Day 1 of Piano Academy 2026
Day one of Piano Academy 2026 in Amsterdam covered AI products, paywall strategy, data quality, and the revenue publishers lose to failed payments. Here's what five sessions shared.
Hilary Mason: building great products that combine AI and human creativity
Hilary Mason, CEO of Hidden Door, opened with a case for treating AI as a leadership question. She founded the applied ML research lab Fast Forward Labs in 2014, which Cloudera acquired in 2017.
Her argument: "AI is now a leadership issue – not an IT project." It touches customers, operations, risk, and talent. She called it the first technology that challenges every layer of an operation at the same time, and no company is set up for that. Model people ignore the product, product people ignore the business, the board demands "AI," and IT runs the data. As she put it, "You have to consider the whole burger."
She was honest about the models' limits. LLMs are biased to the mean, so outputs come out "sames-y." They're good at homework and bad at creative work. In her words, "Every output is a hallucination. Some of them happen to be correct."
Her practical advice:
Start with your customer.
Much of what makes a company "AI ready" is what employees already need: clean database columns, documented data and definitions, and clear evaluation criteria.
"Building is nearly free. Maintaining is costly." Buy the plumbing, build the context. Open formats, open source, and portable business logic are how you keep leverage, because "Software should work for you, not the other way around."
Protect your taste and judgment, and build systems that encode it.
She closed with: "AI is not creative. But we are. AI works with us. To make our work scale, build beautiful things, for people."
PD Digital: using the full stack to turn data into revenue
Jürgen Risinger, Head of Subscription at pd digital Hub, joined Lucia Juelke, Director of Account Management at Piano. pd digital Hub is the central digital unit of Mediengruppe Pressedruck, behind brands like Augsburger Allgemeine, Main-Post, Südkurier, and Allgäuer Zeitung.
The starting point was a market shift. Piano Benchmarks for 2023–2025 show search traffic down 36% and new conversion revenue from search down 16%. Inside the business, the newsroom, ad sales, and reader revenue teams all fought for the same top placements. The result was a loud, overloaded site that missed targets for every department.
Their answer is Revenue Driven Design. Instead of one layout for everyone, the site adapts to each visitor based on five loyalty clusters. As Jürgen put it, "Revenue Driven Design helps us monetize our visitors as effectively as possible, based on their loyalty." Segmentation, targeting, and analytics tools, including Piano's Ad Revenue Insights, decide which template, content, ads, and subscription offers each user sees.
The results beat their own hypotheses:
About 27% more time spent per subscriber per day (they'd expected 15%)
About 11% more page impressions per subscriber per day
32% more revenue per page view (they'd expected 10%)
About 17% higher average CPMs and about 27% better ad visibility
Two learnings stood out. On augsburger-allgemeine.de, the most loyal 3.9% of users drove 28.5% of advertising revenue (January–May 2026). And subscribers account for nearly 10% of total digital advertising revenue.
Clear data from Piano Analytics also means fewer disputes between ad operations and reader revenue. In Jürgen's words, "Decisions are no longer based on opinions, but on Revenue Driven Design."
To try it yourself, start with the three questions behind Revenue Driven Design:
What is the user's intention at a given moment?
What exactly do we expect the user to do at a specific moment?
What is the conversion goal of the page the user is visiting?
BBC Studios: scaling the BBC subscription product
George Agate (Lead Marketing Analyst) and Joseph Iles (Lead Subscriptions Analyst) walked through 12 months of growth. BBC Studios, the BBC's commercial arm, launched a paid subscription for US audiences in June 2025. Today it's live in the US and Canada on web, iOS, and Android.
From metered to dynamic paywall. The team scaled a metered paywall from 10% to 40% of the audience, then switched to a paywall based on Piano's Likelihood to Subscribe (LtS) model. In late 2025, they tested it against Piano's Dynamic Paywall, which did so well they ended the test early. It more than doubled the number of people who saw the paywall, and ad revenue per user held steady. As the BBC Studios team put it, "Dynamic Paywall didn't just optimize existing audiences. It uncovered new subscription opportunities that propensity modeling alone could not identify."
