The Economics of Buying Better Digital Ads. You Can Too
programmatic supply chain waterfall chart

The Economics of Buying Better Digital Ads. You Can Too

Previously, I wrote about the Economics of Ad Fraud (Forbes, June 2020). We talked about the ease of carrying out ad fraud, how scalable it was, and how low cost and low risk it was for criminals to continue it. Today I will talk about the economics of buying better ads and why advertisers should pay higher CPMs and buy from real publishers with real human audiences.

CPMs are prices, not cost

You want to do better digital marketing, right? You want your digital marketing campaigns to work better, right? Well, that means more business outcomes and lower cost per outcome, not lower prices, more quantity of ad impressions, and more clicks. Digital marketing is not a video game, where bigger numbers mean you win. I see far too many advertisers boast about their "spend" and not enough of them boast about their outcomes. But outcomes are what matter, not how much you spent or how big your digital ad budgets were.

Even though CPM stands for "cost per mille" or "cost per thousand" it's a horrible misnomer that has misled advertisers for the last 2 decades. CPMs are prices, the price you pay for every thousand ad impressions. Price times quantity gives you your cost. Years ago you purchased ads at $30 CPMs (price). Now you buy ads at $3 CPMs (price). But you are buying 10 times the quantity. So your total cost is still $30 ($3 x 10). In other words you did not save any costs by buying at $3 CPM (price). $3 is not your cost. But, by buying at such low CPMs, your ads are not going to legitimate publishers with human audiences. That is because they cannot afford to sell ads at such low CPM prices. They have real journalists, and real editors, and real costs of producing the content. Fake sites have little to no costs of content because they just plagiarize it all, text and pictures too.

Greater than 50% tax

When you buy through programmatic channels, you are also paying at least a 50% tax, which is hidden from view and typically ignored by the advertisers. For every dollar you spend, less than 50% of it goes to the publisher for showing the ads. Think of this as the equivalent of "working media" in traditional advertising like TV advertising. You want most of your dollar to go towards showing ads, not to line the pockets and profit margins of ad tech middlemen, right? As long as you buy through programmatic channels, you are paying at least a 50% tax to these middleleeches, and it's usually hidden from view. How does that happen?

Let me use some round numbers to illustrate. You tell your media agency you want to pay a maximum CPM of $10. They say OK, and they turn around an extend a bid on your behalf of $8, assuming they only take a 20% margin for themselves. Then, the DSP takes 25% for themselves, which means they extend a bid of $6 on your behalf ($8 minus 25% of $8). Then, the SSP which represents the publisher takes 25% for themselves, so the publisher is left with $4.50 ($6 minus 25% of $6). In the simplest of examples above, the advertiser effectively paid a 55% tax -- they spent $10 CPMs and publishers got $4.50 CPMs, the rest went to middlemen. The reality is far far worse than this, but let me not burden you with too much reality at once. Suffice it to say the advertisers never get to see what the publisher got; and publishers never get to see what the advertisers bid. So the "con" continues. You can avoid this scam entirely by buying direct from publishers, no bidding necessary, and no middlemen necessary. (I can show you how, DM me).

Fraud, waste, and other crap

The above had not taken into account ANY other problems like ad fraud, waste, and "unknown delta" (ISBA could not figure out where the money went). I wrote a whole slide deck 2 years ago on this, but I am sure you didn't read it. ;-) That's fine. The punch line is in the slide below. Taking into account costs that can be known (like what we said above) and other factors like domain spoofing, ad blocking, IVT (bots), ads not being rendered on screen, not viewable, and not on target, we get to a number of about 1%. This is the percentage of ads that can be plausibly considered productive. Again, this is too hard to swallow, so let me not burden you with too much reality at once.

No alt text provided for this image

Let me show you just how few humans we can confirm with FouAnalytics in programmatic media buys compared to humans on real publishers' sites. In the table below, we looked across 22 different programmatic sources and a dozen real publisher sites. Note the yellow and green highlight under the column "unique humans." The percentage of humans on real publisher sites is around 100X that in programmatic channels. There are so few humans in programmatic channels, your chances of getting an ad in front of a human is less than your chances of building a space ship to go to Mars yourself. Substitute any other analogy in here that would make the same point. Considering that humans use ad blockers and bots don't, your ads are shown even more disproportionally to bots through programmatic channels, even if the legacy fraud verification vendors fail to mark most of them as IVT ("invalid traffic").

No alt text provided for this image

Still don't believe me? Let's look at the clicks that come from programmatic campaigns. You've seen this chart before -- it shows the clicks arriving on the landing pages of 6 different programmatic campaigns. Orange means declared bots. Red means bad bots. And dark blue means humans. You can see the blue if you squint. There's 1 - 4% dark blue, which means 96 - 99% of the clicks coming from paid programmatic campaigns cannot be confirmed as humans.

