Share agency perspective on BT based on most recent campaigns and experiences.Not a planner or buyer. Oversee a team of digital media analysts. Campaign reports and analyzing results for optimization.Talking about BT today from the perspective of those folks on my team.Objective thoughts on it’s performance within the broader media mix.We see that the advantages of BT and it's likely impact on a client's business can be obscured by current-state standard measurement. Peel back layers, or look at it from different angles, to get a truer understanding of it’s value.
First some definitions.I’m going to be talking about "Behavioral Targeting" in generalized way, which I’m sure is going to irritate many of the specialists in this room.But that kind of 10,000 foot view is necessary when evaluating the broad array of tactics used in today’s large scale campaigns.So, you should know that, for the purposes of this presentation…
Start with the notion of unpredictability.BT is a component of virtually all of our display advertising recommendations these days. But it does not dominate the media mix. While retargeting almost always performs well, more general BT results can vary vs. other tactics.Three campaigns: tech products, targeting fairly precise consumer segments, similar mid-funnel engagement objective. But very different results.As analysts, we sometimes have trouble identifying really clear patterns in the results.There seem to be so many complicated and fluid data quality variables underneath the hood of these buys, can be hard to predict results campaign to campaign.E.g. The recency of the behavioral data, the varying definitions of common audience segments (e.g. an auto purchase ‘intender’) across networks and data providers), the fuzzy delivery chain.
But if I had to point to the most typical performance scenario, it would be something like this.Relatively strong performance vs. endemic display tactics, but limited volume and spend. In this respect, optimization challenge can be similar to search.Performance challenge is about how to scale without sacrificing too much in terms of cost-per-conversion.
One proven way to scale is to expand the target definition. And now that we have the ability to look at the behavioral profiles of responders and converters from all types of media, it should be a relatively straight-forward exercise.Here’s a good example.Essentially discovered a whole new segment to go after.The most interesting thing about results like these is that they tend to lead to some of the most fascinating, most politically charged campaign performance discussions.Come from intense, expensive traditional research work streams. They’re shared by multiple disciplines, from web content to advertising to brand identity. Tons of momentum behind targets and the predefined ideas about what the data segments should be. So, even when you have pretty strong evidence that expanding the target should be expanded or modified, it’s often times not enough to escape all that organizational pressure and weight.
To be fair, the adherence to a tight audience definition often has very solid marketing rational. And this leads to the first way in which the advantages of BT are obscured in standard measurement.Many of our campaigns are based on driving content engagements (demo videos, trial downloads). And, more specifically, engagements with a specific audience. Engagements outside that audience are considered low value given the objectives of the campaign. But if the day-to-day performance reporting of engagement volume and efficiency is for all up engagements, not engagements by the target audience, the superior targeting of BT isn't reflected in results. Let’s look at an example.Moving to a 'targeted engagement' metric for tactic-level analysis is something we often hypothesize about, and would reveal the benefits of BT, but is hard to instrument effectively.DMPs could obviously step in here to reveal the advantages of BT. But gets into sticky privacy conversations fast with clients. And cookie deletion, etc. Watching this space because could put BT in context with rest of the plan in dramatic way.
Continuing with the theme of under-represented performance..No discussion of display advertising can escape the nerve-racking issue of attribution. There are folks on my team who suspect that – at least for certain types of products and conversions -- the effects of last click attribution punish BT more than other display tactics.Theory: The superior targeting improves organic/latent response, and the lack of contextual relevance suppresses immediate clicks. Here’s the type of stuff they’re reacting to…Time and time again, looking at network conversion counts from there piggy backed pixels vs. the conversion counts the ad server is attributing to them shows a pretty shocking discrepancy.And actual multi-touch analysis across the plan can prove it out.Giving BT the same conversion weighting as other standard display tactics may undervalue it's true performance.
Given the huge generalization I just made, I’d be remiss to not point out that different flavors of BT can have different attribution impacts on each other, and complicate the performance picture of the tactic.This is something my friends at Cadreon often remind me of.In general, we see that… Retargeting steals credit from straight-forward BT, and both steal credit from predictive targeting.Another good reason to always important to keep the user experience in mind.
