Why Did Trump Pay Less than Clinton For Ads on Facebook?

An article in Wired has sparked controversy with its claim that Trump paid lower prices for its Facebook ads than Clinton:

During the run-up to the election, the Trump and Clinton campaigns bid ruthlessly for the same online real estate in front of the same swing-state voters. But because Trump used provocative content to stoke social media buzz, and he was better able to drive likes, comments, and shares than Clinton, his bids received a boost from Facebook’s click model, effectively winning him more media for less money. In essence, Clinton was paying Manhattan prices for the square footage on your smartphone’s screen, while Trump was paying Detroit prices.

The claim is plausible but although written by a Facebook expert it never really explains why Google and Facebook prices their ads in this way. The reason is what I call the “mesothelioma lawyer” problem. A click on an ad for a “mesothelioma lawyer” is extremely valuable because people who aren’t interested in hiring a mesothelioma lawyer are unlikely to click and those who do click are likely to become profitable clients. Thus, anyone searching for mesothelioma is likely to see an ad for a mesothelioma lawyer.

But suppose that Google or Facebook simply charge for ads by the click. Someone who searches for “funny hat video” isn’t likely to click on an ad for a mesothelioma lawyer but the people who do click are still likely to be very profitable to a mesothelioma lawyer. As a result, the mesothelioma lawyer can outbid the seller of funny hats for ads connected to “funny hat video” even though the search has nothing to do with mesothelioma.  If Google or Facebook only charged by the click it would be mesothelioma lawyer ads everywhere, all the time.

To avoid this problem, Google and Facebook calculate how many clicks or interactions your ad is likely to receive and they charge lower prices the greater the predicted number of clicks. As a result, sellers of funny hats get lower prices than mesothelioma lawyers for ads that pop up after the user watches a funny hat video and mesothelioma lawyers get lower prices than sellers of funny hats for ads that pop up after the user searches for information on mesothelioma. In the long run this system better targets ads to customers and thus maximizes the value of the platform to both advertisers and customers.

As the Wired piece eventually states this isn’t even new:

“I always wonder why people in politics act like this stuff is so mystical,” Brad Parscale, the leader of the Trump data effort, told reporters in late 2016. “It’s the same shit we use in commercial, just has fancier names.”

He’s absolutely right. None of this is even novel: It’s merely best practice for any smart Facebook advertiser.

Addendum: See also Hal Varian’s discussion of the underlying issues in the Online Advertising section of this paper.


Well, the real story here is that this story even exists.

It exists because Dems are still desperate to click on anything that proves that their disastrous candidate didn't REALLY lose the election.

Right. It's pathetic. The suggestion that Trump had some sort of unfair advantage is beyond ludicrous.

"Unfair advantage..." you mean like only having to win less votes than the opposition to win?

I know its legal but its not "fair." If every team playing the Yankees had to score one more run because New York was able to holdout at the formation of the MLB in exchange for that deal, it would be "legal" but not "fair."

You mean in the same way that Massachusetts doesn't have any Republicans in Congress even though the governor is a Republican.

Just so we know what you're talking about, would you define "fair" for us?

One person, one vote. Unless that approach doesn't work. Then it's judicial activism. Unless that approach doesn't work. Then it's pen and phone. Unless that approach doesn't work. No really, there is some principle of fairness here somewhere, I swear!

How did he win fewer votes? Basic Government classes in high school teach us that it's the states who vote for the president. Trump clearly won that vote. You can't just aggregate the votes across a set that is meaningless for the purpose of the presidential elections and ignore what those votes represent.

Yeah, but don't you think it's time to diminish the power of states? It really stops us urban folk from dictating rural life styles and that's a drag.

Poor analogy, because Hillary was the Yankees, outspending the opposition by a wide margin.

It would actually be more akin to the Yankees saying they won because even though they scored less runs, they had more hits. Sure, hits and runs are correlated but the correlation is not 1.00. Sometimes, you will do things in a baseball game to minimize hits in an effort to maximize runs. If both teams wanted to count the winner as he who scores the most hit, then both teams would adjust.

Put another way, if you change how the game is scored, then you change how the game is played. In the same way, if you change how elections are scored, you change how campaigns are run. Thus, the fact that HIllary won the most votes is not necessarily telling who would've won the most votes if that was the unit for scoring. Moreover, it is far from clear that the primary process would even be the same if you changed how the election process is scored. Thus, discussing total votes won is as useless as claiming that the Yankees won because they had the most hits.

Clinton's "win" in the popular vote was, in large part, testament to the incompetence- and insecurity- of her campaign.

