In November 2024, Coca-Cola released an AI-generated remake of its iconic "Holidays Are Coming" Christmas commercial. The backlash was immediate, global — and unusually sharp in Germany. Critics across social media labeled it soulless, creepy, and devoid of creativity.

Coca-Cola did not retreat. In 2025, the company returned with a new AI Christmas campaign: roughly 100 team members, over 70,000 generated clips. Germany's reaction was worse than the year before.

This is not a story about one failed campaign. It is a story about what happens when AI enters advertising without craft — and what that moment means for every DACH brand commissioning video today.

What Coca-Cola actually built

The 2024 campaign tried to remake a commercial that European audiences had watched every Christmas since 1995. That original ad — real trucks, real locations, real actors — had become part of the seasonal fabric of European advertising culture. The AI version used four different generative AI models and three separate production studios.

German audiences read it immediately. Social media described the animation as inconsistent, with visuals switching between realism and cartoonish awkwardness. The word that recurred most was "soulless."

The 2025 campaign attempted to fix the problem through scale. Over 70,000 clips generated; roughly 100 people involved — five AI specialists at the core, with a broader oversight structure around them. The result was broadly received as worse than 2024.

Why volume without direction fails

The logic behind the 2025 campaign was understandable. More clips, more refinement, more human oversight — the assumption was that scale improves output. It does not.

Current video generation models have a structural limitation: they reproduce the appearance of motion without understanding the physics beneath it. Shape degrades at the edges. Small visual incoherences accumulate into something the eye reads as wrong — before the brain can name why.

The 2025 campaign's 70,000 clips could not overcome this. The problem is upstream of generation, not inside it. No amount of iterating inside the generation phase resolves a structural gap in how models process physics.

Seventy thousand clips cannot solve a structural limitation in the underlying models. What corrects it is direction — human judgment applied before and during generation, not after it. Volume without direction is iteration without a compass.

Coca-Cola's head of generative AI claimed the craftsmanship was "ten times better" in 2025. Audiences across Europe disagreed. The gap between internal assessment and public reception is where the real lesson lives.

Why Germany's reaction was different

The global backlash was significant. Germany's response was qualitatively different — and worth reading as a market signal rather than a cultural footnote.

Germany has a specific tradition in advertising craft. Germany's advertising market was valued at USD 31.74 billion in 2025 — one of Europe's largest. That scale has bred decades of elevated creative standards.

In Germany and Austria, craft quality in advertising is a baseline expectation — not a premium. Brand managers and agency creative directors work to that standard as a matter of course.

A campaign that looks unfinished does not just fail aesthetically in this context. It signals disrespect — for the brand's own heritage and for the audience's intelligence.

Christmas advertising carries particular cultural weight in DACH. It is the one moment in the advertising calendar where emotional expectations are highest. Craft failures here are most visible — and most consequential.

The "Holidays Are Coming" campaign had been part of the seasonal advertising fabric in Germany and Austria for three decades. Failing to meet quality expectations on that campaign had a compounding effect. The same miss on a less emotionally loaded brief might have passed more quietly elsewhere.

The Coca-Cola AI campaign was produced without the director-level art direction that makes AI output look considered rather than accidental. German audiences saw the absence of craft — not the presence of technology. That distinction is commercially important for any brand currently planning AI video production in DACH.

What the data says about AI video trust

Germany's reaction aligns with consumer research that has sharpened considerably since the backlash.

Animoto's 2026 State of Video Report found that 83% of consumers can now identify AI-generated video. The most common detection signals: robotic gestures (67%), unnatural motion (55%), and lack of emotional tone (51%).

Thirty-six percent said an AI-generated video lowers their perception of the brand behind it. That is more than a third of any audience actively discounting brand trust on first contact. The trust penalty is immediate — triggered at detection, before any other impression is formed.

The trust equation is asymmetric. eMarketer research puts it directly: visible AI in marketing is four times more likely to cost a brand trust than build it. Undetectable AI has a neutral to positive effect on perception. Detectable AI has a measurably negative one — and the detection rate is now at 83%.

The visibility problem is a craft problem

The backlash against Coca-Cola was not about AI being used. It was about AI being visible.

