Headline Test

The Headline Test is not a structure. It is a test — three questions, asked in order, of any candidate title. Does it have a real verb? Does it make a defensible claim? If the reader saw only this line, would they leave with the slide's point? Pass all three and the title is action-grade. Fail any one and the slide is a folder name pretending to be an argument. The test takes ten seconds; running it on every slide, every time, is what separates a deck that lands from a deck that gets nodded through.

Where it comes from

The Headline Test was not invented in a consulting firm — it was borrowed from the copy desk. The Associated Press stylebook, canonical American newsroom reference since 1953, instructs writers to build headlines around active verbs and to avoid "label" headlines ("Election results", "Cost dynamics") that announce a topic without asserting anything. The New York Times and Financial Times copy desks apply the same rule under different names; Reuters and AP wire-style guides treat the noun-phrase headline as a draft state to be revised before the piece moves. Roy Peter Clark at the Poynter Institute codified the underlying discipline as Tool 3 of his Writing Tools (Little, Brown, 2006) — Activate your verbs — first published as a 2004 Poynter column. The point Clark hammers: a headline without a working verb has not finished its job, because the verb is what carries the finding.

What the strategy firms did, around the same time Barbara Minto's pyramid was solidifying in the 1970s, was transplant the newsroom discipline onto the slide title. Gene Zelazny — McKinsey's director of visual communication for forty years — drilled it into every analyst: the title of a chart slide is not the topic of the chart, it is the point the chart proves. "SG&A trend, 2019–2024" tells the reader nothing. "SG&A grew 40% faster than revenue, eroding 300 bps of margin" does the work. Zelazny codified the rule in Say It with Charts (1985); The McKinsey Way (Rasiel, 1999) made the title-bar review a firm-wide checkpoint — partners read only the titles in sequence and asked whether the deck still argued.

What it actually is

The Headline Test is three questions, asked in order, of any candidate title. The order matters: Q1 catches the labels, Q2 catches the tautologies, Q3 catches the technically-correct lines that nobody remembers. A title has to clear all three.

Q1 — Does the title contain a real verb? A real verb describes a state change, a finding or a recommendation — grew, eroded, compressed, outperforms, can recover. Copulas — is, has, are, was, have, had — assert existence, not motion, and do not count. "Volume" fails (a noun is a label). "Volume trends" fails (label, dressed up). "Volume is up" fails (only verb is a copula). "Volume grew" passes. The verb test is the cheapest filter and rejects roughly half of any first-draft set.

Q2 — Does the title assert a defensible claim? A title can have a working verb and still hide. "Volume grew" clears Q1, but the inverse ("volume shrank") is the only other possibility — no information. "Volume grew 8% driven by Europe" specifies magnitude and cause, and a reader could disagree with either. The diagnostic is the inversion test: state the opposite. If the opposite is incoherent or trivially uninteresting, the claim isn't doing enough.

Q3 — If the audience read only this line, would they leave with the slide's point? The kiosk test. Cover the chart, read only the title, walk away. Did the reader get the slide's conclusion? "Volume grew 8% driven by Europe" clears Q1 and Q2, but if the slide's point is that Europe is the only growth engine left and US turns negative next quarter, the title is forgettable — what without so what. "Europe carried Q3; US volume turns negative in Q4" clears all three. Q3 is where most technically-correct titles die.

Pass all three and the title is action-grade — a sentence the pyramid can ladder onto. Fail any one and the slide is a topic dressed as an argument.

Why it works

The title is the only line every reader reads. Nielsen Norman Group's eye-tracking work — the 2006 F-Shaped Pattern for Reading Web Content and the 2017 follow-up confirming it survives mobile — shows readers scan as an F: heaviest fixation on the first line, lighter on the second, spotty down the left edge. In any room with more than two readers, the title bar is the only line every set of eyes touches; slides whose titles don't earn the first stripe are statistically never read past it.

The test is a quality gate, applied twice. On draft: can I write a working title? If not, the analysis isn't done — the slide goes back. On review: if I read only the titles in order, does the deck still argue? The review sweep takes ninety seconds for a thirty-slide deck. Skipping it is the most expensive thing a team does in the last hour before a board.

It also catches a failure mode that pyramid rule-checking alone misses. A pyramid can be perfectly MECE, perfectly laddered, and still consist of clean groups of invisible labels — "Cost dynamics", "Margin considerations", "Strategic implications". Structure passes; the deck does not argue. The Headline Test catches what structure cannot: nodes have to say things, not just categorise things.

When it's the right tool

Almost always. The Headline Test applies to every slide title, section header, memo subject, blog H1, PR brief, and dashboard tile — any line whose job is to summarise what follows. Narrow exceptions: divider slides ("Part II: Recommendation" — a topic word is the function), agenda slides (intentionally topic-shaped), and neutral-framing research titles that shouldn't pre-judge data. In those, the title is navigation, not summary. Everywhere else, the test is the entry gate.