I spent virtually the entire election season of 2016 within the state of California. There was absolutely no scenario where my state was going to have a close election. Despite this widely understood reality, Clinton pored millions into television ads in California. I saw hundreds and I am alight TV watcher at best. Some of this ineptitude owes to Clinton needing to let her donor base get a good luck at what they were giving her those five figure checks for (which is another interesting issue) but the larger portion of the blame for this foolishness was Clinton looking to pile up completely useless votes. In so doing, she squandered funds that could have been used to secure an actual election victory.

An incompetent dolt until the bitter end. Which in this case is never, because the ability of any well-funded candidate to lose to Trump will make that candidate- and her enthusiastic supporters- laughable buffoons until they are being mocked in tongues not yet known.

God bless Hillary. The perfect candidate for the modern progressive movement.

Actually, I think he's saying that if the Red Sox win more games (states/electoral votes) in a season, but the Yankees have a larger total run differential (popular vote), then the Yankees should still win the division because every run (vote) should count equally. Running up the score in blowout games (winning red/blue states by large margins) should somehow net against decisive winning runs in close games (winning swing states).

He didn't win less votes, but more. Otherwise he would not be president. I am talking about electoral votes.

Counting direct voters in this election is absurd: how can you add voters in different states where the rules to register and to vote (including the duration of the vote, the possibility to vote by mail, etc) depends on heavy of each state, and moreover the incentive of people to vote also depend heavily of the state (for instance is weaker in deep blue and deep red state, an stronger in swing state, or depends on the other elections on the ballot which change by state, etc.)? The very idea of adding votes in this condition show a profound disdain for one of the tenet of democracy, the equality of all votes.

Interestingly, you don't need to count direct votes -- if state electoral votes went to candidates on a proportional basis in all states instead of on a winner-takes-all basis in most, then Clinton would have received more electoral votes (though third party candidates would have gotten electors in NM and UT neither would have gotten a majority). It's both true that Trump won the election and that a lot of Americans (maybe even the majority of Americans eligible to vote!) didn't want him to be president.

It's very reasonable to talk about how the output of a system would be different if the system worked differently!

You mean like in the way that Wild Card teams that didn't even win their division sometimes win the Super Bowl or the World Series?

The Framers changed the rules of the game to eliminate the holdout, Rhode Island, from sinking the whole thing. It only took 9 out of 13 states to ratify. It's still called the United States of America. If you want a unitary rather than a federal system, then change it. Just be prepared to accept your favorite pet state policies to #resist the next president you hate being mowed over.

Actually it *is* fair. Your implicit assertion is that each person's vote ought to matter equally. But you are Simply Wrong about this. The Constitution does not prioritize the preferences of individuals (as you implicitly assert), it prioritizes the *weighted preferences* of groups of people (grouped into "States"). Each state *determines* the preferences of the people through popular vote, then each state *decides* how to present its collective preference to the community of states.

If every renounced the "winner-take-all" strategy they almost all use, then it would be closer to a popular election. But 47 states have *individually* decided to collectivize their citizens' votes (rather than individualizing them heavily).

That's about as fair as you can get-- letting the people vote, and even letting the people (collectively) decide how their votes should be counted. Your assertion that it isn't fair just amounts to, "I want my ALL VOTES to be counted MY WAY, regardless of what other people want!"

* Simply Wrong, because you are arguing that the thing that constitutes us "shouldn't" constitute us the way it does--apparently because you, personally, don't like something about it right now.

Yeah, none of this makes any sense. It's like if Budweiser and Pets.com both go out and buy a Super Bowl ad for $1 million. The Budweiser ends up getting a big sale boost because its ad was more engaging and its a more popular widespread product. In terms of converted customer, Budweiser did pay a lot less. No one with common sense would say that the NFL charged Pets.com more money than Budweiser.

"A click on an ad for a “mesothelioma lawyer” is extremely valuable because people who aren’t interested in hiring a mesothelioma lawyer are unlikely to click and those who do click are likely to become profitable clients. "

Once in a while, when I remember my disgust for politicians and lawyers, I will google mesothelioma, and click on all of the adds, to take some money out of the pockets of lawyers. I just wish there was a way to punish politicians likewise.

Mesothelioma is more valuable than the lung cancer of smokers, because Asbestos exposure is the only known cause, whereas lung cancer can be caused by "other", so there is plenty of reasonable doubt, despite a smoking client.

I don't find the mesothelioma explanation Alex gives complete, so I''m going to try and add a little more. :)

Clicks on advertising are a supply and demand issue. There is a lot of demand for mesothelioma clicks, but a very small supply of them, so the price gets bid up.

From an ad network perspective (Google/Facebook), they want to get the best prices from advertisers, but they also want to maximize the number of clicks from the ad viewers. If ad viewers click on ads and are then not interested in the resulting page, the ad viewers start ignoring ads more over time and stop clicking on them. So the system the more sophisticated ad networks have setup is that "better" ads, more related to the keywords and more beneficial (in a utility sense) to the ad clicker, get a discount and are shown more frequently/in better positions than less beneficial ads. This helps keep the overall supply of clicks up so the ad networks keep having clicks to sell.

