What a paywall costs you in advertising

Sample text, to show how a post reads. Replace with your own.

The objection to a paywall is almost never that readers will not pay. It is that metering will cost more in advertising than it earns in subscriptions. The objection is reasonable. It is also, in most cases, answerable with data the publisher already has.

The fear assumes that every impression lost to the meter is revenue lost. That assumption breaks in two places: not all impressions are worth the same, and not all readers stop reading.

Where the number actually comes from

Start with the distribution of page views per reader per month. In most news operations it is extremely skewed — a small share of readers generates the majority of views, and the long tail arrives once from search or social and never returns.

A meter set above the tail leaves that traffic untouched. The impressions genuinely at risk are those belonging to readers above the threshold, and those readers are precisely the ones most likely to subscribe rather than leave.

The advertising you lose to a meter is concentrated in your most loyal readers. So is the subscription revenue you gain. The two are not independent, and modelling them separately is what produces the scary number.

Making it measurable before launch

• Model lost impressions at several meter levels, not one — the curve matters more than any single point.
• Value those impressions at the yield they actually earn, not at your average CPM. Tail traffic is usually worth a fraction of loyal-reader inventory.
• Net the loss against conversion at the same threshold. The trade only makes sense evaluated jointly.
• Run the meter on a share of traffic first. A live test settles arguments that a spreadsheet cannot.

Done properly this is a week or two of analysis, and it converts a debate about fear into a decision about a number.