Is All Programmatic
So, what exactly is programmatic advertising?
Programmatic advertising uses software to automate the decision-making process of media buying by targeting audiences through data. One way of purchasing ads programmatically is through a set of rules or targets that automatically perform real-time bidding (RTB) for online display ads, social media ads, tv ads and more.
Conceptually speaking, the promise of programmatic has been embraced by a large portion of the advertising industry, more so with larger companies than with agencies. Individual companies hear or read stories about the new goose that lays the golden eggs (i.e. programmatic) and these companies, of course, want to get one of these geese for themselves. Agencies are more aware of the intricacies involved with advertising such as the technology and methodology utilized to make the RTB decisions, serving the ads, and tracking the ads, as well as attribution. While some agencies have opted to go “all-in” by using various programmatic partners or even creating their own exchanges and technology platforms, other agencies have been much more cautious in their blind acceptance of programmatic.
The 2015 Strata survey stated that about half (46%) of agencies are unsure if they trust programmatic technology to execute ad buys.
An ANA/Forrester survey found that about 40% of client-side marketers admitted to low or no understanding of programmatic. These numbers indicate that as an industry overall, both companies and agencies, we really are not fully sold on this relatively new and evolving tech, yet programmatic spending continues to blindly increase at an astounding rate with $22.10 billion expected to be spent towards the trend in 2016 according to eMarketer (April 2016).
While in principal this does sound like an excellent method of buying advertising with many apparent benefits, there are fundamental issues that your local programmatic salesperson is not really disclosing in their sales pitch. The root of this issue is that programmatic has really done nothing to fix the inherent flaws of running display advertising in a network or in automating the buying process. In fact, it appears to have made it worse (more on this later). Sadly, most programmatic buying is being purchased with no regard to the issues at hand. It has replaced a bulk of media plans for no other reason than the fact that it is the current “buzzword” or because someone’s boss heard about it and caught on to the misguided trend.
According to eMarketer, the leading obstacles with programmatic advertising are fraud, site quality/brand safety, waste (ads that target the wrong demo), and viewability.
Incapsula reports that in 2014, 56% of all web traffic was bot traffic. That is an obvious concern since the belief for those purchasing programmatic is that their ads will be seen by an engaged consumer. Any ads that are being shown to something other than a human is the equivalent of marketing dollars being thrown out the window. Adweek states that bot fraud will cost the industry $6.3 billion in 2015.
Site quality is an issue because the types of sites you are expecting to be on, based on the preview inventory, often ends up being incorrect due to the frequent misrepresentation of available inventory by the programmatic vendors. Often viewability, brand safety, premium inventory and other critical measures that could be put in place may not be initially offered or carefully considered.
Viewability is a big issue with programmatic buying since inventory across all of the exchanges often deal in remnant ad space.
More often than not, the remnant inventory is located in positions or pages that are not optimal thereby having less impact on the user. Of course, some programmatic media providers can offer to sell you guaranteed viewable ads for a substantial premium. If they are offering these premium viewable ads, then they are evidently aware of the fact that much of their non-premium inventory is, in essence, garbage and it is not having any positive impact on the consumer response and brand exposure. One could say that programmatic was originally created to automate the process of buying bad media because of the viewability aspect.
The proliferation of programmatic advertising and heavy fraud has actually led to a decrease in viewability. According to DoubleVerify, all display based advertising has seen an average decrease of 5 percentage points in viewability in 2014 and we can expect to see this trend continue if the industry doesn’t adapt. Viewablity is a particularly sensitive subject for programmatic buying since the vast majority of programmatic ad conversions are captured off of a “view” only interaction, which becomes completely meaningless when the ads were never possibly viewed if below the fold or never even loaded.
For those of you that have been around for some time, you might find this somewhat familiar since many of the issues with programmatic buying are similar to the issues that were associated to poorly run affiliate programs in the past. The improper usage of the brand, the bait and switch revolving around what was expected to be delivered as advertising versus what was actually delivered, and the constant need to police the buys in order to ensure that rules, targets, geos, etc. are being complied with are all similar to the affiliate advertising challenges. Ironically, programmatic was heralded as the automation of the media buying and optimizing process, yet it only gathers the data and operates within the boundaries of established rules while the real insights are still being discovered by the team of humans working with these systems.
Despite the numerous obstacles that programmatic advertising presents, is there still hope for the trend? In our next article, we examine the current state of affairs, and whether programmatic will become a truly successful and effective media buying technological platform.
Despite the numerous obstacles that programmatic advertising presents, is there still hope for the trend?
In our first article, we examined the rise of programmatic advertising despite increasing evidence proving the complications in the technology. Here, we look deeper into the hurdles, and examine whether there is any shed of light at the end of the tunnel.
