When digital advertising arrived, the industry asked one question obsessively: how much more efficient is this than the old way? You could target a single person instead of a whole newspaper's readership, measure clicks instead of guessing at reach, personalize the message down to an individual. Everyone got obsessed with the efficiency gains.1 Rory Sutherland's point is about the question that got skipped in all that excitement: what did the old, inefficient, expensive mass media actually do well that the new efficient version doesn't do at all?
Digital media can do things conventional media genuinely couldn't — a physical newspaper can't show every reader a different ad, but a webpage can.2 That's real, and it's not in dispute. What Sutherland says got missed is the reverse comparison: conventional mass media was good at conviction, trust, prompting social conversation, being part of a shared culture — and he doesn't think digital media, on the whole, replicates any of that nearly as well.3
The mechanism runs through cost and publicness, which this vault's own material on costly signaling already establishes: a message delivered to millions of people simultaneously, in a shared public space, carries an implicit promise — you're making this claim somewhere you can be held accountable for it, in front of everyone at once, the way a wedding vow said in a church full of witnesses means something a private promise doesn't.4 A one-to-one targeted digital ad is the opposite structure: cheap, private, deniable, invisible to anyone but the single recipient. It can be more efficient and simultaneously less credible, because efficiency and credibility were never the same axis.
The industry's mistake, on this account, wasn't choosing efficiency — it's a legitimate goal — the mistake was treating the shift to digital as a pure upgrade rather than a trade. You gained precision and lost the specific kind of public, costly, communal credibility that mass media had by construction. Sutherland doesn't argue for going back; he argues for noticing the trade was made at all, since an industry that doesn't know it made a trade can't compensate for what it gave up.
You're planning a campaign and the media-buying recommendation is entirely digital, entirely targeted, entirely optimized for cost-per-click. You ask a question the brief didn't ask: what is this campaign trying to build — a transaction, or trust? If it's a transaction (drive this specific sale this week), the digital-efficiency case is probably right as-is. If it's trust — a claim about who the brand is that needs to be believed, not just clicked — you notice the plan has quietly optimized away the one property (public, costly, shared visibility) that trust-building messages have historically needed.
You propose something that looks inefficient on a cost-per-impression basis: a piece of expensive, public, shared-context media — sponsorship, a physical placement, something that costs real money in a way everyone watching can infer. It won't outperform the digital plan on cost-per-click. You're not measuring the same thing anymore. You're buying the specific, harder-to-fake signal that "we spent real money making this claim in public," which digital efficiency, by its very design, can't manufacture.
The comparison that actually explains the trust gap isn't "digital versus traditional" as categories — it's cost and publicness as the load-bearing variable underneath both. A message that cost real money to place, delivered where a large shared audience can all see it was placed, functions as a public bet: you couldn't have afforded to make this claim, in front of this many witnesses, if it weren't true, or at least if you weren't confident enough to risk your reputation on it.4 A one-to-one targeted digital ad strips out both properties at once — it's cheap to produce and invisible to anyone but its single recipient, which means it can be wrong, misleading, or simply forgettable with zero public accountability attached. The efficiency that made digital advertising so attractive is the same efficiency that removed the exact mechanism mass media used to manufacture trust.
Not every piece of digital media is trapped by this — the diagnostic is specific, not categorical. A transaction ad (drive this sale, this week, to people already close to buying) doesn't need the costly-signal trust mechanism; efficiency is the right optimization target and digital is the right tool. A trust ad (this brand is who it says it is, believe this claim about our identity) needs exactly the public, costly, hard-to-fake property that cheap targeted digital structurally lacks. The tell is in what the ad is asking the viewer to believe versus do: "buy this now" is a transaction claim efficiency can serve well; "trust who we are" is a credibility claim that efficiency quietly undermines the more efficient it gets.
The binary isn't absolute, and it's worth naming the exception precisely: some digital formats are both digital and genuinely costly and public — a Super Bowl-adjacent platform takeover, a major sponsored placement visible to a whole audience simultaneously, anything where the spend is large enough and the visibility broad enough that viewers can infer real cost and real public exposure even though the medium is technically digital. These formats recover the trust-generating mechanism precisely by reintroducing the two properties (cost, publicness) that cheap targeted digital removed — which confirms the deeper claim: the relevant axis was never "digital versus traditional," it was cost-and-publicness all along, and digital media happens to have made it unusually easy to build cheap, private, low-trust advertising by default.
If trust-building genuinely requires cost and publicness, and transaction-driving genuinely rewards efficiency, then a brand trying to do both with one undifferentiated media budget is optimizing against itself — every dollar shifted toward efficient targeting is a dollar quietly withdrawn from the trust-building mechanism, whether or not anyone frames the trade-off that way internally. The brands most exposed to this mistake are the ones whose entire media plan migrated to performance-optimized digital over the past decade without a parallel, deliberate line item for costly, public trust-building media — not because performance marketing is wrong, but because a media plan optimized purely for efficiency has, by construction, stopped investing in the property that made people believe the brand's claims in the first place.
Sutherland offers this as a diagnosis from decades inside the advertising industry rather than a controlled study comparing digital and mass-media trust outcomes — [POPULAR SOURCE], treat accordingly. The honest tension: some digital formats (large-scale platform takeovers, major sponsored placements) are both digital and genuinely costly/public, which complicates a clean digital-vs-mass-media binary — the real variable underneath Sutherland's claim may be cost-and-publicness specifically, with "digital vs. conventional" only a rough proxy for that deeper axis.
This is the media-economics application of the same principle documented in Costly Signaling — where that page establishes the general mechanism (wasted resource as proof of resource abundance and genuine commitment), this page shows the specific historical moment where an entire industry built its efficiency case on precisely the axis that mechanism warns against optimizing away.
Behavioral-mechanics — Costly Signaling. This is the direct application: conventional mass media's credibility ran on cost and publicness functioning as a costly signal — you couldn't have afforded to say this if it weren't true, and everyone watching knew you'd made that expensive, irreversible bet in public. Digital advertising's efficiency gain is, read through this lens, a signal-cost reduction, and the trust erosion Sutherland describes isn't mysterious — it's exactly what costly-signaling theory predicts happens when the cost of sending a signal drops: the signal itself becomes cheaper to fake, so it carries less information. The insight neither page reaches alone: an entire industry can spend a decade optimizing a variable (media cost) without realizing that variable was secretly the source of the thing (trust) it also depended on.
Business — Marketing as Philosophical Outlook, Not Tactical Function (same batch). That page argues marketing's real contribution is a way of seeing the customer, easily crowded out by ROI-language. This page shows a specific historical instance of exactly that crowding-out: the industry's ROI-and-efficiency framing of digital media is precisely the kind of narrow, legible justification that page warns displaces the harder, truer, less measurable value (trust, conviction) mass media used to provide by construction.
Sharpest implication: every efficiency gain in communication is potentially a trust cost nobody priced in, because efficiency and credibility have historically been produced by the same mechanism — visible expense — running in opposite directions from each other.
Generative questions: