Somewhere in New York, Washington, and San Francisco, the chief revenue officer at a digital platform is sending a note to his political sales lead. It goes something like this, “The Borrell Report says $1.41538 billion was spent on campaign advertising this cycle. Why didn’t we get even 1 percent of that?”

The answer, as several political digital ad sales teams learned the hard way last year, is that trying to grab a made-up percentage of a questionable pie is a surefire recipe for leaving you hungry. That’s because tracking political digital dollars -- and by extension, projecting them -- is essentially impossible. Here’s why:

FEC expenditure reporting is almost worthless

FEC expenditure reporting is absurdly loose about naming and categorizing vendors. With the exception of California, the lack of sub-vendor reporting allows the general consultant to hire everyone else, meaning it’s difficult to tell where the funds are actually going. (This is why you rarely see oppo researcher on FEC reports.)

For example, OpenSecrets data doesn’t accurately tally the digital advertising Audience Partners, and its campaign division, CampaignGrid, sold this cycle. Nor does it account for everything our competitors sold this cycle. The same database shows just over $7.6 million for Facebook and just over $1.6 million for Google. I’ll speak for them both and say that’s not even close.

Most state laws regarding campaign expenditures are even murkier, especially for sub-vendors. You should question any analyst using the FEC or OpenSecrets as the primary guide for campaign expenditures because, under current law, the information there is never going to be representative of what’s really spent.

So how else do you determine digital political spending?

Data from public companies like Facebook

The Borrell Report suggested about $560 million would be spent on “social,” but doesn’t break it out to specific networks. Although Facebook is a public company, their official political numbers won’t be out for a while. I suspect their political revenue this cycle may be under $350 million, and Twitter and the other social networks won’t come close to making up the difference.

Google and others will also provide a broad picture of what they generated from the campaign cycle, but the data is often so generic that it’s impossible to use for an industry model. Meanwhile, Rocket Fuel, another public company, cited a “bizarre” election season as the cause for weak Q3 revenue numbers, but didn’t provide much more. But in a moment of candor, a source told AdExchanger: “There is a 0% chance that many resources are dedicated to political again.”

Public companies cover more than half of political digital spending. If we have junk data from the FEC and OpenSecrets, and overly broad data from public companies, where else can we learn about total digital spends? 

Data from privately owned demand-side platforms

DSPs will have a window to what is being bought, but they have little interest in tracking political-versus-commercial spending. They have more important things to worry about, like preventing the next Methbot and improving reach. Even if all buying platforms had clear spending data categorized by candidate (they don’t), why would they release this information if the FEC doesn’t require it (they won’t)?

Unlike broadcast buying, there are too many on-ramps to the Internet to ever regulate effectively. In other words, you all can stop freaking out about regulating political speech on the Internet. Even if the FEC decided to regulate paid digital messaging more closely (they won’t), I’m doubtful the technology exists to implement that decision.

Moat and other tracking services

Go to Moat.com and type in Hillary for America. You’ll see 617 creatives. You’ll find fifteen different creatives for Pat Toomey. It’s helpful, but you won’t see spend data, CPM’s, or creative from most down-ballot campaigns. We’ll see an expanding market for new tracking services, but nothing will be bulletproof.

There’s also no agreement for categorizing campaign expenditures. Imagine a sales representative at a local television station, who fills his days selling auto ads. He lands some digital ads for a congressional candidate. From there, the congressional candidate lists the television station as the vendor, and the sales rep doesn’t bother clarifying in his reporting that this was political rather than auto. The data is lost.

Most digital spending in politics and public affairs is not reportable. So while sub-vendors are under-represented in virtually all FEC and state-level reporting, large digital buys for advocacy purposes aren’t included in industry estimates.

Make an estimate based on stuff you know

Tracking broadcast spending is more precise because it’s required by the FCC, and the technology to do so is far less complex. If we know total expenditures according to the FEC, and we know total broadcast spends, we can subtract the difference and understand what’s left over for other items. From there, you can allocate the market share according to prevailing thought on the demise of all things that aren’t addressable.

Most digital staff have exceedingly intimidating non-disclosure agreements, so interviews are not typically fruitful. We can also look at hiring patterns of digital shops. If an agency closes its D.C. branch or evaporated altogether, the year was probably not so good.

You’re doing it wrong

Add all of that up, and you’ll see very quickly just how futile it is to report, or build your business around an exact industry digital spending number. Plus, it’s getting harder as less political spending is subject to FCC disclosure, and we all know the FEC won’t be getting nitpicky anytime soon.

So if you’re crafting your revenue plan based on grabbing some percentage of the Borrell Associates billion-and-a-half dollar digital spend, you're doing it wrong.

Jordan Lieberman is politics and public Affairs lead for Audience Partners (@AudiencePartner) and the self serve platform CampaignGrid Direct.