Troubling trends, predictions and questions for marketers in 2026

A decade ago, this annual post started out to provide counterbalance to excitable trends in social media posts (Video! Influencers!), but very quickly social questions weren’t really social questions, but wider business or industry questions.

So, what are the questions at the start of 2026? Are we on the cusp of a substantial shift in the world of work, and if we are, what is it? What role does geopolitics play? Given this newsletter is, in theory, about marketing, how different will the industry look at the end of the year, if at all?

There are no real easy answers here. But a lot of AI talk.

This is quite a long piece. Here’s quick guide to what you’ll read in case you want to scroll to a section that interests you.

1. Major developments in AI in 2026.

2. Who will win the AI race?

3. What exactly is reality today?

4. How much will trust actually matter in marketing?

5. Economic realities for marketers that refuse to go away.

6. Economic realities that marketers should pay attention to (but won’t).

7. Social media is dead but also not dead.

8. Why the pivot to clipping is another trend for failing publishers.

9. Baxter Dury dancing in a Venice canal.

A broad question, but then there are multiple answers to multiple questions that sit underneath this.

Firstly, what can AI actually do? Depending on who’ll you ask, you’ll get a different answer. Do these answers depend on a better model, what does better actually mean and is it useful?

Drug research, which has large, complex data sets, is an obvious industry to benefit. Well trained LLMs cycle through combinations that would take human researchers years of trial and error.

For white collar industries and day-to-day, the applications are more limited by the human controlling it and their own imagination.

ChatGPT can give you recommendations for meals based on what’s in your fridge and order ingredients, but then Amazon’s Alexa could also do that and ended up being used as an expensive weather forecaster.

Using general AI models like Gemini, ChatGPT, and Perplexity for shopping may seem like a logical step (also, adverts) but this is an ecommerce challenge as much as an AI challenge: less capability but more about layering.

Who owns the workflow, who owns the transaction, and who captures the margin? Is it Google? OpenAI? Amazon? Instacart? Shopify? All? None?

AI can automate many time-consuming tasks, increasing productivity. But it behaves less like an employee and more like a supercharged, but not-always-reliable intern. If hallucinations and errors are a structural feature rather than a bug, then what role should humans play in the process? And if the answer is “better prompts,” is that an AI solution, a learning and development problem, or something else entirely?

Key questions around AI adoption are strategic (what’s the best use of AI in our business?) and engineering one (do we have the capability to do this well?). Using AI is easy. Implementing it well is not. The same was true of the internet, mobile apps, chatbots, and voice technology.

Many AI startups will fail to find a product–market fit. We’re still at an early phase of figuring out what it all means and what it all can do. Often the advances will be accidental. Nvidia was a primarily in the graphics card business until AI scientists found their the chips were very good for research.

But equally big tech will continue to funnel billions into capex to produce what exactly?

Google has a clear use case in search, and an improving model (and GEO is essentially a variant of SEO). Amazon and Microsoft have obvious enterprise and commerce applications but have struggled to produce truly differentiated consumer-facing products. Meta doesn’t have a clear use case but are funnelling money towards AI to produce… something?

OpenAI has capex that runs into trillions with no clear view of where the money is coming from, outside of further investment. Only 5% of individual users pay for ChatGPT, and corporate users may be more inclined towards enterprise versions of Gemini or Copilot. ChatGPT doesn’t have a moat and LLMs are essentially commodities.

AI isn’t going anywhere. AI-orientated businesses though? Much like the dotcom boom of the early noughties, many won’t survive.

First movers in new tech aren’t businesses, they’re bad actors. Scams, disinformation, and exploitation (explicit deepfakes) thrive as the models eliminate a lot of the telltale uncanny valley signs of synthetic production.

So what happens when we can no longer identify what’s real? Cognitive exhaustion asking “can I trust this?” several times a day. And when this is the question, it destabilises, well, everything.

False content can spread quickly (this is particularly an issue for teenagers and bullying), real material can be dismissed as fake. AI slop is surfaced over reality. Bad actors can plausibly deny and those with good intentions find it harder to put forward the truth. It’s easier for countries to destabilise other countries than it is to wage actual war. Polarised audiences can twist truth to suit their narrative.

The implications on a geopolitical level are, to be frank, quite terrifying. The implications at an individual level are more nuanced.

If trust is at a premium, then trusted connections should, theoretically, matter more. Is that a good thing? It depends who the connections are. Trust could mean quickly becoming radicalised into an extreme. Or it could mean connecting with trusted friends or peers.

The human element becomes more valuable, but it also risks these smaller, fragmented groups becoming more insular.

