how a page is revealed in a room rather than read on a screen.
- Speaker notes, the ghost deck, the red-team markup and the pages that were killed.
- The cutting itself. You see 224 pages and 64, never the argument about what to drop.
- Source files. No public template, no chart library, no colour token set.
Which public documents are really decks
Most writing about McKinsey presentation examples never establishes whether the documents are slides at all. Page geometry gets you most of the way. A page of 612 by 792 points is US Letter portrait, a document you read. A page wider than it is tall was laid out to be projected.
Several flagships are landscape rather than portrait, deck-shaped even where the writing runs report-length. The State of Fashion 2026 runs 143 landscape pages. The global balance sheet 2026 runs 42. The clearest case is Voice of the global farmer 2024: 20 landscape pages, one survey finding on each, almost no connecting prose.
Set that landscape survey beside the portrait State of AI 2026, 31 pages of US Letter, and the conventions are identical: a title that states a finding, one message per page, the source line on the chart. The conventions are the thing worth learning, not the shape of the page.
There is no published template file anywhere, but there is repetition, which teaches the same thing. The Technology Trends Outlook 2025 gives each trend an identical four page module across 108 pages, and the Global Private Markets Report 2026 splits its infrastructure chapter into a separate twelve page document. Read two modules and the master slide underneath is obvious. The methods hub collects the page types worth standardising first.
Action titles, and how to verify the rule yourself
The single best teaching artefact in the public corpus is the Global Banking Annual Review 2026, and you do not have to open the PDF to use it. The exhibit titles are carried on the landing page as image alt attributes, verbatim, so the whole title layer reads in the page source. Fourteen exhibits, fourteen complete sentences, zero noun phrases.
Exhibit titles as published, Global Banking Annual Review 2026
- "Margins are declining, which could pressure banks to further reduce costs."
- "Global banks have receded, while continental banks have gained."
- "Neobanks are achieving significantly faster growth in both scale and profitability than traditional players."
- "Banks should adopt a three-speed organizational approach, matching the innovation type with the right organizational setup."
Each title names a subject, asserts a movement, and carries a consequence. "Bank margins" would be a label. "Margins are declining, which could pressure banks to further reduce costs" is a claim you can agree with or dispute. That difference is the whole action title rule, enforced at production level rather than recommended in a style guide.
The test on your own pages takes five seconds. Cover the body of the slide with your hand. If the title alone still tells the reader what is true, it is an action title. If you have to look at the chart to learn what the page claims, it is a topic label wearing a verb. Most corporate decks fail on more than half their pages.
A title layer only reads as an argument if the pages underneath were grouped properly, which is a structuring problem rather than a writing one. If your titles are sharp but do not add up, the fix is upstream: check the grouping against the MECE test, rebuild the decomposition as an issue tree, and start from a stated hypothesis using hypothesis-driven problem solving.
One message per exhibit, built as a format
Advice to put one idea on a slide is everywhere and changes nobody's behaviour. What is unusual here is that the constraint became recurring formats.
The Five Fifty is a web-native slide format whose name is its editorial promise: five minutes of scroll, then fifty minutes of depth. Each panel carries one number or one claim and nothing else. Chart of the Week reduces the discipline to its atom and publishes it weekly: one exhibit, one title that states a finding, one source line.
The pattern survives at document scale. The 2025 Global Payments Report carries a full annual argument in fifteen pages, the tightest page-to-message ratio of any flagship in the set. It does not get there by having less to say. It gets there by refusing a second idea onto a page.
Source lines on every exhibit, method kept separate
The State of AI 2026 is a 31 page survey chart pack and the clearest public demonstration of source discipline. Almost every page is one exhibit plus one sentence, with the sample size and method footnote restated on each chart rather than collected at the back.
That repetition is worth copying for an unglamorous reason: slides travel alone. A page gets forwarded by itself, and a caveat that lived only on the first page of a run is now lost. The same habit runs through the quarterly Economic conditions outlook, which runs a near-identical exhibit set each wave with the sample and fielding window restated every time.
