Something funny happened on the way to full automation.
'Slop' was named word of the year in 2025. By early 2026, outlets including CNN were describing the moment as the year of anti-AI marketing. Consumer enthusiasm collapsed: surveys now put the share of people who feel positive about AI at around 19%, down from roughly 50% two years earlier, with more than half of Americans reporting outright AI fatigue.
The turning point most people point to was Coca-Cola's AI-generated version of its Christmas advert in December 2025. The reaction was brutal. One widely shared line described it as the most profitable commercial in Pepsi's history. By spring 2026, brands including Aerie, Equinox and Almond Breeze were running campaigns that explicitly positioned themselves against AI slop, which tells you the backlash had become a marketable position in its own right.
I am not anti-AI. I use these tools every day and part two of this series is basically an argument for building your operations around them. But we should be honest about what happened. Everyone got the same capability at the same time. When everyone can produce infinite competent content, competent content stops being worth anything.
What actually got scarce
Economics is not complicated here. The value moved to whatever did not become abundant.
Original evidence. Not opinions about the market, actual data from your business. Benchmarks, survey results, aggregated anonymised customer numbers. A model cannot generate your proprietary data, and as I said in part one, statistics get cited and citations name their source. This is the highest-return content asset in B2B right now and almost nobody is doing it properly.
Named human expertise. Gartner expects a large majority of enterprise marketers to have built dedicated content authenticity functions by 2027, purely to stop brand dilution. Meanwhile buyer research consistently puts industry expertise ahead of both price and product fit as the deciding factor in vendor selection. Not 'thought leadership' as a content category. An actual named person who visibly knows the field.
Peer proof. G2's 2026 research: 92% of enterprise software buyers consult peer review platforms before shortlisting, 72% will remove a vendor over a negative review. Your reviews are simultaneously your most persuasive human proof and, per part one, a major machine citation source. If you have no systematic process for generating reviews, you have a gap in two channels at once.
Being in the room. Freeman's research found around 80% of B2B respondents naming live events as their most trusted marketing channel while roughly 74% distrust social media. Events convert to qualified pipeline at around 5.5%, against 0.5% to 2% for most digital channels. Nearly 60% of people now say they do not trust the authenticity of online content generally.
That last one is the whole story in a sentence. Trust in mediated channels fell, so the premium on unmediated ones rose.
What buyers actually trust in 2026
Freeman · G2 · consumer sentiment surveysThe bit I got wrong
I want to be straight about something, because I think a lot of us made the same mistake in 2024 and 2025.
I assumed the efficiency gain was the prize. Produce four times the content with the same team, and win. That logic worked for about eighteen months, and then it stopped, because the constraint was never production capacity. It was attention and trust, and both of those got scarcer precisely because everyone else also quadrupled their output.
What I would tell myself two years ago: use the tools to buy back time, then spend that time on things that do not scale. Customer conversations. A properly researched piece of original data. Showing up at the event. Actually writing the difficult positioning document instead of generating around it.
The teams I see winning right now published less this year than last year and more of it mattered.
How to run this in practice
Three rules I use with clients, and they are simple to say and hard to hold.
Keep AI invisible in the output and visible in the process. The brands taking damage are the ones who made AI a feature of what customers see. The ones doing well use it heavily in research, drafting, analysis and operations, and let the customer-facing work read like it came from people who know things. Nobody has ever complimented a company on its content pipeline.
Put names and faces on everything. Bylines from real people with real credentials. Founders and senior operators visible in the market. In a world where anonymous polished content is free, a named human with something to lose is the signal.
Publish your AI policy. Since 68% of enterprise buyers now weigh AI governance in vendor selection, a clear page about how you use AI, what humans review, and what you will not automate is a sales asset. Most of your competitors do not have one.
What the whole playbook comes down to
Three parts, so let me pull it together.
Your buyers now build a shortlist of three to five vendors using AI, mostly before you know they exist, and around 95% of the time the winner was on that list from day one. So being findable and quotable by machines is now a pipeline requirement, not a content project.
Your team gets smaller and more senior, systems absorb volume work, and the money moves from headcount to leverage. But the leverage only pays if you point it at the right market, which is a judgement problem, not a tooling problem.
And what you do with the time you save is what decides whether any of it works. Because everyone has the same tools now, and the only durable advantage left is being genuinely worth choosing and having enough real proof that both humans and machines can tell.
That is the playbook. It is less about AI than people expect. Most of it is the oldest advice in our profession: be clear, be specific, be somewhere your buyers are, and give people a real reason to trust you. The tools changed. The job did not.
If you want to talk through what this looks like for your company, that is what we do.
Related
Frequently asked questions
Is the AI marketing backlash real or overstated?
Real and measurable. Positive sentiment towards AI dropped from roughly 50% to about 19% in two years, over half of consumers report AI fatigue, and several major brands ran explicitly anti-AI campaigns in early 2026.
Should we tell customers we use AI?
Use it openly in your process and keep it out of the output as a selling point. Publishing an AI governance policy is increasingly valuable, since 68% of enterprise buyers now factor it into vendor selection.
Are events still worth it in B2B?
More than before. Around 80% of B2B buyers name live events as their most trusted channel, and events convert to qualified pipeline at roughly 5.5% against 0.5% to 2% for typical digital channels.
What content is worth producing now?
Original data, named expert perspectives and honest comparison material. Generic explainers have no scarcity value and no citation value.