Understand how predictive paywall works
Predictive Paywall combines content and visitor signals into a score. The score helps choose when to show a wall; it is not a reader's percentage chance of subscribing.
Understand the signal groups
| Signal group | What contributes |
|---|---|
| Content activity | Recorded paywall outcomes, associated signups and subscription purchases, compared with publication activity. |
| Related content | Activity summaries for content sharing Categories, Tags or Authors. Classifications also help select the intended access rule. |
| Visitor history | Available account or guest-session activity, transaction history where available, and earlier paywall outcomes. |
The formula variables set the relative influence of these signals. Categories, Tags and Authors connect content items so related activity can contribute to the score.
The inputs come from recorded activity and stored summaries, so updates need not appear immediately. The content metric definitions explain how the activity is counted.
Read the Paywall percentage as a target
For City Journal, 20% aims to show the wall on roughly 20% of eligible content requests. The score selects which requests are candidates, and the policy also considers recent recorded wall decisions.
The target applies to requests considered by the predictive rule. It counts requests, not distinct people.
Recorded activity contributes to the score. The applicable rule and policy determine access, and the website presents the result.
Understand previously unlocked articles
An article already unlocked for a reader normally stays unlocked. The publication option Do not grant content access based on “Already Viewed” access rule. changes this behavior. When enabled, an earlier unlock does not by itself grant access, and the access checks continue.
For guests, this retained-access rule uses the session's earlier article access. For signed-in readers, it uses the account's earlier article access. This rule covers the previously unlocked article.
Review an unexpected decision
If a decision is unexpected, note the article, Plan/rule, reader or guest-session context and observation time. Use Configure and review predictive paywall to check the settings and give the integration or support owner a precise question.