The bottom line is that more relevant/better targeted ads cost less.

Banned again.

Well, there is always the option to just leave and go on with your life. Honestly, if your not wanted why would you stay. It's not as if this is the only blog on the internet. I'm sure another site would be happy to read your valuable insights.

"It’s not as if this is the only blog on the internet. I’m sure another site would be happy to read your valuable insights."

Don't you think I haven't thought about it? But if I do, they win.

No, don't fall for their tricks. They are the ones who just don't value you. You're better off starting fresh with someone who appreciates you for who you are.

If I give up now, they will have won. They will have prevailed over me. I can not accept that. In the warriors' code, there's no surrender.


Out of the night that covers me,

Black as the pit from pole to pole,

I thank whatever gods may be

For my unconquerable soul.

In the fell clutch of circumstance

I have not winced nor cried aloud.

Under the bludgeonings of chance

My head is bloody, but unbowed.

Beyond this place of wrath and tears

Looms but the Horror of the shade,

And yet the menace of the years

Finds, and shall find me, unafraid.

It matters not how strait the gate,

How charged with punishments the scroll,

I am the master of my fate,

I am the captain of my soul.

Is the controversy that the incompetent one didn't win the presidency?

I suppose that if you sell things to the US Federal government you really want the one who overpays to win.

I think Cowen misses the point. Ask people what business Facebook is in, and they will respond social media. Ask people what business Google is in, and they will respond search engine. No, they aren't, they are in the same business, digital advertising. Of course a company in the business of digital advertising will adjust its advertising rates by the clicks and buzz generated by the ad. But not a company in the business of social media or search engine. If you're not paying, you're the product.

Facebook and Google together capture close to 70% of total revenues from digital advertising, with Google capturing the lion's share. They are very good at the business they are in, which is digital advertising.

I'm not going to bother looking it up, but I wonder what percentage of ad sales are digital. I hear a radio commercial every so often that claims radio reaches more people than TV or online. Even for millennials.

Well, for Facebook and Google, 100%. Alas, for old media a growing percentage, which they are losing to . . . . Facebook and Google.

Tabarrok, not Cowen.

I’m not sure you are correct about this, but never mind. I will say this. TC is missing A point: the left is not just lashing out, it is showing Facebook what is in store if it doesn’t get on board.

Remember when paid search results were discreetly to the right side?

Yeah, now the entire screen is ads.

People who are lazy, will click the ad and the company pays a Google tax.

Seems like clustering by goods/service type solves this problem.

I resent having to say mesothelioma that many times in my head. Couldn't you have found a more succinct cancer?

Pneumonoultramicroscopicsilicovolcanoconiosis is also a disease caused by a specific workplace hazard. We could use that instead.

Try saying brain in your head instead. This has the added virtue of being succinct and ironically amusing!

The pay-per-click model makes a ton of sense if you assume that, for a given advertiser, every click is equally valuable and every non-click impression has zero value. The model starts to get gameable when these assumptions fail to hold: for instance, Google's and Facebook's algorithms reward strategic bidding if you have budget constraints. As we see here, there start to be interesting consequences when people care intrinsically about impressions as much as -- or more than -- clicks. _This is why Facebook lets you bid for impressions rather than clicks!_ And the Clinton campaign could outbid Trump's CPMs if they wanted people to see their ads and not Trump's.

If the Clinton and Trump campaigns actually valued clicks then the argument is...more people clicked on Trump's ads because people were more willing to click on Trump's ads?

Could the issue really be that political ads are in thrall to "the market"? Are political offices "products"? Should they be?

"Why Did Trump Pay Less than Clinton For Ads on Facebook?"

Because "the strong do what they can, the weak suffer what they must"?

The bots generate a lot of Rump, excuse me, Trump traffic,

So maybe it wasn't Trump followers


Rather bots which in effect reduced the price

Or, perhaps it was Trump that overpaid because there wasn't a human at the other end.

"Or, perhaps it was Trump that overpaid because there wasn’t a human at the other end."

Trump paid roughly the correct amount. Hillary horribly overpaid.

"I think Trump conspired with the Russians to have bots click through Trump ads in order to cost his campaign money for fake views!"

Dude, you need to call Mueller and let him in on this.

Interesting hypothesis, but not what I said.

What I did say was that bots amplify traffic, and a price based on traffic that does not recognize distortion is not a true metric.

Except you specified Trump specifically. What was the magic shield that kept the Bots from clicking on them? If both sets of ads were subject to roughly the same amount of Bots activating them, then it really is just noise.

