Trends

A trend is a change in behavior that lasts for years and survives its first headline. A fad is a wave that lasts weeks: it shoots up, everyone talks about it, and six months later nobody uses it. A megatrend is a shift over decades: urbanization, Europe’s aging population, digitalization. For a marketer, the difference is crucial: you invest a campaign in a fad, a channel and people in a trend, and a strategy in a megatrend. The most common mistake is to take a fad for a trend and build a yearly plan on it. The second most common is the opposite: to declare a trend a fad and ignore it for five years.

The most widely used map of how innovations spread is the diffusion of innovations theory, published by sociologist Everett Rogers in 1962. Rogers divided people by when they adopt something new: 2.5% innovators, 13.5% early adopters, 34% early majority, 34% late majority and 16% laggards. The curve is bell-shaped: a few enthusiasts, then the masses, and finally the people who buy a smartphone only when their feature phone stops working. In the late 1980s, consultants at the firm Regis McKenna, and Geoffrey Moore in his book Crossing the Chasm (1991), added the chasm between early adopters and the early majority: enthusiasts want something new, pragmatists want a working solution with references. Most innovations end up in the chasm.

How do you tell where a trend is on the curve? Ask who’s using it. If only the tech press writes about it and the only users are people who try everything, you’re at the 2.5%. When your aunt starts using it, you’re past the chasm. The second test: the five-year search interest curve in Google Trends. A fad has a spike and a drop; a trend climbs in steps. In 2012, the QR code looked like a fad nobody scanned; after 2020, thanks to restaurant menus and payments, it became something everyone takes for granted. The same goes for podcasts, covered in detail in the Media chapter.

The key word is adoption, not existence: video calls existed long before 2020. A trend isn’t born from an invention, but from the moment the majority changes its behavior. That’s why today we won’t ask “what’s new?” but “what have people started doing differently?”

Predicting trends is a business. The best-known tool is Gartner’s hype cycle, proposed in 1995 by analyst Jackie Fenn. It has five phases: the technology trigger, the peak of inflated expectations, the trough of disillusionment, the slope of enlightenment and the plateau of productivity. A 2024 analysis by The Economist, cited by Wikipedia, found that only about a fifth of breakthrough technologies make it through the whole cycle, and six in ten of those that fall into the trough never recover. The curve doesn’t hold for every technology, but as a language for discussion it’s excellent.

The second source is trend reports. Read the ones that have data and say who paid for them: Reuters Digital News Report (every year, free, with a Czech chapter), Kantar Media Reactions (consumers and marketers in one survey) and Trends: Artificial Intelligence from the Bond fund (May 2025, 339 pages of charts on AI).

The third source is your own signals, and they’re free. Google Trends shows search interest over time and by region. Similarweb estimates your competitors’ traffic. Meta Ad Library and Google Ads Transparency show which ads competitors are running right now, and the Wayback Machine shows how their website has changed over ten years. When three independent signals point the same way (search interest is growing, competitors are buying ads, a report flags growth), you have a trend. When only one does, you have a headline.

How to read a report: methodology first (how many people, where, when, who paid), then the numbers, and the conclusions last. A software vendor that finds that 90% of companies need its software isn’t a source, it’s an ad. And write down the year next to every number: a report from 2020 is history today.

Natural scientists have a better model for markets than economists do: evolution. The environment changes, and species that don’t adapt die out; not because they’re weak, but because they were built for a world that ended. The famous quote “it’s not the strongest or the smartest who survive, but the one who adapts best to change” wasn’t written by Darwin, even though it’s attributed to him. According to Quote Investigator, management professor Leon C. Megginson said it in 1963 as his own interpretation of Darwin. But the idea holds: the market is the environment, and the company is a species.

Two textbook examples. Kodak: in 1975, its engineer Steve Sasson built the first portable, battery-powered digital camera. Before 1997, the company had an 80% share of the US film market; digital threatened its most profitable business, so it kept putting it off. On January 19, 2012, Kodak filed for bankruptcy protection. Blockbuster, which you know from the Marketing lesson, turned down buying a young Netflix for $50 million in 2000 and went bankrupt in 2010. Both had the new technology, and both knew about it. They didn’t change their behavior, because the old behavior still made money.

How do you adapt in a controlled way? With the OODA framework of US Air Force Colonel John Boyd from the early 1970s: observe, orient, decide, act, and again. Whoever runs the loop faster than the opponent “gets inside their decision cycle.” In marketing: measure every week, not once a year; always keep small experiments running; decide on last week’s data, not last year’s survey.

From the Marketing chapter, recall the two kinds of innovation: incremental (a better camera in your phone) and disruptive (streaming instead of DVDs, the phone instead of the camera). Every company can handle incremental innovation; disruptive innovation kills the one with the most to lose. A framework for you: keep the core stable (brand, website, relationships), let the edges mutate (formats, channels, tools), and when an edge starts earning more than the core, move it inside before your competitors do.