Breaking news. At first, a 40% traffic spike came with a 25% drop in daily new subscriptions. During big events, the site focused on live news pages. After the team added onward journeys from those pages, news traffic doubled during the next spike, and so did daily new subscriptions.
Native apps. The paywall went into the BBC app in January 2026, with Composer controlling who gets access and how the paywall appears, and Apple and Google handling payments. A/B tests put the best meter at about four page views, and 42% of active subscribers converted in the apps. In the team's words, "Native apps quickly became a meaningful subscription channel."
Retention. After moving billing to Stripe and turning on Stripe Smart Retries in August 2026, dunning success rose from 43.7% to 98.5%. "The migration strengthened our payment infrastructure and almost eliminated involuntary churn," the team said.
Marketing. The team moved from lower-funnel-only ads to a full-funnel strategy, sending high RevOps score signals to Meta and Google through a Conversion API. Early results: 200K+ weekly visitors from upper-funnel campaigns, time on site up from 1m 51s to 2m 26s, twice as many visits with 2+ pageviews, and $1.80 in revenue for every $1 spent on marketing in 2026.
Next up: a registration-to-subscription journey, Dynamic Paywall in the native apps, and Stripe Checkout.
Janus de Visser: when everyone uses the same AI, your data makes the difference
Janus de Visser, Chief Product Officer at LevelUp Group and founder of Cloud Nine Digital, looked at what happens as marketers hand decisions to AI.
Adoption is already high. 80% of Google advertisers use AI solutions, and 8 million advertisers use Meta's. According to IAB and BWG Global's State of Data 2026, 50% scale AI in marketing measurement and 70% expect to. But 75% say these solutions underperform on trust. Concerns include AI accuracy and transparency (49%) and data quality and accessibility (45%)
His point: major platforms are black boxes with limited model access, and your competitors use the same ones. What you control is the data you feed them. Every layer, from your website and CRM to the data layer and tag manager, can introduce errors, and automation amplifies them.
His framework has three steps:
Map: know your data sources, routes, and blind spots.
Monitor: automate validation and keep continuous control.
Mature: strengthen the foundation, then enrich and scale.
Client results on his slides: 60% obsolete tracking found, a 26% cost reduction, a privacy risk resolved, and campaign performance restored after an add-to-cart tracking failure. As Janus put it, "That advantage requires reliability that is repeatable and scalable."
Stripe x Piano: the silent growth lever
Tim Rowell, SVP of Subscriptions at Piano, and Jana Aleksic, EMEA Customer Success Manager at Stripe, focused on revenue publishers have already earned and then lose to failed payments.
Piano benchmark data shows the scale of the problem:
About 1 in 3 churned subscribers is lost to a failed payment, not a decision to cancel.
Involuntary churn costs subscription clients 10%+ of revenue every year.
11.5% of annual renewals are lost to payment failure, even after a grace period.
2.4% of monthly terms are lost per renewal cycle.
Part of the issue is ownership. As Tim and Jana put it, "In media subscription businesses there is not enough focus on passive churn. It is discussed but not owned." Yet every recovered payment is revenue with zero acquisition cost.
The fix depends on why a payment fails. Network tokens and card account updaters keep card details current, Smart Retries time each retry to when it's most likely to succeed, and targeted dunning handles cases that need the subscriber to act. Payment, billing, messaging, and access also need to stay in sync, so a subscriber in recovery doesn't lose access or get the wrong email.
Stripe Billing is now Piano's billing engine. As of September 28, 2026, 891,279 subscriptions renew on Stripe Billing through Piano, and 9,999 of every 10,000 switches in September went through without an error. Across paired apps (median), Smart Retries recovered 55.1% of failed renewals, up from 50.4% with Piano's previous retry logic. The switch itself takes about three days.
Their four questions to take back:
What are your first-attempt and final authorization rates by market and renewal type?
Why do your renewals fail, and which causes can you influence?
What percentage of initially failed payments do you recover?
During recovery, do billing, messaging, and access stay synchronized?
Piano Academy is Piano's annual conference for customers, partners, prospects, and industry professionals. This year's fifth edition is running October 6–7 at the Beurs van Berlage in Amsterdam, built around the theme "Navigate What's Next: Driving Monetization in the Age of AI."
Day 2 recap is coming soon.