Your assumption is that the Russian bots forwarded Hillary's links and material in the same proportion as Trumps.

My assumption is that the results are the same unless proven otherwise. If there's a differential, it's on you to post some proof of it. Otherwise, I'll just assume you aren't smart enough to lay out an effective case.

JW, are you really that ignorant of current events or facts to say that you have know awareness of the news that Russian bots retweeted and forwarded to facebook accounts.

Man, you have a problem. You can read real information about this here: http://money.cnn.com/2018/01/27/technology/business/russian-twitter-bots-election-2016/index.html
Sometimes when you engage in a discussion you do not assume that someone chooses to ignore common knowledge, or ignores the news around him so that you can have a discussion.

"Twitter stressed throughout the report that the behavior of the accounts in question represented "a very small fraction of overall activity" on its platform in the lead-up to the election. The 2.1 million tweets sent by the automated accounts represented 1% of election-related activist tweets on its platform, Twitter said. "

Why don't you blame the weather instead? It would be more persuasive.

The link you pointed to doesn't have anything to do with Russian bots clicking Trump ad banners. So, put up or shut up.

Trump knows the art of the deal, and gets better prices.

It makes sense, I think.

so we can infer that Wired is looking for lower FB ad pricing by writing such a provocative claim that Trump paid lower prices -- suggesting perhaps FB favored them or is a better negotiator or something?

Wonder if the controversy would have been the same if the claim was more like "Trumps ads on FB are worth less than Clinton's ."

I use FB and Instagram to promote my business. When I saw the amount of money the Russians allegedly spent to manipulate the election I laughed out loud. The only reason we’re still talking about it is because the media can’t let go of their narrative.

Now consider this for malign influence on the electorate. After the tax cuts were announced it was very clear that a broad-based cut was coming to most Americans. But polls showed most people didn’t think they were getting a cut and opposed it (I think one poll had it about 80-20 against)

Why? Because the media - not Facebook - chose to distort the message and spread disinformation. Polls now show most Americans support the tax cuts but only after they saw their paychecks.

Same goes for Russian collusion. Over half of Dems think the Russians actually hacked the tally of votes.

The media has become a truth-distortion field.

The only reason we’re still talking about it is because the Trump appointees in the intelligence community are still warning about Russian interference in American elections.

Wait, that wasn't the narrative, it is reality.

No, the narrative is that Trump won the election due to Russian help. And 52% of Democratic voters (in a poll from before Trump took office) believe that the Russians actually hacked the voting machines.

Of course, the narrative is completely wrong. There's no evidence that Russian interference had any effect. And it's laughable that spending $100K on Facebook ads would move the needle on an election where $2.6 billion was spent on the Presidential campaign.

Exactly. If you imagine the media's job is to educate the electorate, this would be fraud.

"The media has become a truth-distortion field."

You know, like it made Saudi terrorists into Iraqis a fewyears ago because Saudi Arabia is our friend and always will be?

Well you gotta give Trump credit where it's due. He probably made more money from selling hats than he spent on his own campaign, and he ended up winning. The man's a winner, all things considered.

But was that worth it?

Years ago when I was getting paid (normally a pittance) by Google Ads for keywords, I made sure that any post I wrote that mentioned Stephen Jay Gould also mentioned that he had died of mesothelioma in the hopes of getting one of those $73 payoffs for a reader clicking on "mesothelioma."

Never happened. Oh, well ...

Former VP of Ads at Facebook just contradicted this. Trump campaign paid a higher CPM than the Clinton campaign. https://twitter.com/boztank/status/968577962223136768

So the original article was classic Fake News. Great. I look forward to the retraction, and for all the people who were upset about Trump getting a better deal to now be upset about Clinton getting a better deal.

Honestly, this update, couldn't be funnier.

Facebook does not include predicted engagement as a factor in its ads auction to solve the “mesothelioma lawyer” problem. Granted, it might help; but Facebook builds ads targeting products, ads outcomes measurement, and the ability to bid for outcomes as the intended solutions to this problem.

The reason Facebook includes predicted engagement as a factor in the ads auction is because engagement is the opportunity cost of ads to Facebook. That is, every time Facebook shows an ad, it costs Facebook a little bit of user engagement, due to the fact that the ad is generally less engaging than whatever content Facebook would otherwise show in its place. The ad has to pay for this opportunity cost, and it does so via its bid at auction. The opportunity to show an ad goes to the highest bidder who also covers the opportunity cost; if no ad bids high enough to cover the opportunity cost, Facebook will not show an ad. This also helps Facebook cap the engagement cost of showing ads. Predicted ads engagement enters the ads auction because that reduces the opportunity cost the ad bid has to cover.

Comments for this post are closed