Method itself is quarantined rather than crammed in. The health of nations ships its 39 page appendix as a separate document, so the 58 page main report never carries the derivation. That separation is what lets a title be blunt: the qualification lives below the chart, so the headline does not hedge. A hedged action title has already collapsed back into a label.
Distillation, published as a diff
The most transferable habit in the corpus is that the compressed version ships as its own artefact, finished to the same standard as the long one, often enough to be a policy rather than a courtesy. The net-zero transition exists as a 224 page full report and a 64 page executive summary. The health of nations goes further still: 58 pages of research, three pages of summary.
Read a summary and its full report side by side and you are looking at a decision about what mattered, made by people who had all of it. That is the closest the corpus comes to showing judgement rather than output, and the best warm-up for writing your own opening page, which our guide to the executive summary slide walks through as a method.
What the pair never shows is the argument that produced it. You see 224 pages and 64, never the meeting where somebody defended a page that got cut.
Which exhibit type, and why
The lesson here is selection logic, not a catalogue of exotic chart types. The most common exhibit across the flagships is a plain labelled column chart, and the discipline sits in the title, the labelling and the emphasis.
Reach for a bridge when the question is why a number changed, so the delta breaks into signed drivers rather than two bars. The economic potential of generative AI builds its 2.6 to 4.4 trillion dollar value pool in layers rather than asserting it, which is the same move. Our guides to the waterfall chart and the synergy bridge cover the mechanics.
Reach for a left-to-right layout when the argument itself has an order. The global balance sheet 2026 lays its bridges out horizontally across 42 landscape pages, so reading the chart in reading order is reading the argument. For the other recurring page types we have walkthroughs on market sizing, scenario analysis, heat maps and the competitive landscape.
One honest negative. Reviewing these documents did not produce a verified example of a true Marimekko, the variable-width stacked column that circulating advice treats as a house signature. It may well be in there. We will not assert a chart type we did not see, and neither should a page you learn from.
Where the public corpus goes quiet
Everything above is exhibit craft. Push one level up, to how a set of pages becomes an argument, and the public material stops answering.
The reports are reports. You do not see a governing thought decomposed into branches, a storyboard where consecutive titles read as one continuous argument, or a page whose only job is to rule an option out. The Five Fifty is the closest thing to a public storyline, and it is deliberately shallow.
There is no presenter layer either. No speaker notes, no build order, no sense of what gets said while a page is on screen. These documents are made to be read, not walked through, which matters because a working deck is built for a room.
And there is no process. No drafts, no ghost deck, no red-team markup, no record of the forty pages killed to reach the twelve that shipped. The 224 page and 64 page pair implies that discipline exists. It never shows the cutting.
The gap is the point
The part that is published is the part that is safe to publish. What is withheld, the storyline built for one client, the recommendation, the argument about what to cut, only exists inside an engagement, and an engagement is confidential. That is not a limitation of the research library. It is what the confidentiality obligation looks like from outside, and it is why no amount of searching will turn up the deck you are imagining.
That leaves a clear division of labour. Take the exhibit craft from the public work. Build the storyline layer from method and your own evidence. The frameworks field guide and the strategy on a page walkthrough are the places to start on that half.
Seven McKinsey deck examples, and what to study in each
Seven documents rather than a hundred links, each published by the firm on its own site, each with a note on what to look at. Checked 8 September 2026. If a link has drifted, search the title on the firm's own site rather than reaching for a re-upload.
Fourteen exhibits, every title a complete declarative sentence, carried verbatim in the landing page image alt attributes. Start here if you have ten minutes.
A recurring web-native format built on a hard constraint: a five minute scroll, then a fifty minute deep dive. One number or one claim per panel.
Published twice in the same house style: a 224 page full report and a 64 page executive summary. Read the summary, then find what the full version dropped.
Twenty landscape pages, one survey finding each, almost no connecting prose. The shortest example of every page earning its place with a single message.
A 31 page survey chart pack. Almost every page is one exhibit plus one sentence, with the sample size and method footnote restated on each chart.
