Dead to Rights

It's not just about making millions in questionable arbitrage. It's about corporate outcomes, too.

“I’m sending over the spreadsheet,” said the media supervisor.  “There’s something I need you to see.”

When she said it, it was with an overarching sense of dread, like a mutual friend had been in a serious car wreck.

She attached it to an email, sent it over, and then joined me in my office to look over my shoulder as I inspected the spreadsheet.  She wanted to know if this Excel formula was doing what she thought it was doing.  Because if it was, the implications were staggering.

Slow and Steady

Our approach to programmatic was measured and careful.  Most of our clients knew media rather well, and they had tight controls over where their ads ran.  For our largest and highest-profile client, digital was the only medium they didn’t buy in house, so they knew a lot about media value and how smart media buying contributes to that value.

When game-changing technology came along, we were usually among the first to pilot it with clients.  When it came to buying programmatically on the open web, we had taken an uncharacteristically slower pace because of our concerns about brand safety and fraud.

In fact, when we had selected our first partner, whose platform we had envisioned handling the bulk of our agency’s programmatic buying, we had strict requirements about transparency.  We wanted pricing transparency from the DSP all the way back to the ad exchange, and so we would have our DSP partners dump transaction logs into flat files and batch them over to us.

We also used a whitelist approach.  And we developed a process where we would look at venues where we could source inventory, and teams of people at the agency would vet the sites and knock out any that weren’t appropriate venues for our client brands.

No, you don’t “get it.”

These requirements sent many of the DSPs seeking our business running away screaming.  Many of them would cast doubt on our ability to “get it” when it came to scale.  They told us we were greatly limiting our opportunity in programmatic and that we would come to regret putting these controls in place.

We knew that we were limiting scale.  That was the point.

In fact, if you were in sales at a DSP and had called on me at that time, you would have heard me describe our own requirements as “draconian.”

I even used that word with the consultant we hired at the time to administer our first reviews of potential partners.  If you work in the programmatic end of the business, you’ve probably heard his name – Matt Prohaska.

The Bozo Filter

Our main DSP partner and a small handful of other DSPs agreed to the requirement, though.  Over time, we discovered reasons to add multiple DSPs to our approach.  Each had to agree to the draconian requirements.  Most complained vigorously about limited scale.

But back to that spreadsheet the media supervisor wanted me to take a look at.

We were in the process of building technology to process the logs from programmatic buys and identify inconsistencies and other red flags.  But at the time, we were in full manual review mode, where our buyers would make regular spot-checks of the logs.  One of these checks resulted in flagging that spreadsheet.

A Gut Punch

First off, the spreadsheets weren’t supposed to have complex formulas in them at all, beyond summing up impression totals and such.

This one did.

That formula referenced a hidden tab on the spreadsheet.  That tab contained the whitelist we had supplied the partner with.

It used VLOOKUP and a couple other functions, and it took a second for me to realize what it was doing.

It was copied through every entry in the log, and if the domain for that impression didn’t match an entry on our whitelist, it simply changed it so that it did.

Clearly, this had been sent off to us before it had been cleansed of the formulas.  We weren’t supposed to see this.

I Suddenly Don’t Feel So Hot…

I was overcome with a feeling of dread, not unlike that feeling when you pat yourself down and realize your wallet isn’t on your person.

There had been a few occasions over the years when we had caught media partners dead to rights trying to defraud our clients.  When it happened with a pharmaceutical company, there were always hours on top of hours spent on the phone with in-house and external counsel, where we had to explain ad tech concepts to people who typically didn’t deal with them day to day.  I was already dreading those calls.

But not as much as I dreaded explaining it to my client, because that explanation would come hand-in-hand with the sense that I had let them down by recommending they do this at all.

Yes, it was a shitshow.  I won’t get into it.

Why All The Obfuscation and Fraud?

It’s been over a decade.  I’ve had plenty of time to think about what this all meant.

This wasn’t about merely selling some advertising fraudulently, although more than a decade later, we still haven’t cleaned up this swamp and opportunities to defraud advertisers in the programmatic space are still everywhere.

This was about putting up numbers for the investment community and engaging in fraud to produce a “liquidity event” for a small handful of people in a hot sector, where showing revenue against “proprietary tech” gets you a high “multiple.”

So when you look at ad fraud at a high level and wonder why, it’s not just about somebody in some deep dark corner of the mobile app space making insane margins on arbitrage or creating MFA opportunities at scale.  It’s often about the corporate sector looking for a big payout – and all the incentives to cheat are still there.

Seven Deadly Sins will next publish in January, after we’ve taken some time to decompress with our families. Happy Holidays!