This is Part 3 of our Media Growth Intelligence series. Part 1 covered what your audience is actually worth. Part 2 covered the engagement milestones that predict who stays. Now: detecting who will expand — and who’s about to leave — 30-90 days before it shows in revenue.
Acquisition tells you how many people found you. Activation tells you who engaged deeply enough to stick. But neither answers the question that drives sustainable media economics: what’s about to change?
The two metrics in this article are forward-looking. One detects expansion signals — the behavioral patterns that precede upgrades, referrals, and spend increases. The other detects engagement decay — the slow fade that precedes churn by weeks or months. Most publishers react to cancellations after they happen. The ones that retain profitably are reading the signals early enough to intervene.
Quick glossary: Expansion signals = behavioral indicators that an audience member is about to increase their value (upgrade, refer, sponsor, increase engagement). Engagement decay = the measurable decline in interaction frequency, depth, or breadth that precedes a cancellation or lapse event. NRR (Net Revenue Retention) = (Starting ARR + Expansion - Churned - Contraction) / Starting ARR — the single number that captures whether your existing base is growing or shrinking.
Metric 7: Leading Indicators of Expansion
What it measures: Behavioral signals that predict which existing audience members will increase their value within the next 60-90 days, across four expansion vectors.
1. Upsell / Upgrade
For newsletter publishers: Free subscribers clicking premium content teasers, visiting pricing pages, or engaging with upgrade prompts in emails. The conversion window is typically 30-60 days from first premium click. Real median free-to-paid conversion is 2-3%, but top performers exceed 5% — and the gap is almost entirely explained by how well the free product advertises the paid product.
For ad-supported media: Anonymous readers creating accounts, registered users deepening session depth, newsletter subscribers engaging with premium content. Each step is a 2-5x ARPU multiplier. Piano’s data shows registered users convert to paid at 19% versus 2% for anonymous — the registration itself is the expansion signal.
For paid subscription: Monthly subscribers viewing annual plan pages. Single-topic readers expanding into new content categories. Individual subscribers browsing team pricing. Monthly-to-annual conversion is the highest-leverage move: annual subscribers show 70% year-2 retention versus 34% for monthly and are 2.4x more profitable.
For membership businesses: Event attendance increasing, bootcamp enrollment after events, colleague introductions, sponsor interest. Hampton built 1,000+ members and ~$8M ARR entirely through word-of-mouth — every referral is both an expansion signal and an acquisition channel.
The compounding rule: When 3+ expansion signals appear within 60 days, conversion to higher-value status runs 40-60%. A single signal might be noise. Multiple signals are intent.
Benchmarks:
- 20-30% of active audience members trigger at least one expansion signal per quarter
- Expansion should account for >15% of base revenue annually (not just new acquisition)
- Newsletter free-to-paid: 2-3% median, 5%+ top quartile
- Membership upsell (event/bootcamp): 15-25% of active members per year
Where the data lives
Expansion signals are scattered across platforms. Email platform (beehiiv, ConvertKit, ActiveCampaign) tracks opens, clicks, and premium content engagement. Analytics (GA4, Plausible) tracks pricing page views and session depth. Payment platform (Stripe, Memberful) tracks plan changes. Community platform (Circle, Discourse) tracks event signups and referrals.
No single platform sees the full picture. Your email tool knows they clicked a premium teaser but not that they visited the pricing page three times this week. Your analytics knows the page views but not their email engagement history.
A warehouse (BigQuery + dbt) joins these signals into a subscriber-level expansion score: content engagement + pricing intent + referral behavior = predicted upgrade probability. That score reverse-ETLs back into your email platform as a custom property, triggering targeted upgrade sequences to the audience members most likely to convert. This is Phase 3-4 work in the implementation roadmap.
Metric 8: Churn Prediction (Engagement Decay)
What it measures: The measurable decline in engagement signals that precedes a cancellation or lapse event by 30-90 days. The earlier you detect the decay, the cheaper and more effective intervention becomes.
The core insight: Revenue events (cancellations, non-renewals, unsubscribes) are lagging indicators. By the time someone cancels, they checked out weeks ago. Engagement decay is the leading indicator — and it follows a predictable curve.
