Evan Williams: Medium's key advertising sales metric is not page views, but time spent on page
Evan Williams: On billionaire-backed journalism, Shanley Kane, and the problem with vanity ad metrics — Many people use the newish blogging site Medium each day, but almost as many of them are confused … Tweets: @ftrain , @megan , @jonathanglick , @joshelman , @pkafka and @digiphile Tweets: Paul Ford / @ftrain : “Brand advertising has never worked on the Internet anyways, because banner ads don't work.” http://fortune.com/... Megan McCarthy / @megan : Medium sold advertising to BMW using time spent on the article, not pageviews, as its sales metric: http://for.tn/1srfJJW Jonathan Glick / @jonathanglick : Compare @ev's take on the purpose of paid content on a platform (http://fortune.com/...) with @nicknotned's (http://nick.kinja.com/...) Josh Elman / @joshelman : Great interview with @ev on @medium and describing the big opportunity he and a great team are going after http://fortune.com/... Peter Kafka / @pkafka : Sharp Q&A on @medium with @ev, via @eringriffith. http://fortune.com/... We'll do one live in Feb at #codemedia http://recode.net/events/ Alex Howard / @digiphile : “...the main thing we measure is time spent on stories. I think it's the best proxy for value."- @ev http://fortune.com/...
Context & Ripple Effects
In December 2014 Medium had just enough advertiser pull to challenge the industry's default currency: Evan Williams told Fortune the site sold BMW a campaign priced on how long readers stayed with an article, not pageviews — a direct rebuttal to Paul Ford's tweeted line that brand advertising has never worked online because banners don't work. That argument became a through-line in Medium's decade-long run of strategy shifts.
The stakes are visible in what came next in related coverage: Williams reportedly pitched Medium to Twitter for $500M during Twitter's own growth troubles, then handed the CEO role to Tony Stubblebine in 2022, whose plan for Medium to turn a profit in 2024 leans on distribution and membership rather than the ad business this interview was defending. The metric debate was never academic — it sat at the center of how the company could sustain itself.
First-order effects
- BMW's campaign becomes the proof case: Medium's sales team must demonstrate dwell-time value to brands accustomed to buying audited impressions.
- Advertisers lose the ability to verify delivery from their own click logs, so trust shifts entirely to whatever measurement Medium reports back.
Second-order effects
- Rival platforms face a fork — adopt attention metrics or argue that pageview-priced display inventory undervalues engaged audiences — even as budgets drift toward targetable, customer-acquisition-style spending, the shift Benedict Evans traces as Amazon's ad revenue passes YouTube's.
- Ford's objection sharpens the competitive frame: if banners never worked, every publisher claiming a better metric is implicitly conceding the old brand-ad format failed.
Third-order effects
- If attention-priced brand deals stay hard to scale, platform publishers default to subscriptions — the path both Medium under Stubblebine and the NYT's subscriber-first Beta Group push followed in the years after this interview.
- The unresolved question Ford raised — whether brand advertising has a genuinely native format online — keeps resurfacing each time a new metric or channel claims to fix it, from dwell time to today's algorithmic targeting.
The trend: Digital publishing has been migrating away from impression-count currencies toward attention and outcome metrics — and, where those stall, toward subscriptions.