Why Doesn’t Digital Advertising Work Better?

There is no more important and existential question in marketing than “Did my digital ad campaign work?”  Digital advertising holds the promise of eliminating waste and increasing Return on Investment (ROI) by targeting the best prospects at the right time.  With the level of data available to marketers and AdTech firms, why are campaign results so inconsistent?  It turns out that many buyers of digital advertising, are not getting what they think.  A common axiom in AdTech has been that media was differentiated and that data/targeting were commodities.  Hence you see all types of premium media packages, placement, and offerings. However, as you will learn in this post, media is often the commodity, and targeting/data is most likely to be the difference between success and failure.  This holds true across all the common types of digital media including display, over-the-top (OTT), social, native, programmatic audio, and digital video ads.

Here is a quick primer on how ads are targeted.  Many organizations buy an audience from data companies like Acxiom, Experian, or Comscore.  Then brands or agencies use data onboarders like LiveRamp, Oracle, or Adobe to create digital audiences in a Demand Side Platform (DSP).  Finally, the DSP serves digital ads to this audience via their platform.  Results are then commonly measured using tools like Google Analytics, Adobe Analytics, or HubSpot.  The problem is between the data company, the onboarders, and the DSP, things can go terribly wrong.  The remainder of this blog describes some of the most egregious examples.

Often Data is Simply Wrong

Stacy, our CEO, recently requested his data from Oracle to see how accurate they were with the behavioral advertising segments he was in.  As you can see from the notes below, a simple review of this data reveals massive inconsistencies.  In fact, the data was so wildly inaccurate Stacy had no problem sharing it in this blog post.  Keep in mind these segments are often curated and then provided by third parties like PlaceIQ, ComScore, Epsilon, or Ground Truth in addition to audience segments developed by Oracle.

First, let’s look at income, which is an important targeting variable in most digital advertising campaigns.  As noted in the screenshot below, according to InfoGroup, Stacy is listed in 5 different income brackets (anywhere from under $35K to $499K).  Obviously, he only has one income.

Net worth is another common tool used to curate digital audiences.  According to Oracle, Stacy’s net worth ranges from 0 to $27MM+, with him being listed in 4 different categories.

Additionally, InfoGroup has Stacy listed in every age group from 18 to 65+.

PlaceIQ, another one of Oracle's vendors, has Stacy listed in every income group as well, plus they have him listed as both male and female.

Additionally, PlaceIQ was equally wrong on Stacy’s age, listing him in 5 different categories.

This goes on and on with much of the data that brands would use for digital targeting being dead wrong about Stacy.  If you would like to see your online audience data from Oracle, you can request it using the following link.   

A look at Oracle’s audience segments – shared with a third-party DSP partner – is also confounding.  For example, consider Oracle’s Automotive Ownership Segment.  According to audience counts pulled in October of 2023 from one of El Toro’s DSP partners, Oracle has an audience of 383,700,000 targetable people in the US who own a Ford Fiesta and the same number who own a Ford Flex (see screenshot below).  For some perspective, Ford only sold about 296,000 Ford Flex vehicles (ever), which means under optimal conditions only 1/1200 of the people in this segment own a Ford Flex.  Ford only sold about 700,000 Fiestas in the US from 1978 to 2021, meaning if all of them are still on the road (which is unlikely) 1/548 of these digital prospects would be accurate.

How About Other Data Providers?

You may be telling yourself: I don’t use Oracle; my AdTech partner is more accurate.  Well… to take this a step further, we reviewed the audience counts for about 500,000 3rd party segments that are available via one of our DSP partners.  And it turns out that many other data providers have similar issues.

The following examples are derived from the same DSP audience counts for US targetable devices from October 2023:   

Data company Bridge has identified 268MM new parents and 265MM parents-to-be. There are only 258MM adults in the US and 3.6MM babies born in the US annually.  Meaning that under optimal conditions this data is only 2% accurate.

AdTech company Semcasting has 880MM devices associated with people in their 18+ targeting segments. This is more than 3X the number of adults in the US.

ShareThis, with data onboarded via Acxiom, has identified an audience of 365MM people in the United States who may be interested in purchasing a Ferrari, but Ferrari only sells about 5,000 vehicles in the US each year.

Healthcare data provider iMedical has identified 534K US neurologists that can be targeted with ads; however, there are only 3,050 practicing neurologists in the US. 

Digital audience provider All Alike has identified devices for over 1.1 BILLION people in the US who like punk music, and another one of their online audiences contains devices for over 1.1 BILLION people in the US who like Rugby.  Put another way, according to this data, 400% of the adult US population like Punk and Rugby. 

Mobile location analytics company PlaceIQ has an audience of ½ billion devices in the US for people who are likely to spend more than $1,000 on a cruise, but the US only had 12 million cruise passengers last year.

We could go on and on, but you get the point.  With data like this, much of the money brands and agencies spend on digital advertising is simply wasted on people who will never buy your product, come to your location, or vote for you.

So Why Does This Happen?

The Marketing/AdTech ecosystem is needlessly complex (see the LumaScape below).  With the average brand leveraging 6+ companies to target, place, and measure digital ads, there is much room for confusion and little accountability.  This complexity also causes issues with data integrity.  El Toro provides audience onboarding for several leading data providers, and when we compared their known audience counts with the counts once onboarded inside a DSP via other data onboarders, we noted in some cases the counts from the DSP were exponentially larger.  Note: if you are an audience provider and would like to increase your accuracy and results, we should talk.

Most importantly, brands are measuring the wrong things when it comes to the performance of digital advertising campaigns.  Most digital ad campaigns are judged by metrics like clicks, click-through rate (CTR), viewability, CVR for form fills… almost anything but the one metric that matters.  The seminal metric for digital advertising campaigns at El Toro is now and always has been Return on Ad Spend (ROAS).  By conducting a Pre/Post Net of Control (PPNOC) analysis of who bought either as measured by data from a CRM/offline dataset or via a footfall analysis, we focus on efficacy and profitability for our clients. 

How Can Brands Protect Themselves?

First, make sure your advertising agency is focusing on the metrics that matter.  Second, ask good questions about the technologies and tools you are using.  One of the most frustrating parts of trying to optimize performance for many marketers is that you simply don’t know who you are targeting.  That is why El Toro provides visibility and accountability in targeting so that brands can have more control over results.  Finally, stop focusing on cost and focus on value.  One of the biggest mistakes agencies, brands, and marketers make is thinking that data, targeting, and digital advertising are commodities and then measuring them based on which provider has the lowest price. However, value can be best defined as quality divided by cost.  Quality in this instance equates to driving more sales, visits, votes, or whatever you are trying to influence with digital advertising.  ‘

As a good example of real-world value, all we have to do is look back at our Ford example. The Ford Fiesta was cheap; in 2019, its final year of production in the US, the MSRP was $14,260, making it one of the least expensive cars on the market.  Yet, Ford still only sold 60,148 of them in 2019, because consumers saw greater value in more expensive models like the Explorer which cost twice as much but provided much higher value.  AdTech is the same way, often the cheapest option will provide the least value. 

If you would like to get superior measurable results from your digital advertising campaigns and eliminate wasteful spending, come run with the Bull.