Rough waters can be seen ahead for programmatic advertising, since the new understanding is that programmatic needs to evolve in order to have any continued success. Problems with inventory quality, fraud and true brand lift are drawing quite a bit of attention, and when it comes from a big brand in digital advertising, people will tend to listen. In early 2016 Facebook announced that it was canceling their plans to build a new Demand-Side Platform (DSP) that would leverage both Atlas (the ad server they acquired in 2013) and LiveRail systems (the video exchange they acquired in 2014), which had been in development since the latter half of 2015. The combined efforts of these three companies and the subsequent testing of this new system brought to light the fact that DSP ad inventory wasn’t up to Facebook’s standards. Dave Jakubowski, head of Ad Technology at Facebook, shared:
“We came up against many bad ads and fraud (like bots). While we were fortunately able to root out the bad actors and only buy quality ads, we were amazed by the volume of valueless inventory… We knew that in good conscience, we couldn’t sell what Atlas and our people-based measurement told us was valueless. Unfortunately, those ads were almost certainly dumped into another low-quality exchange where all of them were most likely purchased.”
When an industry giant is willing to miss out on exorbitant amounts of ad revenue, it speaks volumes about where programmatic is right now.
While the evidence doesn’t seem to be in programmatic advertising’s favor, it is of note that most programmatic advertising is not being purchased nor run correctly. The great many hurdles within the programmatic space exist largely due to poor implementation, lack of data analysis, and vendor inventory mismanagement. This, in turn, is having a negative impact on the general market opinion towards programmatic advertising. Is there light at the end of the tunnel? There certainly is. The trick here is that real programmatic initiatives aren’t cheap, nor do you simply jump right in and run with it.
Effective programmatic campaigns need to be architected, planned, and delivered in a cohesive manner that ensures relevant data is being leveraged to deliver impactful messaging to the right target audience at the right time… easier said than done.
Typically, before a campaign runs you’re asked to set goals with programmatic vendors. Often times most systems can only optimize towards one goal. Let’s say that goal is return on investment (ROI), if frequency isn’t taken into consideration then you may create message fatigue. Also, there may not be a guarantee that the user saw or was impacted by your ad. While rate structures like cost per acquisition (CPA) or cost per engagement (CPE) are making their way into programmatic, the offering is scarce.
The birthplace of a new programmatic campaign is not within the DSP platforms or with the media partners that exist within the open exchanges, it begins with the advertisers themselves. Programmatic is a concept that needs to be presented, discussed, and finally embraced by the advertiser wholeheartedly before beginning down the path of creating a programmatic program. There is considerable cost, both internally and externally, to design and implement an initiative of this sort – and it will take time, so rushing is not an option.
In our previous two articles, we established what programmatic is, its rise in popularity, and its undeniable hurdles. In our final piece, we examine new technologies and set guidelines in tackling the hurdles and ensuring a safer and more successful experience in automated media buying.
Data is the lifeblood that brings a programmatic campaign to life and data is what keeps it healthy and thriving.
Data truly is the most critical aspect of a good programmatic campaign, but is often the most mishandled or ignored aspect of it. In order to leverage the power of this data, it must first be architected. Identifying what data is available and from what sources is the first thing to consider. Data can come from CRM systems, other departments, 3rd party vendors, and other digital or even offline sources can be integrated as part of architecting the data. As part of this data architecting process, identifying the data will lead directly into defining how the data is provided, who are the responsible parties, who will have access to the data and to what extent that access will be given. And finally, who will own the data.
If data is the lifeblood of a programmatic campaign, then the heart would be the data management platform (DMP) that collects and analyzes all of the sources of data that were architected.
The DMP enables the advertiser to segment, filter, trend, and in all manners, process the data in order to identify customer segments that will be the most lucrative for the programmatic advertising to reach. The DMP will continuously evolve as the data aggregated into it will provide an ever refining and evolving data set that will be usable to drive intelligent targeted marketing decisions for all channels, not just those that are programmatic in nature.
This requires parsing the data or structuring it into a cohesive pool of data points that can then be mined. The beauty of what programmatic could be is being able to, in real-time, take action on what’s going on, not only in the market, but also with your customers.
Having created a strong DMP and filled it with relevant data, the next step will be to integrate with at least one Demand Side Platform (DSP). A DSP enables advertisers to buy advertising across multiple publishers and to specific targets based upon the advertiser’s DMP data as well as other information such as their geolocation and recent browsing behavior, if available. Publishers make a portion of their ad inventory available within ad exchanges and then DSPs automatically decide what part of that available ad inventory makes sense for the advertiser to purchase based upon the advertiser’s specific needs.
Not all DSPs are created equal, so the process of selecting a DSP that is right for an advertiser requires thought and planning. The advertiser must understand the type of data that they need and then determine which DSP partner and which DSPs have that data available. DSP reach is another factor to consider since geo locations covered, channels covered, and which ad exchanges are linked to the DSP are all indicators of the type of inventory that will be available for the advertiser to purchase. Speaking of purchasing, cost is also an area that requires investigation since DSPs have different cost models for inventory and DSPs will vary in regards to which, if any, additional fees they also charge (platform usage fees, per user charges, etc.) Hidden costs can often drive the expected investment for a DSP much higher than originally budgeted for. Finally, support and management are factors in selecting a DSP, since advertisers will have varying levels of expertise working with marketing platforms, so their ability to jump in and work within these DSP platforms will vary dramatically. How much should support hours or plans be needed? Should you consider a self-serve model, a fully managed model where the DSP manages much of the process for you, or perhaps a hybrid of the two? These are all important queries to contemplate when considering which DSP is right for you.
