Marketing is about to get a lot more personal

A marketing budget has always bought the same thing, which is a place in front of people who (most of the times) did not ask to hear from you. For most of the last 100 years those places belonged to a handful of channels, each with a fixed amount of airtime or column inches. The money went to whoever owned these channels and a company's message was compressed into something a committee would approve.

That is where the brand voice comes from. It's polished, careful and spoken by nobody in particular on behalf of an abstract company entity. Audiences never liked it much, but with a few channels and no way to talk back, they had little to compare it against.

The places in front of people are still scarce. If anything they are scarcer, because everyone now advertises online and the receiver still has the same hours in a day. But AI is changing what fills them. The message itself now costs close to nothing to produce, so the volume competing for each moment of attention has multiplied and a convincing fake is as cheap to make as the real thing.

This is not only an advertising problem. Between social media and AI, the same person now meets an endless stream of posts, reviews, videos and news with no reliable way to tell what is true. Most people have been misled by that stream often enough to start from doubt rather than from trust, so every message a company sends lands in that general doubt.

That is why marketing's challenge is moving from producing the message and buying the placement to getting the message noticed and believed once it arrives. So, what will get noticed and believed from here is going to be personal in two senses: 1) it will come from a person with a name rather than from a brand and 2) it will be made for the receiver specifically rather than for everyone at once. Neither is what marketing departments are used to spending money on.

This piece walks through both these points, the third thing that ties them together, why budgets have not caught up yet and what marketing will look like once they do.

Content is becoming cheap and the brand voice is losing its value

A brand is an abstraction. Nobody stands behind its words, the words have passed through a team whose job is to make you think what the company wants you to think and the purpose of that team is to sell rather than to inform truthfully. People always discounted the brand voice for this reason. What changed first was social media, which gave every claim a comment section and every product a review, so the gap between what a company said and what was true started getting caught in public within hours instead of never.

Then AI arrived on top of that and made the supply of content effectively infinite. Graphite, an SEO firm that tracked a large sample of new web articles, found that AI-written pieces overtook human-written ones in November 2024 and made up just over half of everything published by May 2025. And the readers have noticed. In Kantar's 2025 Media Reactions study, 44 percent of consumers said AI-generated ads bother them and 57 percent said they worry about being shown fake ones.

When the cost of a piece of content is zero and the reader assumes AI created it, the message of that content no longer says anything about how much the sender cares. Trust has to come from somewhere else. In practice it now comes from three places: a person who puts their name to the claim, evidence that the sender did the work on you specifically and a third party confirming that the result was real.

Trust moved from brands to people

A person has something a brand does not, which is a name that exists outside the company and loses or gains credibility depending on whether what they say is truthful and consistent with their character. You can check what they said last year, you can see their face and you know they will still be around after the campaign ends. That is why trust has migrated toward creators, toward founders who speak for their own companies and toward media brands built around individuals rather than mastheads.

The acquisition market is following the same logic. OpenAI, a company that can generate unlimited content, bought TBPN in April 2026, a daily tech talk show with two hosts and eleven staff that made around $5 million in advertising revenue the year before. The price was reportedly in the low hundreds of millions. Paramount Skydance paid $150 million for The Free Press in October 2025, a Substack publication whose value sits almost entirely in Bari Weiss's name. Publicis has bought Influential, Captiv8 and 160over90 in two years for well over $1 billion combined. Each of these buyers already had the means to produce content and paid instead for people an audience already believed. Taken far enough, the logic turns the industry upside down and creators start launching products of their own, which is what Beast Industries is doing at a $5.2 billion valuation.

Founders are the version of this that a company can build rather than buy and almost none of them have started. LinkedIn has said that only about 1 percent of its members post in a given week. Yet in Edelman and LinkedIn's 2025 study of business decision-makers, 61 percent said a person's thought leadership tells them more about what a company can do than its product marketing does, so most of that demand is currently going unmet.