Seems like almost all the conversations I have with folks in the BT world focus pretty narrowly on DR prospecting and acquisition. To a certain degree it makes sense.Something inherently sexy about ferreting people out who are at the bottom of the funnel in places we wouldn’t normally notice them.And we know that attitudinal results are contextually sensitive, so BT is at a disadvantage. But this narrow focus of BT is hard for me because so much of the activity we at UM manage is designed for the top of the funnel.And can make the argument that the zero waste element of BT gives is a solid place in brand campaigns.So it leaves an important question on the table… When does BT's superior targeting outweigh it's lack of contextual relevance in driving attitudinal results? Getting stable survey results at the level of granularity to prove BT's worth on these metrics can be hard, and without them, other tactics (like premium video and custom integrations) enjoy a perceptual advantage. But occasionally we’ll see some pretty interesting results like these.From a campaign last year in which Nielsen took the brand recall lift of control vs. exposed by publisher, applied target comp and media cost to it, and derived an efficiency metric for brand recall impact.Top three performers were BT networks.This is certainly not the case every time. Since you don’t know where you’re running, hard to diagnose whether there are different degrees of contextual relevance influencing the results. But results like this give the tactic a whole new interesting dimension. Clearly something going on around the economics of frequency and targeting.The frustrating thing is that getting reliable measurement options at this granularity is rare. Never ending headache for us.If you can start to answer this question in an actionable way for advertisers, gold mine waiting for you.Getting reliable measurement options at this level of granularity is a never ending headache for us.Hard to predict because you don't know where you're running. ----focus on efficient targeting and delivery rather than the effectiveness of an advertisement. Need immersiveness and emotionalimpact.The investment focus has been on ever finer-grained targeting of those who are already likely to buy, rather than “top of funnel” branding.
I’m big on lists, so thought I’d end with a couple.When it comes to BT, what are my teams top wishes and top worries.Rely on rich media to push out content experiences for engagement campaigns. I’d be nice if it could be tied into BT network optimization. Mentioned measurement before. There’s got to be better tools for us to optimize media based on important upper funnel brand building metrics.Would be fascinating to see audience segments become available across platforms; could trigger major innovations in tracking.Fuzzy chain. Makes it hard to fully understand why something is performing the way it is.Some goes for segment definitions. Perfect example is ‘automobile intender’. Everyone defines it a bit differently. Hard if your building a plan for Chrysler. For me, another thing that makes analysis that much harder.Sit next to broadcast buyers. Amazing the level of clarity and confidence they have as they go about negotiations. Also amazing the volume of deals they seem to execute each day. They know exactly what they’re buying. They don’t have to ask scores of questions just to understand a unit, placement, or segment.Finally, my boss reminded me that I’d be foolish to at least not mention privacy regulation. It’s the elephant in the room. Very worried about a potential premature and ill conceived end of cookies. For my world in particular it’s hard to fathom, as I’m sure it is for all of you as well.
Behavioral Targeting from the Trenches:Reflections of an Agency Analyst Dane Hulquist, SVP, Director of Ad Services Universal McCann San Francisco July 20, 2011 OMMA Behavioral, San Francisco
Definitions and Context For this presentation, "Behavioral Targeting" is… Buying custom audiences via ad exchanges using a variety of leading 3rd party data providers Buying pre-packaged audience segments directly with networks It does not include… Retargeting (unless noted) The cases and generalizations are from campaigns with a broad media mix and $100k+/month (digital).
Silver Bullet or Wild Card? Fluid data quality variables can make performance hard to predict. Consumer Tech Engagement Campaigns Oct 10 – Mar 11
Unlocking Brand Ad Budgets When does BT's superior targeting outweigh it's lack of contextual relevance in driving attitudinal results efficiently? 2010 Consumer Brand Favorability Campaign Brand Recall Efficiency (Nielsen IAG) (Size= Imps)
An Analyst’s BT Lists Wishes Network-side optimization based on in-unit interaction High-fidelity attitudinal measurement Cross-platform segments (and tracking) Worries Fuzzy ad delivery chain Lack of standard segment definitions Death of the cookie