The morning at a glance
The morning had three parts. Maris Tuvikene, Campaign Measurement Lead at Nepa, shared what more than 1,700 campaign evaluations reveal about the best and weakest campaigns. Andreas Nordfors, Managing Director for Marketing Acceleration at Nepa, showed how AI now turns creative data into a formula for success. A panel with Laura Penttilä (Valio), Katariina Kukkonen (Metsä Tissue) and Elizabeth Harjunpää (ex Wolt), moderated by Nepa’s Kiia Rajaranta, closed the morning with a discussion on the evolving role of insights.
Safe is the new risky
The biggest risk in advertising is not a bold ad that divides opinion. It is an ad that makes people feel nothing. Maris Tuvikene compared the best and weakest quarter of more than 1,700 campaign evaluations from 2019–2026, each ranked against comparable campaigns in the same region, sector and period. Around half of viewers feel no energising or surprising emotion after seeing an ad, and that holds for good, consistent, well-known brands too.
Activation draws on the brand you have already built
The campaigns that best drive short-term action share a well-known brand, an ad that gets noticed and is seen as coming from that brand, a clear message and a good brand fit. More media types raise attention, but well-known brands start from a clearly higher level. Between the best and weakest activation campaigns, the media mix differed only slightly.
Seen is not enough
On average 47% of viewers notice a campaign, but only 19% also know whose it is. That drop loses six in ten viewers before the message even has a chance. After that point few drop off: 15% get the message, 12% like the ad and 10% are moved by it. Distinctive, consistently used brand assets are what close the gap, a theme Nepa explores in the webinar Stop building your competitors’ brands.
Brand building is not reserved for big brands
For long-term brand effect, the best campaigns did not differ from the weakest in how well known the brand was or how clearly it was recognised. The difference came from the creative: people liked the ad and it stood out from other advertising. Fitting the brand mattered less, which suggests brand-building work can stretch beyond the brand’s usual comfort zone.
Liking is the shared ingredient
Ad liking separates the best from the rest for both activation and brand effect, and it is earned by the creative. The emotional picture backs this up. High-energy positive feelings such as happiness, enthusiasm and surprise have the strongest link to both effects. Negative feelings cost less than many fear. Neutral reactions – unaffected, indifferent, bored – are both the most common and the most costly. Nepa measures these reactions with its Emotion Palette.
Roots and blooms
Tuvikene summed it up with an image of roots and blooms. Consistency is the roots: built over time, so people know it is you and it feels like you. Creativity is the blooms: liked, standing out and full of energy. Roots alone feel safe, and blooms alone are anonymous. Her advice: build the brand consistently, then dare to stand out – the same case Nepa makes in Why playing it safe is holding your brand back.
In the Q&A, a challenger brand asked where to start. Recognition comes first, so that the effect of every ad lands with the right sender, followed by building the brand in the same consistent way over time.
AI turns creative data into a formula for success
AI can now explain why some creatives perform and others don’t, at a scale no analyst team could match. Andreas Nordfors estimated that AI is roughly ten times more capable than a year ago. Creative analysis has moved from spotting objects in ads to holistic analysis of hooks, storytelling and mood, and is heading towards autonomous systems that generate, test and refresh creatives continuously.
Nepa’s Creative AI approach works in three steps:
- Find creative patterns at scale. AI classifies every creative: basic video properties, faces, logos and on-screen text, and with multimodal models also what is said, whether a pain point is raised and how strong the call to action is. It can even track brand-specific elements, such as how long an app is shown on screen.
- Connect patterns to performance. The classifications are linked to outcomes, from digital metrics such as cost per acquisition to pre- and post-test results from campaign evaluation.
- Establish the formula for success. The result is a clear picture of which features lift or hurt performance, turned into playbooks and instant creative scoring – the core of Nepa’s Visual Performance work.
The Wolt case on TikTok
The Wolt case on TikTok showed what this looks like in practice. Across more than 3,000 creatives from one year, the strongest acquisition drivers were showing the product and app experience, visible people and faces, and a lo-fi, user-generated feel. A low frequency of the brand logo worked better than a heavy one, and a video that focused too much on the food itself performed clearly worse. The findings became a playbook for Wolt’s creative teams, markets and influencers, and a scoring tool for new ideas: the best creative scored 84, the weakest 7. You can watch the case as presented at IIEX Europe 2026 or read our takeaways from the conference.
Where AI meets its limits
The audience pushed on the limits. Can AI judge humour and emotion as well as a panel of real people? Andreas’ answer: it depends on the data it learns from. Scoring an ad against generic best practice says little; scoring it against your own history of measured campaign effects says a lot. And AI raises the bar rather than replacing creativity. It can optimise whether a pizza or a burger goes on top, but the bold idea that changes the format still comes from people.
Panel: the evolving role of insights
AI gives organisations more data and faster answers than ever, which makes human judgement more important, not less. That was the shared view of the panel, where Laura Penttilä (Valio), Katariina Kukkonen (Metsä) and Elizabeth Harjunpää discussed with Nepa’s Kiia Rajaranta how knowing turns into doing inside organisations.
A good insight changes a decision
The panellists agreed that an insight is more than an accurate fact about people. It explains why something is happening, what it means for the business and what to do next. A brand metric that rises is data; knowing why it rose is insight. Findings that merely confirm what you already believed are nice to know, while the valuable ones shake up existing thinking.
Learning should be continuous, not one-off
Systematic, ongoing measurement such as brand tracking reveals long-term trends that single studies cannot. The panel described insight teams that never start from scratch but build on everything already known. The next challenge is moving that accumulated knowledge from people’s heads into data and systems, so it stays accessible to everyone.
AI needs shared definitions and quality control
If everyone in an organisation asks an AI tool what an “active customer” is and gets a different answer, conclusions start to conflict. Elizabeth Harjunpää, who has worked at both Nepa and Wolt, stressed the need for a common data structure and common definitions, for example in segmentation. The insight function’s role becomes quality assurance and interpretation: spotting incomplete data, and remembering that consumer survey data is what people say, not absolute truth.
Faster analysis frees time for thinking
AI can turn weeks of analysis into hours, leaving more time for conclusions and next steps – if you protect that time. But the panel warned against accepting conclusions without knowing what they are built on. Routine work can be automated once you understand how it is built; strategic thinking still needs dedicated time, and a clear vision of where the organisation is heading.
Closing thoughts
Their closing thoughts tied back to the morning. Clarify what really matters and do fewer things well. Maris Tuvikene’s roots-and-blooms image struck a chord: a consistently built brand is a strength, but it cannot be rested on. The real question is how to renew without losing what made the brand recognisable. As one panellist put it, the need for insight has not changed in decades, only the tools have. Turning tracking data into agreed decisions is exactly what Nepa’s BrandGrow Program is built for.
A deep dive into the panel discussion, with more of the panellists’ own voices, follows in our next post.
Talk to a Nepa expert →
