Growth Design
Delfi Paywall & Conversion
Rebuilding the subscription funnel, from the first paywall impression to the moment a subscriber tries to leave.
Many paywalls in Delfi portals from 2023 to 2025.
Scope
Paywall, checkout, onboarding, win-back
Platform
Piano.io
My Role
I lead design at Delfi and owned the conversion architecture end to end. I designed the checkout, onboarding, and win-back flows myself. A senior designer on my team built out the paywall variants and campaign library under my direction. Our product owner ran the user research, session analysis, and the A/B testing programme in Piano.io. Marketing wrote the copy.
OUTCOMES: Winning paywall variant: +26% purchases · Monthly churn: lowest in Delfi's history · Subscriber base grew from 92,000 to 125,000 so one-in-10 Estonians now have a Delfi subscription.
Context
Delfi is Estonia's largest digital news publisher. When it adopted Piano.io as its self-service subscription platform, the entire funnel was up for redesign at once: the paywalls that hit mid-article, the checkout, what happens after purchase, and what happens when someone tries to cancel.
The Problem
The original paywall was a single generic gate. It interrupted readers mid-article, then talked about subscription benefits instead of the article they wanted to finish. Checkout took several steps and lost people along the way. And when subscribers cancelled, nothing happened at all. They were simply gone.
The insight
Before Delfi I spent years in e-commerce conversion work, and one principle carried over: people don't buy subscriptions. They buy the thing in front of them with emotion. A reader hitting a paywall is at peak interest in one specific article. Generic benefit copy may waste that moment.
So every paywall had to mirror its context. A news paywall says "Find out what really happened." The recipe site says "Your delicious recipe is right here." The podcast portal invites you to listen to the audio. Marketing wrote each line, which I sometimes made shorter; I designed the placement system and template architecture that made context-matching possible at scale.
The A/B test
We ran a three-way test in Piano.io with roughly 290,000 impressions per variant: the original generic paywall, a benefits-list variant with price anchoring, and a contextual variant with article-reflective copy and imagery.
The contextual variant won with 26% more purchases than the original. The interesting part: it got fewer clicks. The people who did click were more committed. The design was filtering for intent rather than generating curiosity clicks, which is exactly what you want when the metric that pays is purchases, not engagement.
The three-way A/B test paywalls.
From one paywall to a system
A winning test is worth little if every new campaign needs a designer and a developer. The two-column template architecture turned the winning pattern into a system: same structure, different emotional hook.
General articles: "Find out what really happened".
Seasonal campaigns for summer and Christmas.
Delfi's birthday: limited free access with a live countdown, a pattern borrowed from e-commerce.
Main sub-brands with their own value proposition.
Oma Maitse (recipes) and Delfi Tasku (audio), each with its own variant.
Also continuous testing new ideas, like bulleting what are you getting from the article.
The senior designer on my team extended the system across all of these. My job shifted from designing screens to setting the standard and critiquing the work, which is how it should go.
The paywall architecture.
Checkout
Reduced to one screen. Email or social login, one button.
The email field does the thinking. Type an address and the system checks whether it belongs to an existing account: existing users sign in, new ones pay without creating an account first. Abandon the checkout and no half-made account is left behind, so nobody's first experience of Delfi is a password they never asked for.
Apple Pay and Google Pay joined the existing options, because a payment method someone doesn't have is just another exit. And since paying is the most anxious moment in the funnel, the screen answers the three questions people don't ask out loud: it's secure, you can cancel any time, and here is exactly what your package includes. The same promise the paywall made, repeated at the moment of payment.
Onboarding
New subscribers used to land on a homepage that looked exactly the same as before they paid. The four-step Minu Delfi (My Delfi) flow fixes that: enable notifications, pick newsletters from twelve curated options, choose favourite topics, and take a short tour of subscriber features like Delfi Tasku and read-later. Skippable at every step. Each step shows something the subscriber can use immediately.
The four steps of the onboarding flow, on mobile.
Win-back
When most cancellation flows ask "are you sure?" and throw a discount at everyone, ours asks why, and answers that.
First, a reminder of what leaving costs: how long they've been a member, shared reading access, what's coming up worth staying for. Then a short exit survey. Then a response matched to the stated reason:
Too expensive → a six-month discount offer
Wants ad-free reading → a cheaper ad-free package
Doesn't visit enough → the apps and social channels, keeping the relationship without the subscription
Technical problems → a direct line to report the issue, with an offer to stay while it gets fixed
If they cancel anyway, the final screen still surfaces the apps, the newsletter, and social channels. A cancelled subscription doesn't have to be a dead relationship.
After the new flow launched, monthly churn fell to the lowest level in Delfi's history, roughly a third below where it stood before the project.
Diagram of the winback flow.
Outcomes
Two results I can attribute to design. The +26% purchase uplift came from a controlled test where the paywall was the only variable. The churn reduction tracks the period after the win-back flow launched.
In subscription economics, churn compounds. Every cohort that stays longer raises lifetime value regardless of how many new subscribers marketing brings in.
Over the same period Delfi's subscriber base grew from roughly 92,000 to 125,000, close to one in ten Estonians paying for a single digital product. I won't claim that number only for design. It was successful teamwork between editorial, marketing, business development and pricing strategy. But the conversion architecture is the machinery all of it runs through.
