Better Customer Targeting Through Greater Understanding

Linda Dupree, CEO of Nielsen Catalina Solutions, has seen it all, research-wise from, as she noted, “Agency, advertiser and then research supplier.” Dupree and Chief Research Officer, Leslie Wood, have worked together for a long time and together they are a formidable force in the field of sales data analytics. 

Now, as Nielsen Catalina is about to celebrate its tenth anniversary, they have doubled down with efforts to further improve the measurement of CPG purchasing through the consumer funnel.

Measuring CPG on a More Sophisticated Level

Dupree believes that, “In an environment that is increasingly focused on outcomes, NCS is the sweet spot in the CPG market.” While her company initially started in the traditional forms of audience measurement, “increasingly we do targeting, optimization, and eventually we will move more into predictive measurement (applications).”

For Wood, who has been with NCS for eight years, “the company started with overall measurement and targeting and recognizing from the first, the incredible capacity and value of the data.” She added that even years ago, there were multiple sources for every piece of data which made it very useful to help track the consumer intent and journey. And with some of those sources of data overlapping, it was also possible to verify the data veracity. Some of the data comes from panels, some from set top boxes, some from shopping and purchasing data, some from outside suppliers of data and some from Nielsen. “These layers give you the understanding of what the biases are and the things that you don’t do well,” she explained and added, “and that has all formed a canvas which gave us the ability to go out and create so much richer understanding into how advertising works.”

New Products and Services

A range of services were established around all of this research because, as Dupree explained, “the quality of the underlying data resulted in a richness of insights derived from the data.” Technology has also added new tools to examine the data in new ways. “Data guides everything we do,” explained Wood. 

 Some of the new tools released from NCS include Slm, the Sales Lift Metric, which is delivered weekly for a campaign, “so that changes can be made in-flight,” Wood noted. “It gives us the ability to go in and look at what is going on across different tactics all at the same time,” and compare tactics to see what is and what is not working and then re-allocate. The ability to examine campaign efficiency to potentially deploy dynamic ad insertion gives Slm added value to advertisers.

A Changing Marketplace

“The market is looking for faster and better,” Dupree explained. Her team is focusing on that mandate as well as “growth in CPG in a world where consumers have a choice. Twenty years ago in CPG, there used to be about 7,000 items in a grocery store and now, 20 years later, there are about 50,000. (This is because) the types of variety (of consumer products) have changed so much.” This has resulted in much more competition in the CPG space. Marketers need to be sure that their money is being spent wisely and efficiently and results in sales impact.

“There is also an abundance of channels in the media world,” Dupree continued, “the shift in OTT and Linear and Addressable and VOD as well as other traditional channels like magazines and radio, we are awash in choice and that has been important for us to understand and measure.”

What Resonates in CPG

There are some challenges for CPG marketers in this new media environment and some are more focused on price rather than efficiency. But to Wood, this is not a trend across the board. “A lot of the stories about spending less,” Wood began, “may not be from national advertisers. There is a lot of spending from tiny brands where the budgets are very low. When you look at budgets of over a million dollars, it is a different story.”

According to Wood, the creative matters a lot in a campaign. But, as she explained, “The creative is the starting point. We are also seeing that the importance of the media decisions has increased a lot over the last few years.”  And Dupree noted,” There are also so many more levers to pull.” Media has gained in importance especially in regard to targeting. An advertiser would want to target brand loyalists but should also consider targeting non-users of a certain brand as well as users of a competitive brand. This will not only expand the potential customer base, it will also help to build the brand over time. “When you talk about the brand promise,” Wood stated, “it speaks to the essence of the brand and what the brand stands for. The interplay of media with targeting creative and how those people come together has been some of the richest findings we have had.”

Personalization is another big consideration so that the brands can, “meet the consumer where they are,” concluded Dupree, “especially through ad versioning.” All in all, the confluence of data driven advertising within the CPG world is resulting in not only better targeting of consumers but also a greater understanding of their needs, wants and concerns.  For Dupree, “We feel that we are in more of a dialogue with consumers and it is a much more personalized experience,” and that should greatly improve efficiencies by reaching consumers at the right time, in the right place and with the right messaging.

