5 ms·

Of course Arbitrage bothers you a lot, it means that you aren't capturing the full value of the impressions you sell/broker. Bad behavior is not desired by per

Of course Arbitrage bothers you a lot, it means that you aren't capturing the full value of the impressions you sell/broker. Bad behavior is not desired by performance advertisers. The blue chip advertisers you refer too are getting wined and dined by agencies and ad tech companies swindling them into believing it's impossible to quantify emotion and storytelling and brand building. After all, it's not their money, it's their shareholders money. Find me one media buyer, where its actually his own money that doesn't demand accountability. What kills me is that letting this bad behavior thrive is short sighted. It's great for keeping VCs happy about growth, but at the end of the day, it devalues all your other high value inventory and forces advertisers to underwrite the risk in their bidding. It's really bad for business long term...but as long as VCs fund publishers and ad tech companies their incentive is only short term growth and bots and fraud help those numbers. Someone, Somewhere needs to force accountability on every piece of the chain, or we'll continue down this path where 99% of advertising isn't working.

x0x0 10y ago
The idea that media purchasers don't track return on ad spend is utterly disconnected from any reality in the adtech industry. Here's what actually happens: a DSP is given a $30-$50k trial buy from either a company or their agency. While it's certainly true that a dsp is expected to spend the entire budget, the returns on that budget are carefully tracked. If you don't perform, you get $0 next quarter. If you do perform, you call the buyer and demand 2x the budget next quarter where the entire justification is the cost-per-[conversion, click, etc] that you delivered.
- davemel37 10y ago
  You can make data tell any story you want with enough creativity. They use ridiculous attribution modeling to justify all sorts of ROI that doesnt really exist. They aren't actually truly measuring impact, they are justifying an expense where there are no ramifications if their modeling doesn't reflect reality. I worked for a company that lived and died by paid media. If something really didn't work, people lost their jobs. Without that level of accountability...it is just a bunch of BS people sell themselves to justify their decisions and/or make themselves look good. You and I both know that ad tech spin data in all sorts of ways to make them look good. I mean, do you really think a below the fold, unseen ad impression that was delivered to someone who bought 3 weeks later should get any credit for that sale? Do the exchanges report on view through conversions? Of course they do... Even, an above the fold impression, that was seen by a human being, is still 90% of the time ignored by ad blindness, but if that guy buys 29 days later, they'll still call it ROI. Let's not pretend there is true accountability in digital media today. We both know thats just not true. Edit: I thought I was replying to someone else and said "your exchange and corrected it to the exchanges."
  - x0x0 10y ago
    Conversion modeling has it's issues, but run eg a state-wide geo fenced test and the ads work. And, again, all fraud and/or fake conversions or whatever it is you believe is rampant gets priced into what companies pay for ads. With 0% fraud you can pay up to (discount x conversion value) / (ad count). With fraud, the numerator doesn't change. Buyers just spend less on ads. There's lots of ad buyers making lots of money buying ads and they're going to keep making lots of money buying those ads. And, to repeat myself, the idea that buyers don't push on how conversions are calculated plus run their own calculations is true only in a fantasy world.
    - davemel37 10y ago
      > gets priced into what companies pay for ads. This causes adverse selection which devalues high value impressions, which ultimately means more of the advertisers money is wasted on fraud, arbitrage, and poor placements. You can assume that it's all good because they price it into their bids...but if you really think about it, a 3% conversion rate is a 97% failure rate. This massive inefficiency should not exist in todays world of Machine Learning...problem is, the stakeholders all benefit from bad behavior. > the idea that buyers don't push on how conversions are calculated plus run their own calculations is true only in a fantasy world. This is true in a broad sense, but those "conversions" are not always based on whats actually driving sales. Everyone wants to be in a defensible position, so they push for accountability...but its not real accountability, it is anecdotal storytelling that aligns with their interests instead of the bottom line sales. I have clients that are publishers who only want to show growth in unique sessions. I have clinical trial clients who want to make sure their executives and shareholders see their ad, Retailers constantly force manufacturers to invest in TV, OOH and other media in order to even get shelf space, big brands are spending money on the recommendations of kids a year out of college who got dazzled, wined&dined, and flattered by account executives at agencies and ad tech providers, or their friends who got a job at a hot new startup. The incentives to truly be disciplined instead of just telling a good anecdotal story simply don't exist in the real world. It only exists in private companies where the decision maker is spending his own money and actually knows better. I really hope you don't believe that digital advertising today is mostly accountable to statistically significant and disciplined calculations about direct correlation and indirect impact on sales. They spin their tales with data that makes everyone feel good, but doesn't really reflect the true impact of their advertising. It is an industry with incentives aligned with all sorts of things besides true performance. Worse is the first party data movement that is essentially tricking brands into spending money they dont need to spend to drive sales they would get anyway. Find me an ad tech company who says, well save you money and Ill trust them. Find me an ad tech company that says, start with $50k and than next month well do $30k for the same volume of sales and leads... and ill trust them. For now, they all say well make you money, and than they scam their way into taking a piece of a pie they never really earned.

