global sports blog × Adverge · head-to-head split test

+118%

revenue per session over the other partner. The test was meant to run 30 days.

A global sports blog set up a 30-day split test between Adverge and its other monetization partner. Same site, same days, both measured against the same pageviews and sessions.

Adverge led on revenue per session and per thousand pageviews on every one of the 13 days the test ran. The narrowest daily lead was +78%.

After two weeks the publisher stopped the test and moved to Adverge. Seventeen more days of a split meant seventeen more days of leaving revenue on the table.

RPS

+117.87%

revenue per thousand sessions, over the test

Page RPM

+117.87%

revenue per thousand pageviews, over the test

Days in the lead

13 of 13

narrowest lead +78%, widest +204%

Test length

2 weeks

of a planned 30 days, then a full switch

A 30-day test. Called in two weeks.

Split tests are expensive for a publisher. Whatever share of traffic goes to the weaker setup earns less for as long as the test runs. Once the gap was clear and holding, there was no reason to keep paying for the answer.

13 days run17 planned days the publisher chose not to wait for

Two lenses. One answer, every day.

Both partners are shown against the same baseline: the other partner’s average across the test, set to 100. Scroll through RPS, then page RPM, then the daily lead. The shaded bars behind the RPS and RPM lines show impressions, so you can see the seasonal dip and the lead in the same view. Each chart docks as the next one opens.

01 · RPS

Every visit was worth more than twice as much on Adverge.

Revenue per session is the number a publisher can take to the bank, because it is what each reader arriving on the site is actually worth. Adverge opened the test +161% ahead on day one and never dropped below +78%.

The shaded bars behind the lines are impressions. They fell with a seasonal dip in traffic, 32% lower in the second week than the first, and both lines drift down with them. The gap did not shrink: Adverge led by +117% in week one and +120% in week two.

02 · Page RPM

The same gap shows up per thousand pageviews.

Both partners were measured against the same daily pageviews and the same daily sessions. That means the RPM lead and the RPS lead are the same number each day: +118% across the test.

Two different lenses on the same inventory, and both say the same thing, in the strong week and in the seasonal dip alike. The extra revenue was not coming from deeper visits or a lucky day. It was coming from how each page was monetized.

03 · The call

13 days, 13 wins. The publisher stopped waiting.

The lead held on every day of the test, from a narrowest +78% to a widest +204%. In the second week, with impressions down, it ran slightly wider than in the first.

With part of the site still on the other setup, every extra day of testing had a price. The publisher did not want to pay it, called the test after two weeks, and switched fully to Adverge.

RPS+118%

Revenue per thousand sessions

daily, Adverge vs. the other partner

AdvergeOther partnerImpressionslines: index, other partner’s average = 100

050100150200250300350WEEK 1 LEAD +117%WEEK 2 LEAD +120%IMPRESSIONS -32%ADVOTHERDAY OF TEST1471013
Page RPM+118%

Revenue per thousand pageviews

daily, Adverge vs. the other partner

AdvergeOther partnerImpressionslines: index, other partner’s average = 100

050100150200250300350WEEK 1 LEAD +117%WEEK 2 LEAD +120%IMPRESSIONS -32%ADVOTHERDAY OF TEST1471013
Daily lead13/13

Adverge’s RPS and RPM lead, day by day

13 days run of a planned 30

0%+50%+100%+150%+200%+250%PLANNED DAYS 14 TO 30NOT NEEDEDPUBLISHER SWITCHES+204%+78%AVG +118%DAY OF THE PLANNED 3017132030

What Adverge did differently on the page.

No redesign and no new traffic. Three changes to how the same audience was monetized.

01

Smart refresh

Ads refreshed only when it made sense: the slot was in view and the reader was still active on the page. No blind timer cycling ads on tabs nobody was looking at. More sellable impressions from each visit, without wasting buyers’ money on ones that would never be seen.

02

Geo throttling

A global audience does not monetize evenly. Ad requests and refresh were calibrated by country, so high-value markets got full monetization and low-demand regions did not flood auctions with impressions that would clear for very little.

03

Demand throttling

Each impression went to the buyers most likely to bid on it, not to every partner on every request. Fewer wasted bid requests mean faster auctions, ads that render in time to be seen, and demand partners that keep bidding because the traffic they get is worth it.

Same site. Same days. More than twice the revenue from every visit.

This is not a story about a bad partner. It is about revenue each session was worth, which one setup captured and the other did not. Over two weeks, that difference was large enough and steady enough to make the decision for the publisher.

Why the publisher did not wait for day 30.

A split test exists to remove doubt. Here the doubt was gone in the first week: Adverge led on day one and never gave the lead back. The remaining days could only confirm what the publisher already knew, at the cost of part of the site earning less than it could. On average, every session on the other setup earned less than half what the same session earned with Adverge. The faster decision was the cheaper one.

Figures are the publisher’s own daily reporting for 13 complete days of side-by-side data. Both partners’ revenue is measured against the same daily pageviews and sessions, so the RPM lead and the RPS lead are identical on each day. Headline figures compare each partner’s total revenue across the test: RPS +117.87% and page RPM +117.87%. Daily leads ranged from +78.40% to +204.46%. Impression bars combine both partners’ daily impressions and show relative height only. Week figures compare days 1 to 7 with days 8 to 13: impressions fell 32% and the lead moved from +116.8% to +120.3%. Charts index each partner’s daily RPS and RPM to the other partner’s 13-day average, set to 100. Absolute RPS, RPM, revenue and traffic values are withheld, so every figure shown is a relative one. Neither the publisher nor the other partner is named here.

Run the test. Keep the winner.

If you are weighing a monetization partner, put us side by side with what you have now. We will set up the split, report both sides on the same numbers, and let the data make the call.

Get your free audit →

adverge.ai

SportsSplit testRPS