Superweek 2026 – Day One

From Castle Views to Cognitive Work: What Really Matters Now

Day One of Superweek 2026 opened with literal perspective. Morning views from Visegrád Castle over the Danube set the tone: this was always going to be a day about zooming out, questioning assumptions, and getting honest about where digital analytics is actually heading – not where vendor decks say it is.

Across agencies, brands, platforms and tool builders, a few themes kept surfacing:size versus speed, confidence versus accuracy, automation versus judgment, and measurement versus meaning.

Here’s a grounded, practitioner’s read of the day.

Big Agencies, Small Ideas

Doug Hall – Duga Digital

“Big Agencies Kill Innovation: Size Matters and Smaller is Better”

Doug’s talk was an unapologetic open letter to anyone who still equates scale with quality.

After three decades in digital, his argument was simple and uncomfortable:every acquisition promises leverage and ends in gravity. Innovation doesn’t fail because people stop caring. It fails because incentives change. Spreadsheets win. Politics creep in. Decision latency explodes.

Smaller agencies, by contrast, survive by staying sharp. They are:

• Unshackled from internal empire-building

• Forced into real collaboration rather than performative alignment

• Close enough to delivery that mistakes are visible and learning is fast

Doug wasn’t being nostalgic. He’s lived on both sides. His point wasn’t that big organisations are evil – it’s that structure determines behaviour. If you want experimentation, speed and truth, you don’t design for comfort.

Big doesn’t mean better. It often just means slower decay.

When the Machine Knows the Answer

Ibrahim Elawadi – Phillips

“The Last Analyst Standing: Thriving When AI Rewrites the Rules”

If there was one talk that rewired the room, this was it.

Ibrahim opened with a sobering line:

We work in an industry where ‘long term’ now means next Tuesday.

He challenged the lazy comparison between AI and previous revolutions (industrial, computing). This isn’t just outsourcing muscle or calculation – we’ve done that before.

This is outsourcing logic.

He referenced the CEO of Anthropic suggesting we may be 6–12 months away from AI generating most software engineering. Tellingly, Anthropic isn’t hiring narrow specialists – they’re hiring generalists who can reason, challenge and adapt.

The most powerful section of the talk was a live demonstration of confidence versus correctness.

• Ibrahim clustered responses from 1,000 “experts” using AI into a 3D visualisation

• Agreement was high. Confidence was high. Accuracy was not guaranteed.

• Bias wasn’t a bug – it was the engine

He then did something clever: he gave the AI a backstory.“You are an SEO who nearly lost their job five years ago.”

Suddenly the outputs became more confident, more opinionated, more human. But not more correct.

The real breakthrough came when he flipped the question:

“What would prove your theory wrong?”

That single change turned AI from an answer engine into a testable thinking partner.

His conclusion landed hard:AI-enhanced productivity is not a shortcut to competence.

The analysts who survive won’t be “prompt jockeys”. They’ll be architects:

• Designing better questions

• Building context

• Interrogating answers

• Staying cognitively engaged

Redundancy, he argued, isn’t hypothetical. For large parts of today’s skill distribution, it’s an 18–24 month clock.

Purpose Is a Performance Multiplier

Marie Fenner – Piano

“What Gets You Out of Bed in the Morning?”

Marie brought the room back to humans.

Her session explored what actually drives sustainable performance: pay, promotion, autonomy, customers, life outside work, and – critically – purpose.

The data was clear. Purpose-driven companies don’t just feel nicer to work for.

They:

• Scale more effectively

• Adapt faster to disruption

• Outperform over time

This wasn’t a soft talk. It was a reminder that optimisation without meaning burns people out and businesses down.

Purpose isn’t a brand statement. It’s an operating system.

What If You Actually Had All the Data?

Russell McAthy – Ringside Data

“100% Performance: What Happens When You Actually Have All the Data”

Russell tackled a problem everyone recognises but few confront honestly: the widening gap between what organisations think they know and what’s actually happening.

Cookie banners are still awful. One German client saw only a 37% acceptance rate. Entire journeys disappear.

Russell explored what happens when constraints on data collection are removed, including:

• POS data flowing via APIs and webhooks

• The challenge of user identity without a clean key

• Legal and ethical questions around PII at point of sale

A key takeaway: offline-to-online attribution isn’t just a technical problem. It’s a governance, consent and value-exchange problem.

This talk quietly reinforced a recurring theme of the day: first-order solutions create second-order consequences. Delivery choices matter more than diagrams.

When Traffic Isn’t Human

Matteo Zambon – Tag Manager Italia

“The Era of AI Agents: Adapting Your Measurement Strategy”

Matteo made it uncomfortably clear: web traffic is no longer mostly people plus dumb bots.

Autonomous AI agents now browse, click and behave like humans. Detection is an arms race – and one we’ll never “win” conclusively.

His approach combined:

• Behavioural analysis

• Server-side logic

• Real-time bot control

One particularly elegant idea:default a request flag to usingjs=false, flip it only when JavaScript genuinely executes. No JS, no session. Pair that with header inspection and server-side routing and suddenly you have a probabilistic signal rather than false certainty.

The key message wasn’t tooling. It was humility.There will never be a standard that identifies 100% of bots. Measurement now lives in gradients, not absolutes.

Asking Customers Instead of Guessing

Ezequiel Boehler – Conductrics

“Integrating Customer Feedback Into the Optimization Flywheel”

Ezequiel delivered a much-needed correction to CRO culture.

Analytics exists for a reason. Sometimes that reason is not another model – it’s asking a human being what they think.

He showed how surveys, experimentation and optimisation reinforce each other when aligned properly. Not vanity testing. Not “number of tests per quarter” theatre.

One stat stuck:

• 22% win rate without customer research

• 43% win rate with it

That delta isn’t tooling. It’s intent.

He also warned about metric obsession. Optimising only for short-term conversions erodes trust. Facebook’s long unwind of reaction weighting was a cautionary tale: engagement without alignment creates systemic damage.

CRO, he reminded us, is not just A/B testing. It’s intentional decision-making under trade-offs.

Engineering Better Signals

Gunnar Griese – 8-bit-sheep

“Signal Engineering: Moving Beyond Basic Conversion Measurement”

Gunnar closed the day by tackling why ROAS keeps declining even when “everything is tracked”.

Most organisations still feed platforms blunt signals: transactions and revenue.

Algorithms optimise exactly what they’re given – often to everyone’s surprise.

His Signal Engineering Maturity Framework pushed teams to think in proxies:

• Soft conversions over final transactions

• Timing aligned to buying cycles

• Suppressing unhelpful signals entirely

• User-value signals correlated with long-term LTV

One important warning: sending platforms data they’re not designed to handle can backfire. Signal design must scale operationally and cognitively. Marketing teams must understand and trust it.

Measurement isn’t about more data. It’s about better incentives.

Day One, Distilled

Across wildly different talks, Day One converged on a few truths:

• Scale changes behaviour – often for the worse

• AI amplifies confidence faster than accuracy

• Purpose is not fluff, it’s infrastructure

• Measurement lives in trade-offs, not perfection

• The future analyst thinks, tests and challenges

As someone put it later in the day:agreement is not accuracy.

Superweek 2026 didn’t open with hype. It opened with responsibility. And that feels like exactly the right place to start.

More to come tomorrow.