Markets aren’t changed only by customers and technology, but also by governments. For a marketer, this takes three forms: the platform you stand on can be banned or sold; the rules for data, advertising and AI change across entire continents; and tariffs and supply chains change prices, the availability of goods and where it pays to sell. This is part of the PEST analysis from the Marketing chapter. Three examples from the last two years:

TikTok in the US. A law signed on April 24, 2024, ordered ByteDance to sell the US business by January 19, 2025. The app went dark in the US on the evening of January 18 and came back at noon the next day, after a promise of a delay. On January 22, 2026, TikTok USDS was formed: Oracle, Silver Lake and MGX with 15% each, ByteDance with 19.9%. The lesson from the Internet chapter: your own website and email list are the only channels nobody can switch off for you.

Tariffs. On April 2, 2025, the US announced “reciprocal” tariffs on almost every country, with a minimum of 10%. According to the Yale Budget Lab, Americans paid an average tariff of 16.9% in January 2026, the highest since 1932. On February 20, 2026, the US Supreme Court struck down the tariffs based on the IEEPA law; the administration replaced them with others, and in August 2026 the average rate was 11.0%. Rules for AI and data in the EU. The AI Act is phasing in step by step, and most obligations apply from August 2, 2026; the Artificial Intelligence lesson covers the exact timeline and the later delays. And on April 22, 2025, Google announced that it won’t remove third-party cookies from Chrome.

What can you do about it? Don’t try to predict politics; build so that it doesn’t catch you off guard: own your audience (website, email, database), have at least two channels, keep data in the EU, and go through a one-page PEST once a quarter. For a marketer, geopolitics is weather: you can’t influence it, but you can have a roof.

Ask “what is marketing using right now,” and you’ll get a list of tools. A better question is “where are people, and whom do they trust?” The Reuters Digital News Report 2026 (48 markets, 97,520 people) found for the first time that social media and video are the most common route to news: 54%, TV 52%. A two-point gap is narrow, and the authors advise reading it with caution. Trust in news fell to 37%, the lowest since 2015, and 10% of people use AI chatbots for news every week (7% last year). That gives us the first trend: creators and social formats instead of media, covered in detail in the Media and Social Media chapters.

The second trend: zero-click search. In March 2025, Pew Research tracked the actual Google searches of 900 Americans. An AI summary appeared for 18% of queries. When it did, users clicked a traditional link in only 8% of cases (15% without a summary) and ended the search right there in 26% (16% without a summary). The result: fewer visits from search, and a new discipline, writing so that AI cites you, called generative engine optimization (GEO). The third trend: first-party data. The industry has learned to collect emails and logins; newsletters are back, and in March 2025 Substack reported 5 million paid subscriptions. Personalization is built on your own data, too: offers based on what customers bought or read from you, not on someone else’s cookies.

The fourth trend: short vertical video everywhere (TikTok, Reels, Shorts, LinkedIn); the fifth, video podcasts (the Streaming chapter); the sixth, retail media: advertising directly in retailers’ online stores and apps. Seventh: events as content factories (the Events chapter). Eighth: sustainability backed by data, not a slogan. Since September 27, 2026, Directive 2024/825 has applied in the EU: generic claims such as “eco-friendly” or “green” without proof, and self-made “eco” labels without certification, are banned (overview).

And above it all, AI: according to Kantar, a net 75% of marketers plan to increase investment in AI assistants in 2027, while only 23% of consumers trust their recommendations. Kantar says “net,” which usually means the share of those who will increase investment minus the share of those who will cut it. That gap between marketers and people is a trend in itself.

According to DataReportal (data from October 2025), Czechia has 9.96 million internet users (94% of the population) and 8.08 million social media user identities. Ads can reach 8.08 million people on YouTube (the same number, by coincidence), 5.00 million on Facebook, 3.95 million on Instagram, 3.20 million on LinkedIn, 2.13 million adults on TikTok, 1.54 million on Snapchat and 1.13 million on X. Facebook isn’t dead in Czechia; it’s second after YouTube. TikTok reaches a quarter of adults, not “all young people.”

The strongest trend of recent years isn’t a tool, it’s tool fatigue. Since 2011, Scott Brinker has been counting the marketing technologies on the market: from 150 products in 2011 to 15,384 in 2025 and 15,505 in 2026. This year, for the first time, growth practically stopped (+0.79%), and Brinker writes about “peak martech.” For every part of the job there are dozens of options, and choosing alone takes more time than the work itself. The answer is an old one: KISS, keep it simple, stupid. In digital strategy, that means one website, one newsletter, one or two social networks done properly, one typeface, one message. People read the first line and skim the rest.