Washington State University research found a clear pattern: when AI is implied in content, emotional trust drops. Purchase intent follows. This is the consumer-side penalty for AI that surfaces on screen — rather than remaining invisible in the production process.

The Coca-Cola ad was detectable within the first few seconds of watching. Jittery motion, inconsistent visual grammar, and characters that felt slightly wrong without being obviously broken. That accumulation of small errors is the signature of AI generation without direction.

Direction is what prevents it. A creative director applies judgment before any generation begins — establishing look development, shot structure, visual logic, and character rules. The AI then executes within that framework.

The result is video that holds together as a coherent visual system — not a collection of generated clips. Without that direction layer, even 70,000 clips produce detectable AI. With it, far fewer clips produce something that reads as considered craft.

What "craft" means in AI production

The word "craft" is used loosely in advertising. In the context of AI video production, it has a specific technical meaning. Craft is the production infrastructure that makes AI generation commercially viable for brand campaigns.

The first element is art direction. Before a single frame is generated, the visual language of the campaign must be established. Look development, storyboarding, and reference imagery define what the AI is asked to produce.

Colour, texture, lighting, and character rules are all resolved at this stage — before generation begins. This phase is not optional. It is where the Coca-Cola campaign failed.

The second element is brand consistency management. A single AI-generated clip can look strong. A campaign of ten clips needs consistent characters, brand environments, and visual grammar across every shot and format.

This is achieved through LoRA training — fine-tuning AI models on a brand's specific visual language. Without it, clip-to-clip consistency breaks at campaign scale.

LoRA is not a feature inside a subscription tool. It is a production process a studio builds and manages per client.

The third element is post-production. Raw AI generation is not broadcast-ready. Compositing, colour grading, audio design, and format delivery require production expertise on top of the generation step.

Skipping post-production is visible — in exactly the way the Coca-Cola ad was visible to German audiences. All three elements require people: directors, art directors, and post-production specialists. The tools are infrastructure; the craft is the judgment that operates them.

What DACH brands should do differently now

The Coca-Cola campaign has done something useful for the DACH market. It has made the cost of AI without craft publicly legible.

Before 2024, brand managers considering AI video production faced an abstract risk. After two consecutive years of public backlash, they have a concrete, documented reference point. Global media coverage and measurable consumer trust data are now attached to it.

The Coca-Cola case has become the DACH market's benchmark for what to avoid. It is also a clear brief for what to do instead.

The right question is no longer "can you produce video with AI?" Every studio with a subscription can answer yes. The better questions are production questions.

Can you maintain visual consistency across a multi-shot campaign without reshoots? Do you use LoRA training for character and asset consistency?

What is your art direction process before generation begins? How do you deliver broadcast-ready formats in every channel specification?

A production company that clears these questions can turn a brief into broadcast-ready assets in 2–4 weeks. A studio that cannot answer them cannot guarantee the output is different from what Coca-Cola produced.

These questions distinguish a production company from an AI operator. The answers determine whether output reads as detectable AI or as a strong campaign. Germany is projected to lead AI video growth in Europe through 2034.

More DACH brands will commission AI video in the next 12 months than ever before. That growth means more decisions about which production partner to trust. The Coca-Cola moment has set the standard: craft is not optional in this market.

The answer Trippy Pictures built

Disclosure: Trippy Pictures is our own studio. We have included this section because the production approach this article argues for is precisely what we have built — but you should know we are not a neutral party.

Trippy Pictures is an AI-native commercial video production studio based in Vienna and Berlin. It is a joint venture between Kaiserschnitt Film and AI creator Andries Ohneisser. Kaiserschnitt Film brings 25 CCA awards, one Cannes Lion, and 15+ years of commercial production in DACH.

Ohneisser brings AI-native workflow depth and 510,000+ followers across Instagram, TikTok, and YouTube. Together, they produce broadcast-quality commercial video using an AI-native pipeline — directed by people, accelerated by AI.

Every Trippy project begins with art direction. Look development runs before any generation starts. LoRA training encodes each client's brand visual language directly into the AI models used in production.

