Rubicon Project eliminated its buy-side fees Wednesday.
“This ties directly to the commitment I made when I joined the company earlier this year to normalize our pricing and bring a higher level of transparency to everything we do,” CEO Michael Barrett said in a blog post released just prior to Rubicon’s Q3 earnings Thursday.
Removing buy-side fees will lower Rubicon’s take rates even more. In 2016, take rates hovered in the 25% range, and the company slowly lowered them as it lost market share to lower-fee competitors using header bidding.
When Barrett joined the company in March, he estimated Rubicon’s take rates would end up between 15% to 20%, but now says those figures are still too high to be competitive.
“In order to be successful, exchange take rates will need to be between 10% and 15%,” Barrett said.
Rubicon’s independent competitors – like AppNexus and Index Exchange – have both been vocal about using low fees to gain market share. In September, AppNexus CEO Brian O’Kelley told AdExchanger, “It’s probably time for a price war.” Another competitor, OpenX, confirmed that it never charged buy-side fees for its inventory.
In September, Rubicon Project experimented with lower fees by conducting tests of modified first- and second-price auctions. Modified first-price auctions with no buyer fees produced the best outcomes.
“We saw increased revenue for publishers and increases in the win rate for buyers,” Barrett said.
Going forward, Rubicon will only charge two types of fees: marketplace fees that reflect a contract signed with a publisher, and platform access fees – technically, a kind of buyer fee – for buyers who don’t spend enough to be profitable for Rubicon.
Rubicon projected a take rate next quarter to be 4% to 5% lower than at the end of Q3, so take rates would decline from 16.9% to 12.9% or 11.9%. In Q1 of next year, take rates would decline another 1-1.5%, ending up roughly in the 10% to 12% range.
Rubicon’s near-term take rates would end up on the low end of the range Barrett outlined.
The decline indicates that buy-side fees roughly equaled sell-side fees back in 2016, before Rubicon started lowering its take rate, and averaged in the 10% to 15% range. Buyers in the open marketplace would have paid even more than 10% to 15%, though: Rubicon said it never charged buy-side fees for its private marketplaces, so those buy-side fees must have been concentrated in open marketplace programmatic deals.
Eliminating the fees signals the end of an era – and the end of a fee that led to a lawsuit from a publisher who felt deceived. Barrett acknowledged that the company would use the dynamic nature of buy-side fees to turn them up or down as needed, a quality that constantly irked buyers.
“We have stripped the fee down to just publisher-side fees, and that business is not nearly as dynamic as the buy-side business,” Barrett said. “Boy, did we live a dynamic roller coaster with that this year.”
Unlike buy-side fees, Rubicon negotiates marketplace fees with publishers and inks long-term contracts, making them more transparent and fixed.
Rubicon said its market intelligence indicates it charges lower-than-average fees in its contracts. That also may allow it to renegotiate contracts at higher rates, Chief Financial Officer David Day told investors.
The elimination of buy-side fees comes at significant financial cost. The company will be using its cash reserves of $100 million to fund operations as it moves to lower fees.
Because Rubicon is lowering take rates, it’s taking a double hit in revenue. Media spend is still declining, and Rubicon is taking a smaller cut of what’s still flowing in.
For example, Q3 media spend flowing through Rubicon declined 20% year over year, from $242.8 million to $195 million. But because Rubicon’s fees averaged 18.1%, instead of 24.9%, the decline in revenue more than doubled. Rubicon posted $35 million in revenue, a decrease of 47%.
“We are sacrificing near-term growth and profitability,” Barrett said, a “bold move” necessary for the company to move past the disruption wrought by header bidding.