Engineering says the same thing as “boring technology” (Choose Boring Technology, by Dan McKinley of Etsy and Mailchimp): every company has only about three innovation tokens it can spend on something new and unproven; everything else should stand on things whose failure modes you know. In marketing: spend a token on AI or on a new format, not on a new CRM, a new website and a new agency in the same year. A new tool is easy to add and hard to run.

Simplicity is also an approach to the customer’s attention. It used to be “the more notifications, the better.” Today, a message from a brand that’s worth reading is rare, and sending noise is expensive: people unsubscribe, and you can’t reach them again. The measure of a newsletter isn’t how many you sent, but how many people still open it a year later. The same goes for websites: a static page doesn’t break; more in the Internet chapter.

How do you follow trends without going crazy? Have three sources, not thirty, and read them regularly. Reddit (since 2005, 121.4 million daily users in 2025): forums where people write about what doesn’t work. Hacker News (since 2007): what techies consider important shows up in marketing two years later. Product Hunt (November 2013, bought by AngelList for $20 million in 2016): new tools every day, ranked by votes; great for an overview, terrible for choosing, and most disappear within a year. And newsletters on Substack or Beehiiv: one good industry newsletter replaces ten websites.

The best idea often doesn’t come from competitors, but from another industry or another country. Don’t copy the content; borrow the format: the way something is filmed, written or built. Sports clubs or tech companies often try new formats before B2B companies do; what works there can be carried over into an industry where nobody has used it yet.

How do you choose a tool when five new ones appear every week? The four questions are in the list below the text. If you answer yes to the last one, don’t buy the new tool.

For reference, here’s my stack, as an example that you can work with very little: Claude Code (websites, scripts, agents), artifacts on claude.ai (overviews, run sheets), Midjourney and Higgsfield (images and video), Photopea, Squoosh and remove.bg (free editing in the browser), Adobe Express, Google Workspace, Trello, Umami (analytics without cookies), 1Password. For verification: Wayback Machine, Similarweb, Google Trends, Meta Ad Library, Google Ads Transparency, Atlas Čechů, ARES, Hlídač státu, TinEye, PageSpeed. More in the Internet and Production chapters.

Automation is a “when A happens, do B” rule that runs without a human: a form comes in → write it to a spreadsheet → send a confirmation → create a Trello card. Three no-code tools can do this; there’s an overview below the text. Zapier is the simplest; n8n is for those who want to keep their data in-house. Before you buy a tool, go through the four questions from the previous section.

What to automate: anything you do the same way three times a week that has a clear input and output. Sign-ups from the website into the CRM, confirmation emails, a weekly analytics report into chat, new article → LinkedIn post, an alert when the website goes down. What not to automate: anything that requires judgment, or where a mistake costs your reputation (a reply to a journalist, crisis communication, the price in a quote). The rule: do it by hand ten times first and write down the steps. What you can’t describe, you can’t automate.

What goes wrong: automations fail silently. An API changes, a token expires, a spreadsheet gets a new column, and the workflow sends empty emails for three weeks. That’s why every workflow needs a watchdog (an alert when it didn’t run) and an owner (a person’s name, not a department). The cost isn’t just the subscription but also maintenance: count on roughly an hour a month for every running workflow.

“Training your own AI” is the most common wish of marketing teams, and also a misunderstanding. Neither a custom GPT nor a Claude project trains the model. You give it instructions (who it is, how it talks, what it must not do), knowledge (files it draws from) and possibly tools (web, spreadsheets, APIs). The model stays the same; the brief changes. OpenAI introduced GPTs on November 6, 2023, and opened the GPT Store in January 2024; according to Wikipedia, the pharmaceutical company Moderna created 750 internal GPTs. Anthropic added Projects on June 25, 2024: each project has a context window of 200,000 tokens, roughly 500 pages, with its own instructions and knowledge. According to the pricing page (October 2026), you get up to 5 projects for free, and the Pro plan costs $20 a month ($17 billed annually).

The list below the text sums up what belongs in a project and what doesn’t. One more rule: a project is a document like any other; someone owns it and updates it once a quarter.

How much should you trust it? In 2025, Stack Overflow asked developers: 84% use AI or plan to, but only 33% trust the accuracy of its output and 46% don’t. The biggest frustration (66%): answers that are “almost right, but not quite.” That’s exactly the risk of a marketing GPT: a press release that sounds good and has the wrong date. That’s why you should test every assistant on ten real tasks.

Virtual reality is the hype cycle in action: a peak of expectations, a long trough and, today, two poles of the market. Apple Vision Pro at $3,499 (on sale in the US since February 2, 2024; it has cost $3,699 since June 2026) and Meta Quest 3S at $299 (since October 15, 2024). The expensive headset shows what the technology can do; the cheap one, what people actually buy. For marketers, VR is still a niche: trade fairs and showrooms (a tour of a warehouse that hasn’t been built yet), not mass reach.

