Programmatic Is Problematic, Even When It Works

This post was previously published in an earlier edition of Online Spin.

The problems with programmatic are well-covered and greatly ignored.  Reported growth in spending continues to fan the frenzy, while sweeping significant problems under the rug.

The biggest problem with programmatic, of course, has to be the non-human traffic.  

Bots surf the Web the way a human would, so cookies get dropped and that browser seems like a legitimate target audience.  The bot then travels to a bullshit site with bullshit content, and programmatic triggers the serving of ads. So advertisers pay real money to fake sites for ads seen by no one.   

Talk to the brilliant Professor David Carroll (@profcarroll), who can break down this problem better than anyone while painting an even darker picture of where these ad dollars end up.

Today however, let’s stick our head deeper in the sand and ignore this non-human traffic thing for a minute.



Programmatic still doesn’t work.

Over 70% of  overall spending in programmatic is for open exchange inventory, with millions of sites competing to earn those ad dollars.  Sites that rely heavily, or exclusively, on this kind of ad revenue are rewarded for creating a disastrous user experience.  That’s because publishers cashing programmatic ad sales checks are incented to jam in as many ad impressions per page view as they can. is an example of one of those sites.  It should have its publishing license revoked.   The fact that this site resides in the Bay Area -- where digital advertising all started -- is a sad and ironic joke.

On any given day,, which I find when Yahoo publishes leads to its stories, will run up to 12 different display ads on its home page. All appear to be sold programmatically.   That’s not counting the native content swimming in this cesspool of a site, either.  

But, OK, let’s stick our head a little further in the sand and pretend 12 ad units per page view is not a big deal.  Programmatic still has fundamental problems.

Close to 90% of ad impressions bought through programmatic open exchanges are based on retargeting.  Visit an advertiser’s site, get a cookie taped to your browser like toilet paper stuck to your shoe -- and now, wherever you go on the Web, publishers over-serve you that advertiser’s ad.

I have two cats, 16-year-old brother and sister named Momo and Maxine.  The latter is starting to miss the box, so I went to a while ago to see what solutions exist.   

As a pet owner, I tend to read stories on Yahoo about pets. I recently read a headline about a dog that was killed by a pet groomer working at Petsmart.  I clicked through to read the whole story, and was met with this:

Before you argue, “this is a one-off impression,” please consider this: Millions of “pet enthusiasts” like me were served a snippet of this story on Yahoo.  Many felt their heartstrings pulled and clicked through to SF Gate to read the story.  A large percentage of “pet enthusiasts” have cookies from a pet site, so it’s likely that many people were served the worst ad placement in the history of advertising.

It gets worse.  I searched for this story a week after I first encountered this ad placement, and Petsmart ads are still serving to my browser.  No one has a clue on where the majority of programmatic ads sold are served, so no one at Petsmart even knows they need to stop this.

The programmatic world is so good at touting their success  -- but they never report on the 99% of the people who did not click, who did not convert, who did not generate attribution credit.

I would love to know, of the 99% of the people who saw this ad and did not engage, how many now have a negative opinion of this programmatic advertiser?  What percentage was so offended by the insensitivity of this ad placement that they won’t ever shop at Petsmart?  What percentage now thinks Petsmart is Petdumb?

Programmatic doesn’t work. Ads are served when a cookie in a browser says it’s the right person -- while so often, the content where the ad is seen is saying it’s the wrong time.

3 comments about "Programmatic Is Problematic, Even When It Works".
Check to receive email when comments are posted.
  1. Billy Tarter from Leads Dynamo, February 13, 2017 at 4:15 p.m.

    Agreed that SFGate has a lot of ads on their home page.  However, it is a very long page which is intended to be scrolled, and it doesn't feel overbearing.  Whether there are too many ads which could turn readers off and lead to diminishing returns for the publisher is up to them to decide, and I'm sure there have been multiple discussions and tests to determine the optimum number.

    Agreed also that the PetSmart ad placement is tragic, but it's also avoidable.  There are numerous brand safety tools and semantic analysis vendors which can be utilized to avoid exactly this issue.  PetSmart's media buyers should be monitoring for this.

    HOWEVER, it's indefensible to issue a blanket statement that "programmatic doesn't work" based on two anecdotal examples and no concrete data or proof.  Programmatic does work – it works every day for the clients that we service, and the data proves that out.  Sure there are challenges with programmatic, but there are solutions as well.  One of those solutions is certainly not to stick one's head in the sand and proclaim loudly that it doesn't work.

  2. Ari Rosenberg from Performance Pricing Holdings, LLC replied, February 13, 2017 at 9:12 p.m.

    Billy thanks for chiming in -- this column ran a few months ago initially, and if I could have reqwritten it I would suggest programmatic doesn't work for publishers nearly as well as it works for advertisers -- your clients as you point to.  The table is slanted and in an attempt to even things out pubs like SF Gate load up their pages so everyone's value get hurt -- most importantly the user -- this is a horendous user experience and there are tens of thousands of sites just like SF Gate who try to make money through programmatic at the expense of their users and then just buy cheap traffic and start the cycle all over again each month.  My head is not in the sand -- and I bet you regret saying that -- just sounds mean.  I looked you up on Linkedin and I plan to call you -- because you look like a nice guy and certainly a smart one.  I respect you taking issue with my column -- really I do -- it sparks debate and keeps me on my learning toes -- so thank you and I would love to have a beer with you one day -- but only if you promise not to yell at me and tell me my columns are indefensible -- wait unti we do a jaiger shot before we get to that point.  Thanks for reading me and for disagreeing.  I am calling you this week :)

  3. Billy Tarter from Leads Dynamo replied, February 14, 2017 at 11:19 p.m.

    Haha, thanks Ari for your response.  Fair enough that my comment was strongly worded (I actually edited it down before posting), but... I don't think it was mean, and I was matching the tone of your article.  I think you wrote this article in an inflammatory tone on purpose, which is your prerogative of course - but if you're going to stir the pot like that you have to be prepared for strong responses.  There's a lot of hyperbole and no sources quoted, so I have call out that kind of thing when I see it, and I still feel it's indefensible.  No personal offense intended, of course.  And for the record, the "head in the sand" analogy came from your article.  :)

Next story loading loading..