Influencer marketing gets a fraction of the budget and most of the trust

And the money has not followed trust yet. Payments to creators come to about $12 billion a year in the US according to eMarketer. Even on IAB's wider count, which adds the media money brands spend pushing creator content, the total is $44 billion this year. Against total advertising that is somewhere between 3 and 10 percent depending on what you count, for the channel consumers say they believe most. Brands still treat it as an experiment funded from what is left after paid search and paid social have taken their share.

Unilever is the exception and it shows how far the rest have to go. Inn his first weeks as chief executive, Fernando Fernandez said the company would move 50 percent of its advertising budget to social media and work with twenty times as many influencers. Eighteen months on, Unilever works with around 300,000 creators, up from 10,000 when he made the announcement. It ran its 2026 World Cup campaign through 50,000 of them to a combined audience of more than 600 million. Handling that many relationships forced it to automate a big part of the work on influencer campaigns, which is the infrastructure the rest of the market does not have yet.

This influencer infrastructure consists of three things that made paid search and paid social easy to fund, namely measurement, a way to find the right people among millions of accounts and someone to run the operation once you have found them. Marketers in survey after survey of the channel name measurement as their main obstacle, which is why the platforms have started building it themselves. Meta's Partnership Ads, the format that lets a brand run a creator's post as an ad with normal tracking, reached a $10 billion annual run rate in the first quarter of 2026.

There's also a talent problem which is even harder. The people who understand a platform's culture rarely function inside a company and the people who function inside a company rarely understand the platform. When a16z opened a fellowship to find the few who do both, more than 2,000 applied for 65 places and portfolio companies were asking to hire the graduates before the programme ended.

As the infrastructure arrives, supply responds. Goldman Sachs expects the creator economy to roughly double from its 2023 size to $480 billion by 2027. Much of the new supply is smaller accounts, as brands increasingly favour micro and nano creators with a few thousand followers each, because they cost less per engagement and are trusted within their niche in a way no celebrity is. Assembling a hundred of them into one campaign, tracking the results and paying them becomes a service business in its own right and that is where we'll soon see a wave of agency acquisitions happen.

Relevance and personalisation is the second route to trust

Trust also comes from evidence that someone did the work on you specifically. Compare two law firms. One offers legal advice to companies. The other offers legal advice to seed-stage UK AI start-ups and opens by naming three problems it has already spotted in the terms and conditions on your website. The second firm has shown you two things at once. It knows your world better than a generalist does and it spent resource on you before you spent anything on it. Even when the reader can tell that AI did the analysis, the fact that a company pointed its resources specifically at them carries the same signal, because the alternative was to send everyone the same message.

Relevance also solves the attention problem that infinite content created and search shows this most clearly. People now type full questions into Google and ask ChatGPT and Claude the same way, because a specific answer saves them reading ten generic pages. Pew Research measured what that does to the old model in 2025. When Google shows an AI summary, users click through to a website 8 percent of the time, against 15 percent when it does not. Adobe paid $1.9 billion for Semrush so that brands can work out how to appear inside those answers.

TikTok as a search engine is a similar example. Two in five Americans use it that way according to Adobe's 2024 survey and what they want is a person narrating the answer with the product in their hands, which brings the argument back to the fact that people trust people.

Most personalisation today falls into the uncanny valley

The bar for this is high because the current attempts are so bad. A first line that mentions your company name and a fact scraped from your website reads as a template and tells the recipient that nobody did any homework. Reply rates to cold email have fallen from around 8.5 percent in 2019 to about 3.4 percent in 2026 according to benchmarks published by Instantly, one of the largest cold email platforms, while the volume sent kept rising. Gartner found in 2025 that 53 percent of buyers had a personalisation experience they considered negative. Personalisation done badly is worse than none, because it looks like an attempt to fake the effort that the good version actually spends.

