A CPM’s-Eye-View of the Web

In e-commerce, it’s common to think of an entire site as a “conversion funnel”: basically, the site is designed with the last user interaction first (i.e. the shopping cart process); the last thing designed is the first part users see—the landing pages. This is a good framework, because it produces purposeful sites: every single page either closes a sale or brings the reader closer to closing a sale on the next page.

Better yet, it gives site owners a framework for analyzing every decision they might make on the site: every single page has a fairly predictable payoff function, so the site’s to-do list is basically an ordered list of expected profits divided by costs.

But there’s no reason to limit this view to the site itself. Where do users come from? Ads, referrals, searches—from the “funnel” perspective, these are just landing pages with low conversion rates. This kind of thinking helps clarify the economics of different ad campaigns: by treating every page you can influence as part of a broad conversion funnel, you can turn any decision into a fairly trivial question of comparing two conversion rates side by side.

The sites that run these ads aren’t used to thinking of themselves as landing pages in a funnel. They’re used to thinking of pages in terms of revenue per pageview (usually quoted in RPM, revenue per thousand, to avoid fiddling around with decimals). But it’s possible to treat ads’ destination pages as equivalent to media pages, just with a higher RPM. That kind of thinking can extend all the way to the checkout page. What is a checkout page if not a perfectly targeted ad with a double-digit click-through rate?

This framework doesn’t capture profits that are derived from competitive advantages—which are opaque to outsiders, and whose magnitude doesn’t really affect decisions. Aside from those profits, the CPM-based view and the conversion-funnel view amount to exactly the same thing. This view of the web is accurate to the extent that all business models are static—but another way of saying that is to argue that the parts of the web that don’t make sense based on this framework are where the action is.