x0x0 10y ago

The idea that media purchasers don't track return on ad spend is utterly disconnected from any reality in the adtech industry. Here's what actually happens: a DSP is given a $30-$50k trial buy from either a company or their agency. While it's certainly true that a dsp is expected to spend the entire budget, the returns on that budget are carefully tracked. If you don't perform, you get $0 next quarter. If you do perform, you call the buyer and demand 2x the budget next quarter where the entire justification is the cost-per-[conversion, click, etc] that you delivered.

davemel37 10y ago
You can make data tell any story you want with enough creativity. They use ridiculous attribution modeling to justify all sorts of ROI that doesnt really exist. They aren't actually truly measuring impact, they are justifying an expense where there are no ramifications if their modeling doesn't reflect reality. I worked for a company that lived and died by paid media. If something really didn't work, people lost their jobs. Without that level of accountability...it is just a bunch of BS people sell themselves to justify their decisions and/or make themselves look good. You and I both know that ad tech spin data in all sorts of ways to make them look good. I mean, do you really think a below the fold, unseen ad impression that was delivered to someone who bought 3 weeks later should get any credit for that sale? Do the exchanges report on view through conversions? Of course they do... Even, an above the fold impression, that was seen by a human being, is still 90% of the time ignored by ad blindness, but if that guy buys 29 days later, they'll still call it ROI. Let's not pretend there is true accountability in digital media today. We both know thats just not true. Edit: I thought I was replying to someone else and said "your exchange and corrected it to the exchanges."
- x0x0 10y ago
  Conversion modeling has it's issues, but run eg a state-wide geo fenced test and the ads work. And, again, all fraud and/or fake conversions or whatever it is you believe is rampant gets priced into what companies pay for ads. With 0% fraud you can pay up to (discount x conversion value) / (ad count). With fraud, the numerator doesn't change. Buyers just spend less on ads. There's lots of ad buyers making lots of money buying ads and they're going to keep making lots of money buying those ads. And, to repeat myself, the idea that buyers don't push on how conversions are calculated plus run their own calculations is true only in a fantasy world.
  - davemel37 10y ago
    > gets priced into what companies pay for ads. This causes adverse selection which devalues high value impressions, which ultimately means more of the advertisers money is wasted on fraud, arbitrage, and poor placements. You can assume that it's all good because they price it into their bids...but if you really think about it, a 3% conversion rate is a 97% failure rate. This massive inefficiency should not exist in todays world of Machine Learning...problem is, the stakeholders all benefit from bad behavior. > the idea that buyers don't push on how conversions are calculated plus run their own calculations is true only in a fantasy world. This is true in a broad sense, but those "conversions" are not always based on whats actually driving sales. Everyone wants to be in a defensible position, so they push for accountability...but its not real accountability, it is anecdotal storytelling that aligns with their interests instead of the bottom line sales. I have clients that are publishers who only want to show growth in unique sessions. I have clinical trial clients who want to make sure their executives and shareholders see their ad, Retailers constantly force manufacturers to invest in TV, OOH and other media in order to even get shelf space, big brands are spending money on the recommendations of kids a year out of college who got dazzled, wined&dined, and flattered by account executives at agencies and ad tech providers, or their friends who got a job at a hot new startup. The incentives to truly be disciplined instead of just telling a good anecdotal story simply don't exist in the real world. It only exists in private companies where the decision maker is spending his own money and actually knows better. I really hope you don't believe that digital advertising today is mostly accountable to statistically significant and disciplined calculations about direct correlation and indirect impact on sales. They spin their tales with data that makes everyone feel good, but doesn't really reflect the true impact of their advertising. It is an industry with incentives aligned with all sorts of things besides true performance. Worse is the first party data movement that is essentially tricking brands into spending money they dont need to spend to drive sales they would get anyway. Find me an ad tech company who says, well save you money and Ill trust them. Find me an ad tech company that says, start with $50k and than next month well do $30k for the same volume of sales and leads... and ill trust them. For now, they all say well make you money, and than they scam their way into taking a piece of a pie they never really earned.