This article first appeared in www.MediaVillage.com

NCS's Leslie Wood on Monitoring Campaigns Using AI Causality

High on the wish list of most marketers is improving the measurement of campaign effectiveness. Fortunately, Leslie Wood, NCS's chief research officer, has been on a mission to perfect it. She has accomplished this with a combination of machine learning methodologies that interface with a campaign in flight, thus enabling marketers to modify and improve their ad placements in real time. This initiative, called Sales Lift Metrics, sheds light on the causal impact of specific campaign tactics by identifying the key sales drivers that amplify incremental sales for CPG brands.

"We have created a Super Learner, which includes multiple different machine learning methodologies, like 'random forest' and 'gradient boost,'" Wood said. "And when we execute the Super Learner, it runs all of those different models and sees which ones are winners — often up to three or four of them — and combines them into an ensemble model." It is apparently flexible enough to apply to any type of CPG advertiser and is constantly refining its application through the use of AI.

How It All Began
The idea and the efforts behind this initiative started about five years ago and, over that span of time, the methodology has evolved through testing and innovation. "When we first started, it was just television because you can't have a control if you have such high reach. After all, who is not exposed? People who were not exposed to the advertising look very different than TV viewers. They are very different kinds of people," she explained.

Wood discovered that this approach took the pressure off the researcher to constantly oversee the model. "Here was this machine learning method that allowed us to put this in places where data was siloed. We could set it up in 'free rooms,' places where there was no researcher," she said, adding that this allowed a door to open where Google, Facebook, and Apple could manage their proprietary datasets without a human interacting or overseeing it. "It is meant for computers to talk to computers."

In-Flight Optimization
Five years ago, the hardware just wasn't there to be able to apply in-flight optimization. But now, this can be executed on what Wood calls, "sales-lift metrics," which provides the ability to "use these machine learnings to quickly, in real time, deliver key performance indicators to tell you what about your campaign is working." It is now possible to compare multiple creative options, consumer targets, or sets of tactics for sales effectiveness and incremental causality.
"We are looking at the incremental value of each of them," she said, "and compare them" in a weekly cumulative sales report. An advertiser can make changes to the campaign as it rolls out and see the impact.

What Sets the Super Learner Apart?
"There is always a full set of statistics" within the Super Learner, Wood noted. This is unusual because "AI almost never has a full suite of statistics. Statistics were developed by professors on small data sets. For big data sets, we develop machine learning and AI methods and those were built by computer programmers. So, we don't usually have confidence intervals, significance tests, or standard error — those measures that you need to say that this is a good model. We built Super Learner, which does all of these steps."

In this way, as new machine learning methods are developed (and, according to Wood, they are being developed all the time), they are added to the system to see if there are new ways to improve campaign efficiency. "It allows us to continually improve without reprogramming or recoding," she said, adding that certain models never worked and others that worked too hard and delivered unreasonable answers. "We removed those."

Among the successful campaign indicators is incremental sales per household, which is a key performance metric. It also reports impressions delivery by target and by creative.

Going Forward
Since the Super Learner has the ability to integrate faster systems and new models, it advances every day. Thousands of new schedules have been run and rerun "to see what works and what doesn't work, what's the minimum threshold, what's feasible, what are indicators inside the statistics that let us know that it's not working."

Wood is committed to keeping up with "this fast-moving world" of data and machine learning with constant monitoring and asking why results change. In the fall, they will be adding more data, more brands, and a price/promotion feature.

And, according to Wood, this entire initiative has been a model for others. "We use the h2o machine learning language; they will tell you that we were the first to create a Super Learner and they followed suit and built one themselves," she said. "We have pushed the envelope on machine learning."

 This article first appeared in www.MediaVillage.com

Nielsen Catalina's Leslie Wood: An Erwin Ephron Demystification Award Winner

If Leslie Wood had taken the same professional path as her parents, she would have been an artist -- a musician like her father or a dancer like her mother.  Instead, Wood is an artist of data and analytics.  She is currently the Chief Research Officer of Nielsen Catalina where she has established a strong reputation in branding and connecting consumer viewing behavior with purchasing activity.  Her out-of-the-box conceptual thinking has led to industry recognition as an ARF award recipient.  Wood (pictured above) is the first female to receive the prestigious Erwin Ephron Demystification Award, which was first established by the Advertising Research Foundation six years ago.