Forty two landscape pages of balance-sheet bridges laid out left to right. The best public illustration of horizontal exhibit logic.
Sixty eight pages whose headline number, a 2.6 to 4.4 trillion dollar value pool, is built up layer by layer rather than asserted.
Two hubs are worth a bookmark. The McKinsey Global Institute research index and the Global Surveys collection index the rest, and chart packs surface there first. One caveat about the Quarterly: full issue PDFs and the compendium collections sit behind a free membership sign-in, so treat them as gated rather than open downloads.
Three exercises get more out of that list than reading will. Read only the exhibit titles of the Banking Annual Review, in order, write the argument in three sentences, then run the same test on your own last deck. Read the 64 page net-zero summary, then find what the 224 page version dropped. Open two consecutive waves of the Economic conditions outlook and list what stayed identical, which is the template you cannot download but can reconstruct.
Common mistakes when studying public consulting decks
Treating a published report as a client deck
These documents are addressed to an industry, not a board. A recommendation deck for one company has a different job, length and tone. Copy the exhibit craft, not the genre.
Copying the surface instead of the function
A colon in the title, a particular grey. Those are house style. The function underneath is what transfers: a skimming reader should always know what a page claims and which branch of the argument it sits in.
Downloading from an aggregator
Search results are full of sites offering hundreds of files, some of which are plainly client work that escaped. A document carrying a confidentiality stamp is not made public by being uploaded somewhere.
Reusing an exhibit instead of rebuilding it
Published freely is not licensed for reuse. Studying how a chart is built is ordinary practice. Lifting the chart, the layout or the numbers into your own deck is republication.
Borrowing the name
Describing your own output by reference to a firm you have no relationship with turns a research habit into a claim. Study the work. Do not imply the relationship.
A related trap sits on the production side. Generic slide generators copy the surface of this style, the dense grid and the confident title, without the sourcing or the single message that makes it defensible, which is the failure mode we take apart in why AI slide tools produce weak slides.
A note on where this leaves you
What you can learn from published research is the craft: how an argument gets built, how a finding gets carried by a chart, how a page earns its place. That is the transferable part and it is fully available in public material. What you cannot learn from it, because it is not there, is what any specific client was told. Build your own work from your own data and your own analysis. The structure is the thing worth studying. The contents of someone else's engagement are not yours to study, and were never on offer.
Frequently asked questions
Are real McKinsey client decks available anywhere?
No. Client data, the findings drawn from it, the recommendations and the deck that carries them stay inside the client relationship permanently, under obligations that do not expire when the project ends. There is no public version and no waiting period after which one appears.
What are the best public McKinsey slide examples to start with?
Start with the Global Banking Annual Review, whose exhibit titles appear on the landing page as complete sentences. Then open Voice of the global farmer 2024, twenty landscape pages carrying one survey finding each.
Do McKinsey exhibits always use action titles?
In the public flagships the pattern is consistent. The Global Banking Annual Review 2026 carries fourteen exhibits and every title is a full declarative sentence. The test: cover the chart with your hand, and if the title alone still tells you what is true, it is an action title.
Is there an official McKinsey slide template you can download?
No. The public corpus is PDFs and web pages, with no published PowerPoint file, chart library or colour tokens. Anyone selling McKinsey style slides is selling a look, not the firm's file. What you can observe are conventions, seen most clearly where a document repeats a module.
Can I reuse a slide from a public McKinsey report?
No. Copyright survives free publication, and reproducing an exhibit in your own deck is republication rather than citation. Study the structure, rebuild it with your own evidence, and link the original if a reader needs to check it.
Why do so few consulting presentations exist in public at all?
Because publishing is a separate activity from client delivery. Published research is written for publication from the outset. Engagement work is written for one client under confidentiality and never enters the publication pipeline.
What does the public material fail to teach?
Deck architecture. You can study one message per exhibit, action titles and sourcing in detail. You cannot study the storyline that orders pages into an argument, the build order, the speaker notes, or the cutting. That layer only exists inside client work.