How we discovered undisclosed buy-side fees

How we discovered undisclosed buy-side fees

09 Jul 2018

Before she joined CtrlShift, Kar Wai Low spent many years on the publisher’s side of the advertising ecosystem. In this post, she shares how a desire to discover just how money is distributed along the programmatic supply chain turned the unspoken assumption of the disappearing ad dollar into a very real fact.

It has always been a frustration for publishers to not be able to see through ad tech’s complexity and lack of transparency.

What exactly were buyers paying or how much were they willing to pay for our inventory? What were the fees in between? Does it all add up? We weren’t sure.

We first investigated whether there was actual leakage in the supply chain. There had been speculation that we were only getting a pittance back from our sold inventory in the open marketplace, and we wanted to see if that was true.

Sorry, I should have clarified that ‘we’ were a premium publisher in the UK, who also had our own little trading desk - making us both a seller and buyer in the digital ad trade.

So what did this allow us to do? It allowed us to buy our own inventory. We created campaigns on our DSPs (demand-side platforms), targeted only our domain in the respective SSPs (supply-side platforms) and bidded accordingly. Lo and behold, we found out that up to 70 percent of ad dollars disappeared by the time they reached us. We were only receiving around 30p on the pound.

The disappearing ad dollar was no secret, we just didn’t know where it went. Data costs and tech fees aside, the large margins lost didn’t quite make sense.

We were also optimisers. Our forte was the ability to increase efficiency and minimise wastage where possible, and we were going to make sure we got every cent out of the upholstery. Highly curious creatures, we just had to find out where that money went.

To be able to see exactly where the margins were lost, we needed to dig deeper into our data. At log-level, we were able to see not just the winning bids but the bids submitted. There were differences based on whether it was a first or second price auction, or if soft floors were at play (see glossary below). However through the log data, one could work out the max bid placed, the bid won, the win rates, and the bid reductions and the price actually paid.

The bid reduction is effectively the consumer surplus which is the amount ‘saved’ between the winning bid and price actually paid. Just hold on to that, I’ll get back to it. Anyway, at log-level, we had an ability to see the entire bid landscape as it lives.

So what were we looking for? Rumour had it that it was the middlemen who were slipping extra change into their pockets, so we set out to prove this.

The methodology was as such:

We stripped down the supply chain as much as possible leaving only four main players - the buyer, the DSP, the SSP and the seller. Both the DSP and the SSP will take a fee during that journey.

We put together various test scenarios, with variables not just being the four main players but including possible influencing factors - bidding mechanisms, trading times, sample sizes, etc.

In theory, the actual bid price paid by the buyer should equate the seller’s revenue, factoring in all fees. In reality, there was a massive discrepancy. The degree of how much the margins varied was not only more than what it should be, but it was also volatile. The only correlation we found was that the degree varied with the difference between the winning bid and the price actually paid, that is, the bid reduction.

We were getting somewhere.

These platforms operate on a second-price auction model whereby the winning bidder pays only 0.01 higher than the second highest bid, despite bidding more. Putting it simply, if Bidder A bids $10, Bidder B bids $5 and Bidder C bids $7, Bidder A wins and should only pay $7.01. Seller gets $7.01. Tech fees aside.

What we found was that Seller was indeed getting $7.01, but Bidder A was paying more. How much more depended on the level of bid reduction - the larger the bid reduction, the bigger the difference between what the buyer paid and what the seller got. But the point was Bidder A was paying more than they should, and someone - a middleman - was taking more than they should.

It did not take too long after to filter out who the naughty kids were. It did not take too much for them to come clean. This ultimately culminated in us suing a particular vendor for allegedly not disclosing its fees, and a counterclaim was received in return - it remains to this day an ongoing quabble.

Whatever the outcome, the message has been conveyed and we got a conversation going. Transparency issues exist in ad tech, and both advertisers and publishers should hold more control in the digital media supply chain. We need to redefine digital advertising, to support its evolution from a fractured and inefficient marketplace to a competitive ecosystem that delivers true value to buyers and sellers.

Ultimately we are simply just users who want more transparency. We want to know where our money goes as soon as we scream show me the money. The sale of media should remain the seller’s right, buyers should get a fair trade, and the technology in between should merely be an enabler.

Kar Wai Low is the Regional Lead, Trading and Operations at CtrlShift.


Demand-Side Platform (DSP)
Platform that allows buyers to automate the purchase of display, video, mobile, and search ads. There is no negotiation of rates so buyers can quickly and efficiently buy targeted impressions across a wide variety of publishers and platforms.

Suppy-Side Platform (SSP)
Software platform that allows publishers to connect their inventory with ad exchanges, networks, and DSPs at the same time to sell impressions at their highest value in real time. An SSP provides publishers with access to many potential buyers, and they can set a minimum price they will sell impressions for.

Price Floor
The lowest price a seller will accept for impressions.

Hard Price Floor
A hard price floor is the lowest price for the impressions that publishers agree to accept from ad exchanges. Bids below this price are simply discarded. 

Soft Price Floor
A soft price floor “catches” the offers that are only slightly below the hard floor and would otherwise get rejected with no yield for the publisher. 

Second Price Auction
The winner of an ad impression in an auction pays one cent above the next highest bidder.


Liked what you just read? Contact us to find out more about CtrlShift and how we help advertisers and agencies achieve their desired business outcomes.