At the recent ARF conference, Wood and I discussed her current work, major initiatives and her "ah ha" moments.  "We've spent the last year and a half looking at how to build brands," she explained.  "A huge study with 50 brands and every measure we could find for a three-and-a-half-year period."  It is such a rich data set that it "will continue to bear fruit" going forward, she said.  Wood is also looking at long-term effects and the right way to measure them.

To do that accurately, "you need to go back at least one year," she noted.  "What we found, though, is that most people aren't willing to measure something over a year old.  'Oh, that campaign?  I'm done with that.  I want to know about this one.'  So, the focus here is, can we estimate the incremental?  It's all about making sure it is right and real."

Nielsen Catalina recently announced that they are measuring incremental sales with Google, building it in a walled garden and "using methods that have not been broadly used before because most tried and true measures require a researcher to be monitoring and you can't do that in a walled garden," Wood said.

Her team has also brought many new industry-wide products to market recently, including in-flight optimization, which is a new measurement methodology that she calls next-gen.  "It allows us to track campaigns in real time.  There are all kinds of products being developed out of that."
What led to the Erwin Ephron Demystification Award was Wood's brand work, which has pushed the limits of measurement and incorporated a range of different data sets.  "There was a core 'ah-ha' [moment]," she explained.  "We had all of this data, including advertising spend, which gave us not only the spend but also the share of voice.  We had from RMS price and promotion and features and store-level sales.  Then from our frequent shopper data, we had all of the measures of household level data, like repeat rates and purchase cycles, loyalty levels and things that require people-level data.  We have advertiser promotion and sales and put them together to look at. and included for every data set any measure that was standardly available."

But how is it possible to integrate all of these disparate data sets?  "When we looked at how those measures were related, they were all in different time frames," Wood admitted.  "Some penetration is past year -- sometimes the past six months like the purchase cycle.  Everything was a different measure.  Some were short-term and some were long and some right now."  To overcome that dilemma, Wood compiled every measure she had into two buckets -- four weeks and one year.  "Dollars is easy," she added.  "How many dollars in four weeks versus dollars in a year and as a rolling four weeks."

Then she did some correlations.  What "popped off the page" was that penetration has an 84% correlation with yearly sales.  "That is why the focus on penetration has been so strong," she noted.  But when she moved it to four weeks, the correlation rose further to 97% over the fifty brands.  What does that mean?  "That it is much more important to be thinking about getting buyers now, rather than only focusing on the people who didn't buy in the last year," she concluded.

This resulted in a new theory: Get your buyers now.  The more you can get now of any kind of consumer, the better off you will be.  "Advertising does best among current buyers," she explained.  "Maybe it's a shorter purchase cycle or maybe you have increased the repeat rate or increased loyalty. Use advertising for what it is really good at.  We know from our work that if you build loyalty and build repeat, you get the long term.  This drives both the short and long term." 

"Erwin was brilliant at picking some single point to focus on.  I picked one measure that I thought, 'Wow, if we really understood this piece it would give us a whole new way of thinking about how to do advertising," Wood said of the inspiration she carries with her from winning the Erwin Ephron Award.  Clearly, Wood's creative and novel way of analyzing data will continue to lead to award-worthy insights and game-changing applications.

This article first appeared in www.MediaVillage.com

Attribution Accelerator - What to Expect?

The annual Attribution Accelerator conference is gearing up for record attendance next week. What can we expect? What have been the advancements over the past twelve months? Sequent Partners who are the event organizers created the Attribution Accelerator three years ago because they realized that the industry lacked a forum that focused solely on attribution. Attribution has become a much bigger issue for cross-platform measurement. There is still a lack of transparency and, as of yet, no industry standard approach. 

“Our event helps the industry get smarter about attribution, by shedding light and providing guidance, as well as encouraging best practices and development in the necessary areas,” stated Jim Spaeth, Partner, Sequent Partners.