Connecting to a single DSP will allow an advertiser to obtain that particular DSP’s offered ad inventory but, as stated in the previous paragraph, not all DSPs are created equal. With differences in reach, audience, geo focus, and channel coverage, many advertisers may find one DSP that fits some of their needs but not all. This is the core reason that many advertisers will end up using multiple DSPs at the same time in order to expand the reach across all of the DSPs selected. This seems like the logical thing to do, considering the circumstances. The problem is that the resources required to manage multiple DSPs simultaneously, as well as having to painfully go through the process of comparing available inventory and costs across platforms that aren’t connected to one another, makes this very inefficient and also less effective than what one would expect. These inefficiencies are what drove the recent development of Meta-DSPs.
Right about now you might be wondering, “What exactly is a Meta-DSP?” A Meta-DSP is a new type of trading platform that connects the leading DSPs within a single interface, which allows for a much more holistic view of the programmatic inventory across all DSPs. To summarize, a Meta-DSP aggregates data from multiple DSP partners in order to provide advertisers with expanded reach, scale, insights, and cost savings. The concept of a Meta-DSP is something that popped up over the recent year or so and is gaining popularity.
Purchasing programmatic advertising through a Meta-DSP helps mitigate the previously noted challenges with managing costs and the manpower resources required to manage multiple DSPs individually.
Meta-DSPs are similar to standalone DSPs in regards to support and management of the product. An advertiser working with a Meta-DSP can work with the system themselves or go with more of a fully managed solution where the Meta-DSP handles the budgets, buying, optimizing, and the campaigns overall on behalf of the advertiser. Meta-DSPs also can provide the full package for advertisers by acting as both, the DMP and the Meta-DSP, although each client would need to assess their own programmatic plan and decide what works best for them when they are in the early stages of architecting their programmatic program. Exchangelab, one of the current leaders in the Meta-DSP market, has been very successful in cross platform/DSP integration.
Meta-DSPs helped solve part of the problems with the current programmatic landscape, but there is another recent development that promises to help programmatic achieve greater success and that is the introduction of header bidding (FYI, this is also known as advance bidding or pre-bidding.) Header bidding is a relatively new programmatic advertising technique in which publishers offer their inventory simultaneously to multiple ad exchanges and, based on the winning bid, then calls the ad server for the relevant ad. How is this different than what is currently being done, you wonder?
Typically, there is a sequential order in which bids are processed, more often than not it follows this model: the unsold inventory from a publisher is offered first to the top-ranked ad exchange, and if it is not purchased, it will then go to the second-ranked ad exchange and this continues down the list of ranked ad exchanges until someone eventually buys. The big problem with this is that these exchanges are ranked according to factors such as purchase volume, which logically makes it so that the inventory is not offered to the exchange or advertiser that might be willing to pay more or that has the higher demand, but rather to the exchange with the most purchase volume overall.
To complicate matters further, many of the publishers utilizing Google’s DFP ad server employ a setting that allows its Ad Exchange (AdX) to outbid all other exchanges by at least a penny if they choose to, which, in essence, always permits Google to get the last bid. This waterfall sequence of bidding has created inefficiencies and skewed the bidding landscape. Header bidding is a way to bypass all of this and some publishers are even including their direct sales volume in this head bidding process, which creates a much more competitive landscape in which ad exchanges and publishers will more openly sell their inventory across a broader landscape. This will enable advertisers to get a shot at the best inventory and poke a stick in the eye of Google Ad Exchange’s last bid rule.
Google has recently announced major changes planned for their Doubleclick for Publishers ad server in which they claim that they will begin to allow real-time bids from publishers outside of their Ad Exchange to compete within their own Ad Exchange for inventory. This announcement was clearly in response to header bidding which was originally created as a way to bypass Google’s “unfair” Ad Exchange policies. While this may be a step in the right direction for Google’s Ad Exchange, it will not eliminate the need for header bidding.
In conclusion, the creation of Meta-DSPs and the introduction of header bidding are clear indications that programmatic advertising is still evolving and trying to come to maturity.
Maintaining the human element throughout the programmatic process will ensure that the purchased programmatic advertising is always the right fit for each advertiser and being delivered as expected according to rules, targets, and audience matching.
With the ongoing introduction of new technologies, like Meta-DSPs and header bidding, as well as having more awareness and planning given to architecting and running a real programmatic campaign from the start, programmatic advertising can be quite successful and stay true to the original promise of programmatic.