One publisher's Discover-driven traffic and revenue, per Adsterra

Adsterra, an ad network, published a case study of an independent publisher it identifies as Karan, based in India. The write-up appeared on MarTech Series on May 5, 2026. It is a vendor’s account of a customer, and it states no dates, so this page reports what it says and nothing beyond it.

What does the case report?#

The article says about 95% of the site’s traffic came from Google Discover and about 85% from mobile. It reports 8.45 million impressions, 703,948 clicks, an 8.32% average CTR, an average CPM of $0.825 and $6,976.06 in revenue. The metrics above use a zero baseline because the source gives totals, not before and after values.

The impressions and clicks here are ad impressions and ad clicks served by Adsterra, not Discover impressions from Search Console. The article describes traffic arriving in unpredictable spikes, sometimes hundreds of thousands of visits within one or two days.

What did the publisher do each day?#

Adsterra describes a workflow built around trend-watching. The publisher scanned Google Trends, Google News and social platforms for emerging topics, then wrote four to seven articles a day with a structured format and several headline variations. Pieces leaned on strong emotional angles such as surprise, controversy or inspiration. Each was submitted through Search Console, performance was reviewed within hours, and topics that gained traction were scaled. The article says small headline updates were used sparingly to revive visibility.

What technical setup does the article describe?#

The site ran on WordPress with Rank Math SEO and Ad Inserter, on managed hosting with a CDN. The article lists a lightweight mobile-first theme, HTTPS, Open Graph tags, NewsArticle schema, featured images at least 1200px wide and no intrusive interstitials or pop-ups. Ads were Adsterra’s Social Bar and Native Banners, sold on a CPM basis across countries.

Why does geography matter to the revenue?#

The revenue figure hides a wide spread. India accounted for $2,114 from 6.2 million impressions at a $0.339 CPM. By our arithmetic from those figures, India was roughly 73% of impressions but only about 30% of revenue. The United States paid a $23.30 CPM and Canada $19.97; the article does not give their impression counts. A Discover audience concentrated in one country can be large and still earn little per view.

How does this line up with Google’s Discover guidance?#

Google’s own documentation says traffic from Discover is less predictable than keyword-driven search and should be treated as supplemental to it. That is the risk this case runs at 95% dependence, and the article itself lists traffic volatility and reliance on one distribution channel as the main drawbacks.

Several of the publisher’s choices match what Google asks for. Google recommends large images at least 1200px wide, enabled with the max-image-preview:large setting, and titles that capture the essence of the content. It also warns against clickbait and sensationalism that inflate engagement with misleading or exaggerated preview details. A workflow built on emotional angles and headline variations sits close to that line, so the headline testing is the part to copy with care. Our summary of what Discover eligibility actually requires covers the documented rules in full.

What should a publisher check before copying this?#

• Open the Discover performance report in Search Console to see how much of your traffic already depends on it.
• Confirm max-image-preview:large is set and that lead images meet the 1200px width.
• Keep headlines accurate to the article, since Google names misleading previews as a problem.
• Model revenue by country, because the same traffic volume paid about 70 times more per thousand impressions in the United States than in India here.

For wider context on how much news traffic Discover now carries, see Discover’s share of Google traffic to news sites.

How should the case be read?#

This is one publisher, self-reported through the vendor that profits from the story, with no timeframe and no traffic baseline. It shows that a site can lean almost entirely on Discover, and it shows the risk of that dependence, but it does not show the growth rate or that the workflow would transfer.