Action titles
"Volume by quarter." Versus: "Volume grew 8% in Q3, driven entirely by Europe." The first is a label on a chart; the second is a claim the chart has to defend. The whole discipline is in that one swap, repeated over every slide of the deck. Read the titles top-to-bottom and you should get the executive summary free, no slides opened. If you can't, the deck doesn't have a story — it has a table of contents.
Where it comes from
Action titles are the slide-level application of the Pyramid Principle. Barbara Minto wrote the principle inside McKinsey's London office through the late 1960s and published the canonical text in 1973; the slide-craft that travelled with it was codified by Gene Zelazny, who joined McKinsey in 1961 and spent more than fifty years there as the firm's director of visual communications, teaching consultants in every office how a slide should think. He died in April 2023, by which point his rules had been copied into the house style of every strategy firm on earth.
His two books are the route by which McKinsey's internal manuals reached the public. Say It with Charts (McGraw-Hill, 1985) is the chart-selection bible — column versus bar versus line versus pie, picked by the message the slide carries, not by the data's shape. Say It with Presentations (McGraw-Hill, 2000) extends the discipline from chart to deck. Both treat the title as the load-bearing element of the slide: chart, table and bullets are evidence; the title is the claim they exist to defend. Get the title wrong and a flawless chart still fails.
The discipline didn't stay at McKinsey. By the late 1980s every top-tier firm — BCG, Bain, Booz, the Big Four advisory arms — had absorbed it as house style. McKinsey calls them "action titles"; BCG, "message titles"; the looser corporate term is "headlines". Same thing. The McKinsey internal practice that seeded all three was the ghost deck: before any chart is built, write the deck as titles only — twelve sentences on twelve slides — and circulate that for review. If the ghost deck doesn't read as the argument, no amount of charting will save the real one.
The deeper logic is Minto's. A pyramid says every idea at every level must summarise the ideas beneath it. Apply the rule to a deck and the conclusion is inescapable: every slide title must be a sub-claim of the deck's governing thought, written as a sentence, because only a sentence can carry a claim. "Q3 results" is not a claim about anything; "Q3 results beat guidance on margin, missed on volume" is.
What it actually is
An action title has three properties, all mandatory:
- It contains a verb. Grew, declined, drove, delivered, reached, narrowed, widened, recovered, lost. The verb is what turns a label into a claim. "Pricing strategy" has no verb; "Pricing changes drove 60% of margin recovery" does.
- It states an insight, not a topic. The reader should be able to take the title alone — without seeing the chart — and know what the slide is for. "Customer segmentation" forces the reader to read the slide to find the point. "Three segments account for 78% of profit" lands the point first; the chart then becomes evidence, not riddle.
- It ladders up. The title is a sub-claim of the deck's governing thought. Cover the body of every slide and read only the titles in order: that sequence should reconstruct the executive summary. This is the action-title test and it is the single most useful diagnostic in deck design.
The action-title test is brutal in practice. Most decks fail on first attempt — the titles, read in sequence, produce a list of topics ("Market overview… Competitor positioning… Our capabilities… Recommendation") rather than a story ("The market is consolidating around two players… we are #4 and losing share… our cost base lets us play offence on price… acquire #5 and use the combined scale to retake #2"). A deck that passes has been thought through; one that fails has been assembled. A corollary, the portability check: a title that could be lifted onto a different deck about a different company is almost certainly a topic title — real action titles are unique to the analysis underneath them.
Why it works
It works because it matches how senior readers actually consume a deck. Watch a director read a board pack on a tablet — eye on the title bar, hand on the next-slide arrow. Nielsen Norman Group's eye-tracking work calls the behaviour the layer-cake pattern: scan the headings, skip the body, drop in only where a heading earns the dive. A deck of topic labels is illegible at that scan speed; a deck of claims is already telling the story before any chart loads.
That cognitive shape exposes two failure modes action titles fix at once. The first is the buried claim: "Volume by quarter" forces the reader to reconstruct, off the chart, what the consultant should already have said. Multiply by sixty slides and most readers stop bothering. The second is a deck without a spine: topic-label titles let adjacent slides float independent of each other, and the deck becomes a binder whose slides could be reordered without changing the meaning — the diagnostic that there is no meaning. Action titles refuse both. The claim lands before the chart; each title must ladder up to the one above and be defended by the ones below; a vacuous title is now visibly vacuous on the page.
When it's the right tool
Action titles fit any decision-grade deliverable — strategy recommendations, board updates, investment memos, M&A reviews, post-mortems. Anywhere the reader is being asked to do or agree something, the deck's titles must carry the argument.
They are the wrong tool for genuine process artefacts: training decks where the slide teaches a topic (the title is the topic, by design), workshop decks where slides exist to provoke conversation, agenda and divider slides whose function is reference rather than argument. Forcing action titles onto these produces fake claims — "Five things every manager should know about feedback" — that are headlines costumed as claims. The tell is the absence of a verifiable verb.