It’s simplistic to say trust and authenticity will become more in demand (how do you actually define authentic), but people will be drawn towards voices they perceive to be more trustworthy. Far from Martin Sorrell’s prediction of death, it suggests that PR practitioners who understand the landscape will be heavily in demand.

Marketers will talk up authenticity, but this matters as much or as little as it ever did.

Prime was authentically Logan Paul, but disregarded the laws of marketing science and category penetration (and, by all accounts, had an unpleasant product) and now resides in discount bins. McDonalds and Coca-Cola have created AI adverts (inauthentic!) but people will still buy Big Macs and Coke Zeros because they’re Big Macs and Coke Zero.

The work of the Ehrenberg-Bass Institute, Binet and Field, Ritson still stands. It doesn’t matter how authentic or not your brand is: if it ignores the 4Ps, is a mature brand obsessed with short-term performance, and isn’t building mental or physical availability, then it probably won’t grow.

Applying economic insights and market forecasts to marketing strategy is a bit of a lost art when it comes to marketing. If the market for a specific product is in decline, that calls for a different strategy than a growth market.

In 2026, inflation, cost of living and disposable income all play a part in maintaining margins. That’s as true for B2B as it is for B2C. Businesses have a finite budget, cash flow and, like consumers, want value for money.

How will this play out? It depends, but not always exclusively, on the brand.

Strong brands typically have higher price elasticity, which means better margins, a form of defensible moat, and an internal defence against discounting (although there is a lot of work and modelling markets need to do to sell this narrative internally).

On the flipside, brands are likely to lean heavier into discounting as a tactic. This is not a good idea (Mark Ritson puts it better than me, so I’ll just direct you there).

This time last year, economists were pricing in interest rate cuts throughout the year. That hasn’t panned out as expected. There’s always another disaster on the way. Gen Z and younger millennials are increasingly finding entry level jobs harder to come by.

This is a trend that requires thoughtful, carefully planned marketing strategies that have an intimate knowledge of their target customer.

Many marketers won’t do this. It’s easy to be blinkered by tactical executions and digital data (our CTR is up, praise the ROI!) while ignoring the bigger picture. Those who understand the economic reality of 2026 and how it affects target audiences will be the ones who win.

We start 2026 with a brutal job market, businesses squeezing their headcount and expecting more with less, and inflation – and interest rates – stubbornly sticky. The number of companies trading at multiples of 34x earnings looks like a bubble, but may not burst.

Even so, it’s one reason why both gold and Bitcoin have attracted a lot of retail investors, as the market feels volatile. Auto loans may not be the next sub-prime mortgage but it’s got Wall Street rattled.

Then there’s the economic reality of climate change (I’m always somewhat surprised that nobody from environmental movements seems to make this case loudly enough: late stage capitalism hates anything that could hurt profits).

From an insurance perspective, there are some areas that simply cannot afford or even get home or business insurance, as the risk of weather-related disasters are just too large. Shepparton in Victoria is one of Australia’s most uninsurable towns. The insurance industry may be an unlikely ally to the Greens, but they have a lot of shared goals.

It’s an interesting year to be a teenager in Australia. At the end of 2025, the federal government banned under 16s from social media, albeit inconsistently: Roblox, Facebook Messenger and Discord are exempt, Reddit and YouTube are included.

Messaging apps are arguably more of an issue for cyber bullying, while YouTube is realistically more an evolution of TV rather than a social network.

Nobody seems entirely sure what the impact will be, but as a national experiment, it mirrors a trend: social media is less about posting and more about consumption.

Instagram, TikTok, YouTube and more are semi-open about aiming to keep people on their apps for as long as possible to generate advertising revenue. This means algorithmically-served entertainment rather than getting updates from friends and family.

It’s one reason why Snapchat has struggled versus the bigger social platforms, as it leans more into friends and connections than serving whatever will get your attention.

But that doesn’t mean people want to doomscroll solo. The internet always has - and will be - based around niche communities. It’s why Reddit is more popular than TikTok in the UK, and why Discord could well have a moment in 2026.

If connections are important, then it also means community managers come back to the foreground.

Nurturing online advocates and being seen to be visible and accountable as a brand will count for more in an age where LLMs like Reddit and authentic reviews and feedback are more important than even in serving as trust signals versus AI slop. Storytelling means nothing without a community to tell it to.

The brands we associate with social media – Facebook, Instagram, TikTok, X – aren’t really social any more, while social connections are more likely to be found on fandom subreddits, messaging apps and live streams. The internet, as ever, evolves.