For newsletter publishers: A 4-week rolling decline in open rate is the first signal. Zero clicks in 30 days is the second. The progression: engaged → opens but doesn’t click → opens sporadically → stops opening → unsubscribes. A healthy unsubscribe rate is 0.1-0.2% per send. Above 0.5% is a red flag. The intervention window is between “opens but doesn’t click” and “stops opening” — typically 2-4 weeks.
For ad-supported media: Visit frequency decline is the primary signal. A reader who visited 3x/week dropping to 1x/week has already lost 60-70% of their ad revenue value. Session depth declining (fewer pages per visit) is the secondary signal. By the time a daily reader becomes a weekly reader, 80% of their ad revenue contribution is already gone. The intervention: re-engagement emails, push notifications, or personalized content recommendations during the frequency decline window.
For paid subscription: Usage decline precedes cancellation by 30-60 days. Articles read per month declining, logins declining, premium features unused. The 60-day pre-renewal window is critical — subscribers who haven’t used premium features in those 60 days cancel at 3-5x the rate of active users. Industry churn range: 3.9-6.9% monthly for digital subscriptions.
For membership businesses: This is where churn prediction matters most (highest ARPU). 47-52% of membership churn comes from “lack of engagement” — not price, not dissatisfaction, just fade. Key decay signals: no event attendance in 90 days, inactive in community for 60 days, no peer interaction 60 days pre-renewal. The median membership renewal rate is 84%. First-year renewal is lower (75%). High-ticket membership target: 85-95%.
The NRR equation:
NRR = (Starting ARR + Expansion - Churned - Contraction) / Starting ARR
Target: 95-105%. Below 95% means your base is eroding faster than expansion can compensate. Above 100% means your existing audience is growing in value without any new acquisition — the holy grail.
The compounding math: A 5-point retention improvement (85% → 90%) generates 15-25% more cumulative revenue over 5 years. For a membership business at $15K/yr dues, that’s the difference between $134K and $200K lifetime value per member — a $66K swing from a single retention metric.
Intervention economics:
The cost of re-engaging a decaying subscriber is a fraction of acquiring a new one:
- Automated re-engagement emails: Near-zero marginal cost. Triggered by engagement score decline. Personalized content recommendations based on historical affinity.
- Personalized content push: Surface their highest-affinity content topics via email or notification. Content recommendation engines automate this at scale.
- Renewal incentive: For at-risk pre-renewal window only. Annual discount, bonus content, exclusive event access. Not blanket discounting — that trains healthy subscribers to wait for deals.
- Personal outreach (membership): For high-value members showing decay signals. A personal email or call from community team. The ROI on a $15K member justifies 30 minutes of staff time.
The 90-day rule: For newsletters and ad-supported media, once a subscriber has been completely inactive for 90 days, the expected re-engagement rate drops below 5%. For memberships, the non-renewal decision is usually made 60-90 days before the renewal date. The intervention window is before that decision crystallizes.
Where the data lives
Engagement decay detection requires time-series data: engagement signals tracked over weeks and months, not point-in-time snapshots. Email platform provides open/click history. Analytics provides visit frequency and session depth. Community platform provides event attendance and interaction frequency. Payment platform provides subscription status and renewal dates.
The decay model is inherently a warehouse model. It requires calculating rolling averages (4-week open rate, monthly visit frequency, 90-day event attendance), comparing current behavior against historical baseline, and scoring each subscriber on a decay curve. This is a dbt model that runs daily, flagging subscribers whose engagement has dropped below threshold and triggering intervention flows.
This is Phase 2-3 work. The basic version (email engagement decline + simple intervention flow) can be built in weeks. The full multi-signal decay model with cross-platform scoring is Phase 3-4.
What These Two Metrics Give You
Expansion signals tell you where growth will come from inside your existing audience — before you spend another dollar on acquisition. Engagement decay tells you who’s fading and when to intervene.
Together, they shift the retention conversation from reactive to predictive. Instead of analyzing why people cancelled last month, you’re intervening during the 30-90 day window when it’s cheapest and most effective. For high-ARPU models like membership businesses, where a single retained member can be worth $66K more in lifetime value, predictive retention isn’t a nice-to-have — it’s the highest-ROI investment you can make.
Final part: Efficiency Metrics — how fast your audience flywheel spins and how sustainably you convert investment into revenue.