More interesting are smart glasses. Ray-Ban Meta (October 2023) have a camera, microphones and an AI assistant in ordinary frames. Meta Ray-Ban Display (on sale since September 30, 2025, $799 including the Neural Band wristband) added a small color display in one lens and gesture control from the wrist. In June 2026, Meta Glasses arrived at $299. If glasses become what the phone was, the content format will change: audio and short text in front of your eye, not a feed. For now, they’re at the early adopter stage of the Rogers curve, but keep an eye on them. Augmented reality (AR), on the other hand, is mainstream; it just isn’t called that: filters on Instagram and Snapchat, trying on glasses or furniture in an online store through your phone’s camera.

Wearables have crossed the chasm: Finland’s Oura sold 2.5 million rings by December 2024 and 5.5 million by September 2025. The Oura Ring 4 (since October 2024) costs €379 in Europe and from $349 in the US, and needs a subscription for full functionality. In October 2025, the company raised $900 million at a valuation of about $11 billion. For sports and health, it’s a gold mine of data (sleep, heart rate, strain). Watch out for GDPR: health data is a special category, and using it for marketing requires explicit consent.

LED walls have changed both stages and film studios. Sphere in Las Vegas (opened September 29, 2023, $2.3 billion): inside, a 16K-resolution screen with an area of 15,000 m². In film, the LED wall replaced the green screen: for the series The Mandalorian (2019), ILM built the “Volume,” with backgrounds rendered in real time. At conferences, an LED wall is now the standard instead of a projector: brighter, no shadows, but expensive, and it needs content in the wall’s native resolution. And esports as the new stadium: the Esports World Cup 2025 in Riyadh paid out $71.5 million in prize money across 26 tournaments; gaming and Twitch are covered in the Streaming chapter.

There are 6.04 billion internet users in the world (73% of humanity) and 5.66 billion social media user identities. Internet users aged 16 to 64 spend over an hour a day on classic social networks, and about 2 h 39 min a day including video platforms like YouTube and TikTok, which is 18 hours 36 minutes a week (DataReportal, Digital 2026). In Czechia, the AMI Digital Index 2026 (1,013 people aged 15 and over, fieldwork in May 2026) breaks this down by generation: Gen Z 223 minutes a day, and 95% of them open social media every day; millennials 139 minutes (152 last year), Gen X 110 (up from 106), boomers 92 (up from 88). For marketers: attention is the scarcest resource, and you compete for it with everyone who has an app.

The countertrend is just as strong. Since December 10, 2025, Australia has banned social media accounts for people under 16 (including Facebook, Instagram, TikTok, YouTube and X), with fines for platforms of up to 49.5 million Australian dollars; according to Wikipedia, getting around the ban is common, but the precedent stands. In fall 2024, the US Pew Research Center found (1,391 teens aged 13 to 17) that nearly half of them are online “almost constantly”; 90% use YouTube, about 60% TikTok and Instagram, 55% Snapchat. Feature phones, focus modes, phone-free schools: people are taking their attention back. Why this is so hard is explained by the dopamine loop in the Social Media chapter.

What this means for communication: less, but better. Limit how often an ad is shown (frequency capping), send the newsletter only when you have something to say, don’t turn on notifications by default, and measure unsubscribes as seriously as opens. Brands that respect people’s peace get attention when they ask for it. And for each of you personally: notifications off except from people, phone out of the bedroom, one morning a week without a screen. It’s a work skill, not magazine advice. This lesson is 90 minutes long; be honest, how many times have you looked at your phone during it?

Trends are a radar, not a steering wheel. The goal (what you want) and the audience (who it’s for) change slowly; tools and formats change fast. Of the six steps of digital strategy (goal → audience → message → channel → form → result, the Internet chapter), trends mainly affect channel and form. If you change your goal to follow a trend, you don’t have a strategy, just a feed. Running a team rests on cadence: a 30-minute retrospective every week, a tool review every quarter (what to cut, not what to add), a plan every year. Experiments get a budget (roughly a tenth of your time), an owner and a review date: after that, either into the core or cancel.

The principles that outlast trends are as old as trade itself: listen more than you talk; a relationship is more than a transaction, because trust pays interest; prepare, but read the room; people can tell when you don’t mean it; and empathy with direction: understand what people want, but lead them; don’t just repeat their wishes back. From this lesson, add: own your audience, have two channels, manual first and then automate, fewer tools, respect attention. None of this will go out of date, because it’s not about technology, it’s about people.

And one last principle from software development that works for marketing, too: never straight to production. Every change goes first to a test branch, then to a test environment (staging), and only after approval into live operation. Translated: try a new trend on a small audience first (one newsletter, one event, one month), measure it, and only then put it in your annual plan. The whole course in one sentence: marketing is a system, and AI is rewriting every layer of it right now; if you understand the system, you also understand what’s changing in it.