Brian Morrissey, who details publishing trends at The Rebooting, has long charted the slow decline of the pageview, which is only likely to accelerate in 2026. As Morrissey writes, “Webpages are not going away, but they’re becoming more peripheral to publisher strategies as search referrals decline. In 2026, expect more shifts to audio, video and live events.”

The likes of Buzzfeed and Vice, which built their businesses on driving clicks to page views from social media, are yesterday’s news. Publishers who still depend on display advertising off the back of page views have seen traffic decimated with the rise of AI summaries and tightened social walled gardens.

Google Discover has become a precarious last ditch dependency – precarious, as Google doesn’t have the best track record when it comes to supporting publishing.

Arguably the greatest success so far has been the New York Times, which is less a news organisation and more a news-focused parent brand with extreme diversification into events, puzzles, membership tiers, newsletters, ecommerce and more.

But what works for well-known, well-established brands like the NYT, Guardian, Mail Online and The Economist won’t necessarily work for other publications.

The current landscape is as fragmented as ever Nic Newman’s forward to Reuters’ 2025 Digital News Report summarises the challenge neatly:

“An accelerating shift towards consumption via social media and video platforms is further diminishing the influence of ‘institutional journalism’ and supercharging a fragmented alternative media environment containing an array of podcasters, YouTubers, and TikTokers. Populist politicians around the world are increasingly able to bypass traditional journalism in favour of friendly partisan media, ‘personalities’, and ‘influencers’ who often get special access but rarely ask difficult questions, with many implicated in spreading false narratives or worse.”

This highlights a number of challenges, mainly:

• Media dollars follow audiences. Right now, the dollars lean towards right wing and even far right outlets.

• Clipping is the trend of 2026 for news organisations, but this presents its own challenges. The clips often achieve earned reach, but on somebody else’s platform. At some point reach needs to translate into revenue. Is that directing back to an owned property, and is that sustainable? Or does this require branded partnerships and, if so, why should an advertiser choose a news organisation over creators?

• If news organisations behave like creators, does this strengthen or weaken institutional journalism? After all, creator-style publisher posts are still just another item in the feed.

• Podcasts are growing but still have lower reach and a harder buying system than other media. They do have trust, but again are competing against much better funded creators (and what constitutes a media organisation in the podcast realm? Is Joe Rogan a podcaster or a publisher?).

• Do media organisations now lead with personalities of star reporters, and how much risk is involved if the talent leaves? Does this become the reporter’s brand over the publisher’s brand?

Meanwhile, the playbook of owning your audience is in danger of capping out. The number of people who willingly pay for news remains stable at 18%, and new independent publications, Substacks and newsletters appear all the time.

This is a challenge: subscriptions add up and people need to be choiceful in where they spend. Is it better to be free to a large audience and pull in brand partnerships or charge and have a small audience? And are newsletters already a commodified market?

What’s emerging are media brands (and some are barely media at all) who are creating for audience consumption today rather than trying to retrofit an older model (For all Buzzfeed’s success in the 2010s, at its core, the business still defaulted to the older metric of page views).

The Daily Aus have a social-first approach posting simple explainers and breakdowns on Instagram, complementing their daily newsletters, and expanding into podcasts (and by default video clipping). There is a website, but this is more hygiene than the focus. It’s easy to avidly read the publication without ever visiting the website, while channels are diversified.

At the other end of the scale are individual personality-led publications, such as Casey Newton’s Platformer and Taylor Lorenz’s User Mag. They’re as much creators as they are journalists.

And then you have not-journalists such as Marques Brownlee, a YouTuber first and foremost, but is more forensic in his tech reviews than many tech journalists.

Unlike the other sections, I’ll make a few predictions here:

• Publications that are chasing distribution trends, such as clipping, without focusing on strategy, will continue on a downward spiral.

• Publications that lean too heavily on one channel that they have no control over - Google Discover, Instagram, LinkedIn, TikTok - will cease to be relevant with one algorithmic update.

• The sheer level of fragmentation and solo publishers or entrepreneurs means we’ll see mergers or joining forces. This will be either in the name of growth (there’s only so far one person can scale without committing significant capital) or necessity (nobody starts a business because they love admin or finance). There is still a level of unbundling to be done, but rebundling is looking like a more attractive option right now.

Of course, to quote William Goldberg, nobody knows anything. One geopolitical event, one market crash, one innovative AI or product breakthrough could change the whole landscape. We don’t know what we don’t know and haven’t yet thought of. But a lot of the future still looks an awful lot like the past.

I discovered this album a little too late to put it into my end of year lists but it’s been on play non-stop in the Betwixtmas period. Allbarone is funny, intimate, and different from previous output, yet very familiar to fans of Baxter Dury. You sense his late dad Ian would be proud of this one.