No alt text provided for this image

Whereas, when we look at the clicks that can be confirmed as humans (from campaigns purchased from real publishers), we can see most of the clicks are light and dark blue (humans). That's night and day different, right? And it is consistent with the paucity of humans in programmatic supply sources in the table above. The low CPM prices, high quantities of ads, and high clicks in programmatic channels are all from bot activity. Time to move on to more mature digital marketing, and that means CPM prices will be higher, the quantity of ads available to buy will be lower, and the number of clicks will also be dramatically lower. But your costs will be lower and your outcomes will be higher.

How do we move forward?

Hopefully the above has convinced you that buying better digital ads is better than buying low CPM (price) ads in programmatic channels. You may have even known that already. But perhaps there are other barriers keeping you from moving forward. Whatever those barriers are, here's how other advertisers have broken through, despite 10 years of buying into programmatic buying.

Create a new campaign line that uses an inclusion list approach (I have lists I can share with you, if that is helpful). This helps you avoid the 99% of the other sites and apps that are fraudulent and fake. Use outcome-related metrics for this campaign line, instead of easy-to-measure quantity metrics like numbers of impressions, numbers of clicks, and average CPM prices. Allocate 1% of your digital budget to this line and run it. If you see that it is producing more outcomes and the costs are lower (total cost and cost per outcome), then allocate another 1% budget to this campaign line, by reducing budget to legacy campaign lines. Even if you are paying higher average CPM prices by buying better digital ads from real publishers, you are buying far less quantity. Higher price times lower quantity will still result in lower overall costs. And because your ads are shown to 100X more humans on real publisher sites, they actually have a chance to drive more business outcomes for you. The 10 - 100X quantity of ads you used to buy through programmatic channels were shown to bots anyway so they won't have any chance of working -- i.e. driving more business outcomes for you. (Don't even get me started about how ad tech companies and fraudsters claim credit for sales they didn't actually cause.)

The above process allows you to transition to better digital marketing. Of course, it will be hard to watch CPM prices go up, numbers of ads go down, and number of clicks go down, because you've been used to those for so long. But it will be refreshing to see real outcomes go up, costs per outcome go down, and overall costs go down. Your boss and your company will thank you for being a better steward of the budgets you were given.

Onward and upward, my friends. Happy Sunday Y'all.

Dr. Augustine Fou

FouAnalytics - "see Fou yourself" with better analytics

1 年

Robert's advice below is SO good, it deserves to be memorialized in rhyme Divide and conquer, that's the rule Split your inventory, don't be a fool Into cohorts of three, or even more Premium partners, long tail, unknowns galore Get direct deals, and PMPs too With the premium partners, that's what you do Move your spend towards them, and watch it grow Your revenue will increase, that's for sure, you know Measure for outcomes, and measure for quality Be aware of weaknesses, don't be hasty Look at the metrics, dwell, viewability, and attention Alongside your attribution, that's the intention Measure incrementality, that's the way To see what works, day by day Measure for fraud too, that's a must Use a vendor like FouAnalytics, with trust Understand and optimize, that's the goal Be cynical, and don't be a fool People are trying to rip you off, that's true But success can be had, if you look for the gold Listen to everyone, but trust no one Validate with what you know, and have fun Follow these rules, and you'll be on top Your inventory will be gold, and never stop. signed: chatGPT

This completely meshes with my experience. I've seen "agencies" slap a fee AND an undisclosed "margin" on the CPM. In one case i saw a 20% fee and a "margin" of 60%. And that doesn't include DSP and SSP fees. I have a fair amount of experience wading into the logs and placement reports and understanding what fees are charged where is pretty hard. And understanding which placements are legit is also really difficult.

Alessandro De Zanche

Independent consultant - Senior advisor - Audience, advertising, media monetisation strategist

1 年

I can’t help but considering quality media owners equally responsible for all the above mentioned issues for legitimising with their presence the programmatic open marketplace.

Robert Webster

AI Solutions for Marketing

1 年

Lots of Good analysis here. Some solutions. 1) Divide your inventory up into 3 or more cohorts. Premium partner publishers, known long tail publishers and unknown . Get direct deals and pmps with 1 and move spend towards these. 2) Measure for outcome amd quality. Be aware of weaknesses in your measurement. Look at quality metrics like dwell, viewability, attention alongside your attribution. 3) attempt to Measure for incrementally. 4) Measure for fraud. Use a vendor like Fou analytics not tied to the ecosystem status quo. 5) With the above understand and optimise. Be cynical. People are trying to rip you off but success can be had if you look for the gold. 6) listen to everyone buy trust no one. Validate with what you know to be true. Problems exist but we can solve this.

回复
James Byrne

Senior Client Partner, Economist Impact

1 年

As usual Dr.Augustine, completely on the money

要查看或添加评论,请登录

社区洞察

其他会员也浏览了