Worked example — three slides on CPG margin recovery
Imagine a North American food-and-beverage business whose operating margin has compressed 240 bps over three years. The CEO is going to the board with a recovery plan whose governing thought is "Recover 180 bps of margin in 18 months by raising list prices 6% and capping discounts at 12%." Three slides; same charts in both versions; only the title row changes.
Topic-title version (before). The analyst's instinct.
| # | Title | What the title bar tells the reader |
|---|---|---|
| 1 | Margin trends | Nothing. Margin has done something. |
| 2 | Competitor pricing | Nothing. Competitors have prices. |
| 3 | Recommendation | Nothing. There is one. |
Slide 1 charts margin sliding from 18.4% to 16.0% since 2022; slide 2 shows our weighted-average list price 4–7% below the category leader across the top-20 SKUs; slide 3 lays out the price-and- discount move. None of it is visible from the title bar.
Action-title version (after). Same charts, same data.
| # | Title |
|---|---|
| 1 | Margin compressed 240 bps over three years, driven 80% by pricing, not by input cost |
| 2 | We are priced 4–7% below the category leader across the top-20 SKUs, with no volume case for the gap |
| 3 | A 6% list-price rise plus a 12% discount cap recovers 180 bps in 18 months at <0.5% volume risk |
Read the three titles in sequence and they are the executive summary: the problem is pricing not cost; the gap to leader is unwarranted; here is the move to close it, sized. A director who only reads the title bar walks out with the argument; a sceptic who wants to challenge driven 80% by pricing knows exactly which slide to open. The two decks contain the same analysis — only the title row changed. That is the leverage, and the reason action titles are the highest-return edit a deck can receive in the last hour before a board meeting.
Common pitfalls
Topic in disguise. "Volume drove growth" looks like a claim because it has a verb, but it carries no specific information — any deck on any company in any quarter could use it. "Volume grew 8% in Q3, driven entirely by Europe" is what an analyst writes when the analysis is done. The test: strip the company name and the period. Could the title still apply? If yes, it is a topic with claim-shaped grammar.
Passive voice with no agent. "Margins have been impacted" is the classic consultant hedge — a verb without a subject, an effect without a cause, a sentence the reader can't act on. "Commodity pass-through gaps cut margins 240 bps" names the agent and the magnitude. Active voice forces the analyst to commit to causation; passive lets them duck.
Multi-claim "X and Y" titles. "Volume grew 8% and margin compressed 60 bps, with further weakness expected if pricing weakens" is three claims welded together. One slide, one claim. If and or but shows up in the title, split the slide or drop the lesser claim into the body.
Mechanical "X drove Y" repetition. Twelve slides of "Pricing drove margin", "Mix drove growth", "Volume drove revenue" read as a fill-in-the-blank exercise. Real action titles vary the verb — grew, narrowed, recovered, broke, held, eroded, overshot, missed — because the analysis varies. Verb monotony tells you the author wrote the structure first and the thinking second.
Overreach beyond the chart. A title that claims more than the slide proves is worse than a topic label — it broadcasts a conclusion the evidence can't bear, and the first sceptic in the room finds it. "Pricing changes drove 60% of margin recovery" on a chart that shows a 40% correlation collapses on first challenge. The title must be defensible from the chart underneath it, not just persuasive.
Related canon
pyramid-principle— the parent discipline. Action titles are the pyramid applied at slide level: every title is a node in the pyramid.governing-thought— the one-sentence claim at the apex. Every action title must ladder up to it.headline-test— strip a deck to its titles and ask whether the sequence reads as the executive summary. Operationalises the action-title test.so-what-test— the line-edit applied to each title: does this sentence contain a so what, or just a what?scqa-framework— Minto's opening pattern; the first three or four titles of a deck typically encode S-C-Q-A.
Canonical references
- Barbara Minto — The Pyramid Principle: Logic in Writing and Thinking — Pearson, 1973 (rev. 2009) — the underlying logic. Author site.
- Gene Zelazny — Say It with Charts: The Executive's Guide to Visual Communication — McGraw-Hill, 1985 (4th ed. 2001) — the McKinsey chart-and-title standard.
- Gene Zelazny — Say It with Presentations — McGraw-Hill, 2000 (rev. 2006) — the deck-level extension. Author site.
- McKinsey — Remembering Gene Zelazny — firm obituary; joined 1961, died April 2023.
- Ethan Rasiel — The McKinsey Way — McGraw-Hill, 1999 — restates the action-title rule as house style.
- Mike Figliuolo — The Elegant Pitch — Career Press, 2016 — modern, post-McKinsey playbook for the headline-and-pyramid discipline.
- Nancy Duarte — HBR Guide to Persuasive Presentations — Harvard Business Review Press, 2012 — the cleanest non-McKinsey treatment of slide titles.
- Nielsen Norman Group — Layer-Cake Pattern of Scanning and F-Shaped Pattern — eye-tracking evidence for the title-bar reading model.