Most retrospectives on digital marketing treat it as a story of tools — the rise of search, the arrival of social, the pivot to mobile, the coming of AI. The more useful reading is that digital marketing has been, from the start, a running argument about who owns attention, who owns data, and who gets paid when the two meet. The tools are downstream of that argument.
Healthcare marketing is a clean illustration of how that argument resolves in practice. Digital has decisively overtaken traditional channels in the category, and 2025 was the first year social media outspent linear TV in healthcare and pharma advertising, according to eMarketer data reported by Fierce Pharma. Linear TV has not vanished — pharmaceutical and over-the-counter brands still put more than $7 billion into it through early December 2025 — but the ordering has flipped, and the forecast gap widens from there. This shift wasn’t led by strategists. It followed the patients, who had already made it.
That is the pattern underneath almost every phase of digital marketing’s evolution. The audience moved first. The budget followed. The narrative arrived last, usually dressed up as strategy.
The early web years, roughly 1994 to 2004, were about presence. Companies built sites because competitors had them. Banner ads were sold on the same logic as print: impressions, placements, gross rating points reimagined for pixels. The measurement was crude, the creative was crude, and the assumption — inherited from a century of mass media — was that the advertiser bought the audience from the publisher.
The second phase broke that assumption. Search advertising, and then programmatic display, transferred power from the publisher to the platform. Google did not sell a magazine’s readers. It sold intent, at the moment intent was expressed, priced by auction. That single mechanical shift is the reason a coffee roaster in Portland can outbid a national chain for a specific query at a specific hour, and it is the reason the open web’s advertising economics never recovered.
The third phase, social, went further. It sold not intent but identity. Facebook’s ascent, and later Instagram’s and TikTok’s, rested on a bargain most users never consciously agreed to: behavioral data in exchange for free distribution and free entertainment. In April 2026, Emarketer forecast that Meta would surpass Google in worldwide digital ad revenue for the first time, at $243.46 billion against Google’s $239.54 billion — a milestone that would have seemed implausible a decade ago and now reads as the natural conclusion of the identity-data trade.
The practical consequence shows up in the unit economics of direct-to-consumer selling. Customer acquisition costs on Meta have climbed steadily since 2019. Email lists, once written off as a legacy channel, have become among the most profitable assets a brand owns, precisely because nobody rents them out. Paid social spend keeps growing anyway, because the alternative — silence on the platforms where customers spend their evenings — is not really an alternative.
This is the quiet condition of modern digital marketing. The dominant channels are simultaneously indispensable and unsustainable at current unit economics for a large share of advertisers. Everyone knows it. Almost no one exits.
The fourth phase, the one currently unfolding, is about the fragmentation of the identifier — though not in the way the industry spent five years predicting. Apple’s App Tracking Transparency and a growing patchwork of state privacy laws in the United States have made behavioral data harder and more expensive to collect. The third-party cookie, however, did not die on schedule. Safari and Firefox have blocked them by default since 2020, while Google abandoned its Chrome cookie deprecation timeline in July 2024, dropped the replacement user-choice prompt in April 2025, and retired most of the Privacy Sandbox APIs that were meant to succeed cookies in October 2025. What marketers actually got was not a clean cutover but a permanently unfinished one: tracking that works inconsistently, differently by browser, and with no settled endpoint. The Supreme Court’s June 2026 decision in Chatrie v. United States, holding that a geofence warrant is a Fourth Amendment search, is not a marketing story, but it belongs on the same map. The legal and cultural tolerance for ambient location and behavioral tracking is narrower than it was in 2016, and the direction of travel is clear even where the specifics remain unsettled.
Marketers have responded with what the industry now calls first-party data strategy, which is largely a rediscovery of things direct marketers were doing in the 1980s. Loyalty programs. Owned email. Preference centers. Postal mail, of all things, has re-entered serious media plans, and DMNews has covered how operators are using direct mail to win at engagement marketing in ways that would have looked quaint five years ago. The postal meter itself has quietly become a software platform, integrated with CRM data in a way its analog predecessors never were.
Subscription businesses have been rebuilding their measurement stacks around this reality for years. Paid social attribution windows keep shrinking. Modeled conversions keep expanding. Incrementality tests and media mix modeling now absorb the hours that went to campaign-level dashboards in 2021. The measurement story has moved from deterministic to probabilistic, and most marketers under 40 have never worked in an environment where it was anything else.
The fifth phase, arriving now, is the collision between generative AI and creator-led distribution. Both are already reshaping unit economics. Influencer spend, once an experimental line item, now sits in core media plans — though the return multiples the industry quotes come from vendor benchmark surveys with incompatible methods and self-reported inputs, and should be read as directional rather than measured. Generative tools, meanwhile, have compressed the cost of producing creative variants to a fraction of what a single finished asset used to cost, which changes what the creative brief is for.
The temptation is to describe this as a revolution. It is closer to an acceleration of a pattern already visible. Channels that concentrate distribution — a search engine, a social feed, an AI answer layer — extract most of the value. Channels that do not concentrate distribution — email, SMS, direct mail, owned communities — return most of the value to the operator. The middle, where third-party publishers sold audiences on the mass-media model, is where the money keeps disappearing.
What has changed for marketers is not the fundamental economics. It is the honesty required to operate inside them. The old story of digital marketing promised measurable, attributable, efficient growth. The current version offers something less flattering: a set of platforms that work, mostly, at prices that keep rising, measured through models that require faith as much as math, subject to policy shifts that can rewrite the plan in a quarter.
Put in operational terms rather than philosophical ones, the job itself has changed. It is no longer to find the cheapest click. It is to build the assets — the list, the brand, the creative library, the customer relationships — that reduce dependence on any single platform’s pricing decisions. That is a different job than the one the discipline advertised for most of the last twenty years.
The evolution of digital marketing has not been a march toward more precise targeting, more perfect measurement, or more efficient spend. It has been a slow transfer of leverage — first from publishers to platforms, then from advertisers to platforms, and now, unevenly and incompletely, back toward whoever owns a direct relationship with the customer.
The marketers who are calm about the next phase are the ones who have already accepted that the platforms are not partners. They are landlords. And the rent is going up.