Worked example — the margin-recovery slide

The chart shows operating margin declining 120 bps over six quarters; the analysis finds pricing (not cost) is the cause and recommends a reset that can recover the full 120 bps in eighteen months. The chart never changes. The title gets rewritten three times.

Draft 1 — "Margin pressures persist." Persist looks like a verb but is functioning as a stative — pressures continue to exist. No state change, no finding, no movement. Fails Q1. The reviewer rejects without going further.

Draft 2 — "Pricing pressure compressed margins by 90 bps." Compressed is a working verb (Q1 pass); magnitude and cause are specific, inverse defensible (Q2 pass). Q3 stalls: a reader who saw only this line walks away with the diagnosis ("margins down because of pricing") and not the slide's point — that the same lever which caused the loss can recover it. The title carries the disease and withholds the cure. Fails Q3 — forgettable. (Side finding: draft says 90 bps, chart says 120; bad titles often hide stale numbers.)

Draft 3 — "A pricing reset can recover 120 bps of margin in eighteen months." Can recover embeds the recommendation (Q1 pass); magnitude, instrument and horizon are specific and the inverse is what a sceptical CFO will push back on (Q2 pass). The reader leaves with both diagnosis and prescription (Q3 pass). The title ships.

Under five minutes for three iterations. The chart did not change; only the line that announces it did. That is the editorial loop, and the failure the test exists to catch.

Common pitfalls

Copula-only verbs. "Margin is under pressure", "There are three drivers", "Costs have grown" — the only verb is is, are, have, has, was. An existence claim, not a finding. The single most common Q1 failure. Fix: "Three drivers explain 80% of the margin loss."

Passive voice with no agent. "Margin has been impacted by several factors" is what an analyst drafts before deciding what the finding is. The passive hides the agent. Force the by what? — commit ("Pricing erosion drove 70% of the loss") or send the slide back.

Hedging adverbs. "Margins may have somewhat softened, in part due to a confluence of factors." Each hedge — may, somewhat, in part, confluence — drains a unit of claim until none is left. If the finding is real, state it; if it isn't, the slide shouldn't be there.

Meta-descriptions. "This slide shows the margin trajectory", "The following analysis examines...". Working verb, defensible claim about the slide, zero information about the world. The most insidious failure: superficially passes Q1 and Q2, dies at Q3.

Bare numeric facts. "Margin declined 300 bps" clears Q1 and Q2 and leaves Q3 dangling: the number is fact, not evidence-for-a-conclusion. Fuse number with implication: "A 300 bps margin decline now threatens 2027 dividend cover."

Topic labels. "Cost dynamics", "Margin considerations", "Strategic implications". Acceptable on a divider slide. On a body slide it means the analyst couldn't write the headline and wrote the folder name instead. Caught at Q1; never let past.

Related canon

• action-titles — the rule that every body slide's title must be a declarative claim. Action-titles is policy; Headline Test is QA.
• so-what-test — the close cousin. Headline Test asks is this a claim?; So-What asks does the claim matter to the decision? A title that passes Headline but fails So-What is technically correct and strategically useless.
• governing-thought — the apex of the pyramid; the most important headline in the deck.
• pyramid-principle — the document-level structure the test enforces at the slide level.
• scqa-framework — the introduction whose Answer beat is the deck's master headline.

Canonical references

• AP Stylebook — apstylebook.com — canonical American newsroom reference (1953–). The headlines entry: active verbs, no label heads.
• Roy Peter Clark — Writing Tools — Little, Brown, 2006 — Tool 3, "Activate your verbs", is the primary-source statement of the verb test (originally a 2004 Poynter column).
• Gene Zelazny — Say It with Charts — McGraw-Hill, 1985 — codified the action title as McKinsey-deck doctrine.
• Barbara Minto — The Pyramid Principle — Pearson, 1973 (rev. 2009) — every node of the pyramid is a headline.
• Ethan Rasiel — The McKinsey Way — McGraw-Hill, 1999 — operationalised the title-bar review as a firm-wide editorial check.
• Andy Crestodina — How to Write Truly Great Headlines — Orbit Media — modern content-marketing restatement of the same three questions.
• Jakob Nielsen — F-Shaped Pattern for Reading Web Content — Nielsen Norman Group, 2006 — original eye-tracking evidence the first line carries disproportionate attention.
• Kara Pernice — F-Shaped Pattern: Misunderstood, But Still Relevant — Nielsen Norman Group, 2017 — follow-up confirming the pattern on mobile.
• Financial Times — FT Editorial Code — public version governing FT subhead practice.