Data Science × Publishing

AI in publishing: Your readers have long shown
what they would subscribe to

6 modules. 6 subscription problems every media house knows. 6 data-driven solutions built on the data you already have - paywall logs, usage data, newsletter clicks, subscription history and CRM. No external data, no new infrastructure.

Calculated revenue potential per year
€0
Model calculation for a sample publisher with 25,000 digital subscriptions · conservatively estimated

Subscription numbers you can trace back

Based on a regional media house: €42M revenue, 55,000 print and 25,000 digital subscriptions, 900,000 unique users per month. Every figure is derived from the documented assumptions - the levers come from published industry benchmarks.

6
Modules
25,000
Digital subscriptions in the sample publisher
€760,000
Revenue potential/year
4 wks
Pilot timeframe

Each module solves a concrete subscription problem

Click a module for the full use case with visualisations, transparent model calculation and business impact.

Revenue potential by module - overview

The subscription growth simulator

Set the sliders to your scale: the simulator projects the digital subscription base over 24 months - once with your current figures, once with the levers of the six modules.

Digital subscriptions today
Cancellation rate per month
ARPU per month
Paywall conversion
Additional subscriptions after 24 months
Additional annual revenue
Cancellations avoided in 24 months
Digital subscription base over 24 months - status quo vs. with AI modules

Model calculation, not a forecast: new subscriptions per month = paywall contacts × conversion; paywall contacts are set at 6 times the subscription base. The module levers match the assumptions derived in the modules: cancellation rate -20% relative (Modules 01 and 05), paywall conversion +30% (Module 02).

No theory - results in weeks

We work with the data you already have. No vendor lock-in, no cloud mandate, no hidden costs.

📰
Industry knowledge + data science
We understand metering, churn curves and circulation seasonality. Our models solve real publishing problems - on your infrastructure, with your data.
Results before perfection
We deliver a pilot dashboard with real numbers from your data. No 6-month concept - you see the ROI before you invest.
🔒
Your data, your systems
Paywall logs, usage data, newsletter clicks, subscription history and CRM - we connect what already arises in your house. No new system, no external data, GDPR-compliant.
Note on the context of this portfolio

The following six modules use a fictitious regional media house (55,000 print and 25,000 digital subscriptions, €42M revenue) to show how data science can be applied across the subscription business - from the dynamic paywall to winback.

All euro figures are model calculations. They are derived transparently from clearly named assumptions (assumption baseline → lever → result). The levers follow published industry benchmarks (including Mather Economics, INMA, FT Strategies, Piano) and are deliberately conservative - no promises.

In a real project, knowledge of your house is the decisive factor: How is your paywall configured today? What data sits in the subscription system, CRM and web analytics? Where do you lose subscribers - at the start, in the base or on price? On that basis we build models that fit your reality.

This portfolio shows which questions data science can answer in publishing - the concrete answers emerge only with your data.

Where the industry figures come from

The levers in the model calculation draw on published case studies from other publishers. Every external figure cited in the text is documented here with its original source - to read and to verify.

  • Financial Times - RFV score (Recency, Frequency, Volume): a user counts as "engaged" from a score of 18.2, which cuts the cancellation rate by about 10%. WAN-IFRA, 2023 · FT Strategies
  • NZZ - propensity paywall from 100 to 150 data points; conversion rate increased fivefold in three years. Digiday, 2018
  • Schibsted - +75% subscription sales from ML-selected front-page articles (A/B test versus the previous model). WAN-IFRA, 2025
  • INMA - propensity-steered paywalls achieved +77% (meter) to +166% (freemium) conversion in A/B tests versus the existing solution. INMA / Piechota, 2020
  • Piano / dmgMedia - highest vs. lowest propensity segment: a factor of 174 in likelihood to subscribe. Piano, 2022 (PDF)
  • Piano - first-month cancellations run roughly three times as high as in the third month (benchmark data). Piano Subscriber Journey Guidebook, 2022 (PDF)
  • Piano - strongest promotional periods, with holiday offers, between November and January. Piano benchmark
  • Mather Economics - a targeted retention intervention cut churn by 17% in an A/B test (1.44% vs. 1.73%). Mather Economics
  • Mather Economics - segmented pricing: +8% digital revenue at only +0.4% incremental churn; segmented 2.5% vs. blanket 5.7% incremental churn. Mather Economics, 2025
  • Mediahuis Aachen - price increase (6% digital / 7% print) with near-zero churn impact. Mather Economics
  • Amedia - its most-produced section (culture) read by only 2.6% of subscribers; +5.6% subscription growth after the realignment (with a live-sports product). INMA, 2018
  • Amedia - systematic onboarding cut first-week cancellations from 25% to 18%; loyalty-sorted win-back calls: 12% reactivation. INMA, 2020
  • Newsday - onboarded new subscribers at 4.5% vs. 8.1% churn: about 44% less attrition. INMA, 2024
  • Wall Street Journal - 100-day onboarding: 18% higher retention for onboarded new members. INMA, 2020
  • ZEIT Online - revised trial journey: +12% transitions to paid subscription, over 7,000 additional subscribers per year. INMA, 2022

The figures from other publishers come from their published case studies and serve as plausibility anchors for the levers - they are not transferred to the sample publisher. Every EUR figure on this page is a model calculation on the sample publisher assumptions named above. Sources accessed: 13 July 2026.

Ready to put your data to work?

We work with the data you already have. No vendor lock-in, no cloud mandate, no hidden costs.

Schedule a data workshop →