Clearing the bar requires knowledge that is not on the open web. In finance, that knowledge comes from expert networks, an industry AlphaSense valued at $930 million when it bought Tegus in 2024. Marketing is building its own versions. One is AI-moderated interviews with a company's actual customers, run at a scale no research agency could staff, which is what Conveo raised a $50 million Series A in September 2026 to do. Others are first-party behavioural data, surveys of a specific niche and simply having met the market in person. Whatever the source, a message written from what the market told you last week beats one written from what everyone can already read.

Automation is pushing the money toward people, data and proof

The third big force in the new age of marketing is automation and it decides where the money that used to pay for production and placement goes next. Meta said on its July 2026 earnings call that its Advantage+ suite, which lets the system pick the audience, the placement and the budget split for an advertiser, was running at a $75 billion annual rate and that 9 million small businesses were already using its AI creative tools.

The ultimate goal, first reported by the Wall Street Journal in 2025, is a brand only supplying a product image and a budget by the end of this year while the system writes the copy, generates the video and decides where the money goes. Google is doing the same with Performance Max. Everything between the brand's budget and the consumer's screen is becoming automated inside the ad platforms themselves.

The agency holdcos results already show it. WPP's revenue fell 8.1 percent in 2025 and the firm has cut around 11,000 jobs since the start of that year. Omnicom completed its $13.5 billion purchase of Interpublic in November 2025 mainly to take cost out. Publicis, in the same market, grew 5.6 percent and the main difference is what it owns - Epsilon's identity data and the creators at Influential and Captiv8. The holding company that bought data and people grew while the one that sold control of Kantar, its data business, back in 2019 shrank.

Interestingly, such automation also creates a type of conflict of interest. When Google sells the ad, prices it in its own auction and then reports whether it worked, the customer who would have bought anyway still counts as a win and nobody inside that loop earns anything by saying the spend was wasted (not to mention the notorious overcounting of leads / results you get through standard tracking on ad platforms). As more buying moves to pay-for-outcomes, the brand needs someone outside the loop to count. That is why the ad result checkers have become one the most sought after companies in the industry. Nielsen agreed to pay $2.15 billion for DoubleVerify in August 2026, Novacap took Integral Ad Science private for $1.9 billion and H.I.G. carved Kantar Media out for $1 billion, all within twelve months.

Finally, automation also makes two of the expensive parts of personal marketing cheap. Building something specific for every prospect was never economic when a person had to write it and running a campaign across a hundred small creators was unmanageable when a person had to track each one. Both are now possible, so the money is moving into the data that makes the personal message accurate and the measurement that helps you choose the creator and shows that they delivered.

What does all of this mean for marketing?

Putting the three forces together, this is how I expect the next few years to play out.

Spending on creators closes the gap with its measured effectiveness and its share of advertising moves from about 3 percent toward double digits. The measurement and coordination layer that makes that possible gets built in the next two years, which in turn brings a wave of new small creators onto the supply side.

More companies buy media brands. The OpenAI and Paramount deals are the first of a series and the next buyers will be ordinary corporates acquiring a newsletter, a show or a publication, because an audience that already trusts the voice is faster to buy than to build.

Personal brands become part of the marketing budget. Companies build small in-house creative teams whose job is to make the content of their founder, their executives and their most visible employees good enough to compete with professional media. The companies that start early hold audiences that the late ones will have to pay for. With about 1 percent of LinkedIn posting weekly, the field is still nearly empty.

Companies start doubling down on personalisation built on data that only the sender has. The messages that get read are built from what a company found out about its customers by itself, whether through customer interviews, niche surveys, first-party behaviour or just the time they spens in the market speaking with potential clients. Outbound does not shrink and gets far more specific, but managing that will not be cheap. Companies will put real resources into it and get steadily better at it, because the generic version has stopped working.

Independent measurement gets written into contracts. Large brands moving to outcome-based pricing will name a third-party checker in the agreement, in the same way that clients of law firms have begun writing AI billing clauses into their engagement letters.

The point is that marketing is getting personal in both senses at once. The message increasingly comes from a person rather than a brand and is increasingly built for one reader rather than for everyone. Everything in between is being automated. What remains is the person sending the message and how well they know the person receiving it.