Post-production handles compositing, colour grading, audio finishing, and broadcast delivery in every required format. Clients receive finished assets — not raw outputs.

Delivered clients include Samsung, Verbund, and Ökostrom — published, approved campaigns for DACH enterprise brands and their agencies. The Wien Nord Serviceplan and Samsung Austria engagement is the reference model for white-label AI production in this market.

The Coca-Cola backlash happened because AI was given a production brief without a director. Trippy was built on the opposite premise: directors lead, AI executes. Every campaign starts with a brief, proceeds through art direction, and arrives at broadcast-ready delivery — not the reverse.

Frequently asked questions

Was the Coca-Cola AI ad genuinely bad, or was the backlash disproportionate?

Both parts of that question are true at once. The public reaction was real and consistent across markets. The underlying generation technology is also genuinely capable.

The problem was not the models — it was the absence of creative direction, which made the models' limitations visible. A directed production using the same technology would produce a different result. The backlash was proportionate to the absence of craft, not to the use of AI.

Why was Germany's reaction sharper than other markets?

German advertising culture places high weight on creative craft, and Christmas advertising carries exceptional emotional weight in DACH. The "Holidays Are Coming" campaign had been part of the region's advertising fabric for 30 years. Failing to meet quality expectations on that campaign had a compounding effect in Germany and Austria.

The same miss on a less emotionally loaded brief might have passed more quietly elsewhere. DACH audiences are also attuned to visual craft in advertising — the inconsistencies were visible early.

Does this mean DACH brands should avoid AI video production?

No. The lesson points in the opposite direction. AI production done correctly delivers broadcast-quality content at a fraction of traditional cost and timeline. Human direction, LoRA brand training, and professional post-production are what "correctly" means here.

The Coca-Cola case shows what AI video looks like without that infrastructure. It is an argument for doing AI production properly, not for avoiding it. The brands that move now with well-directed AI campaigns will hold a first-mover advantage in DACH before competitors arrive.

What is LoRA training and why does it prevent the Coca-Cola problem?

LoRA training (Low-Rank Adaptation) fine-tunes an AI model on a brand's specific visual language. Characters, product appearance, colour systems, and style are all embedded in the training. Every generation from that model then reflects the brand by default.

Without LoRA, maintaining visual consistency across a multi-shot campaign requires precise repeated prompting that breaks at scale. LoRA is what makes brand-consistent AI campaigns commercially viable — and it is not a subscription feature.

What should DACH brands ask when evaluating AI production partners?

Three questions matter most. First: what is your art direction process before generation begins? Second: do you use LoRA training for character and asset consistency across shots?

Third: what does your post-production workflow include, and how do you deliver broadcast-ready formats? A studio that cannot answer all three specifically is an AI operator with a website, not a production company. Clear, specific answers — backed by named client examples — are the signal that real production infrastructure exists.

How does a properly directed AI campaign compare to traditional production on cost and timeline?

Traditional professional video production runs approximately €4,500 per finished minute, covering crew, equipment, locations, talent, and post-production. Director-led AI production brings that to approximately €400 per finished minute — at the same cinematic quality ceiling.

On timeline, traditional production runs 4–8 weeks from brief to broadcast-ready delivery; director-led AI production runs 2–4 weeks. For DACH brands running three to four campaign updates per year, both the cost and timeline differences compound significantly.

Conclusion: What the backlash changed

The Coca-Cola Christmas AI campaign gave Germany's advertising market a shared reference point. That reference point has permanently raised the bar for AI video in DACH.

Eighty-three percent of consumers can now identify AI-generated content. Thirty-six percent downgrade brand perception on contact with it. In a market that has historically valued craft in advertising, the window for careless AI production has closed.

The brands and agencies that move now — with properly directed, professionally produced AI campaigns — will not face a Coca-Cola problem. They will face a first-mover advantage: broadcast quality at AI economics, before competitors find the right production partner.

Germany is the fastest-growing AI video market in Europe. The opportunity is real. The standard is clear. Craft is not optional.

If you are a DACH brand ready for AI video production, get in touch with Trippy Pictures. If you are an agency building AI production capability, the same applies. Samsung, Verbund, and Ökostrom are produced and published — the conversation starts here.