This year, the conference will discuss how attribution is impacting activation and helping marketers manage ROI rather than measure it.  Panels will address how brands are represented in attribution, technical innovations (such as AI), creating a unified model with marketing mix models and how attribution is impacting television optimization. 

Panels to Watch

The ARF is heading a panel called, “Innovation - What’s Next and Will We Love It?,” which will explore what they have learned on the innovation frontier and how the latest developments can drive greater business performance, with a look into the future. According to Scott McDonald, President and CEO, The ARF, “Some of the topics we will discuss include: how AI can be used effectively to solve attribution problems, how precision targeting and unified modeling can improve, and the move from the lab to broad implementation in the real world. We will also examine whether to build or buy when it comes to innovation and how to build an appetite for risky innovation in a risk-averse corporate culture.”

Leslie Wood, Chief Research Officer, Nielsen Catalina Solutions, explained that her panel will cover a recent Nielsen study titled, How to Build Brands which “is our most ambitious research project to date, generating 6 billion rows of data for 50 CPG brands. One of the key learnings is that there is often a  disconnect between the intent of the creative message and the resulting purchase behavior of the audience. Advertising to consumers who don’t respond, wastes your media dollars. Advertising to the right consumers drives sales and growth.”

The importance of attribution cannot be underestimated. David Ernst, Vice President, Advanced Television and Digital Analytics, A+E, noted, “Most advertisers know that TV is a powerful, brand-building medium.  But now, through the use of large data sets and advanced analytics, we can demonstrate its true effectiveness by having a more precise understanding of how and where it works best.”  In his session, he discusses how attribution has the potential to change the dynamics of the television marketplace and why more clients are looking to attribution to get specifics about their goals, what they’re trying to accomplish, and make sure that campaigns deliver on their promise. He will also share his perspective on where attribution is headed going into the future and how they will approach the marketplace in exciting new ways.

“Attribution is continuing to gain traction in the marketing community, with new applications available to demonstrate the productivity of Television. It’s use is evolving from providing insights to delivering very specific analyses regarding the effectiveness of Networks, Dayparts and even programs. This is enabling us to have conversations with clients about productivity and expectations for individual campaigns on our networks,” he concluded.

 “Media companies like CBS are ready, willing and able to partner with their clients on measuring outcomes,” says Radha Subramanyam, Chief Research and Analytics Officer, CBS.  “At the same time, we are in a time of transition when it comes to attribution.  While there are lots of options today, TV attribution still remains an emerging and evolving field with no clear leader to date.  The good news is that early efforts by multiple companies all prove the same thing.  TV advertising works incredibly well.”

This article first appeared in www.MediaVillage.com

Finding Solutions at the Attribution Accelerator Conference

The perfect attribution model, in my opinion, is a chimera. But that doesn’t mean that I don’t hold out hope for an industry multi-touch attribution model that, even if not perfect, is considered pretty good. This past week, executives from data and analytics companies to advertisers and content providers met to exchange pretty good ideas at the Attribution Accelerator Conference in NYC.

The conference strove to present attribution solutions for marketing measurement.  “Attribution is the hot spot,” stated Alice Sylvester, Partner, Sequent Partners, who added, “The development of a unified approach is taking place. We are now focusing on data and data quality and have a more practical, more analytical orientation this year.” But the path to a generally accepted attribution model may be further off than we think because of a fear of change in the industry which has relied on Mixed Media Models for decades and the walled gardens of data that have increasingly siloed and segmented data applicability.

Here are some takeaways from the Attribution Accelerator conference:

Challenges to Attribution

There are many reasons why it has taken so long for attribution to start to gain traction in the industry according to John Leeman, former CEO of Fresh Direct. Companies fear choosing the right dataset, upending the current business model or changing the decision set of marketers which results in less creative control and second guessing of their media choices.  And, once having instituted an attribution model, what if it’s wrong? “It doesn't have to be perfectly accurate to be valuable,” he stated.

But moving towards a prove-able attribution model is pivotal in this new media world of ever increasing consumer touchpoints that make the journey to purchase a meandering minefield. “Get over your fear- trust your gut, imagine, realize and just do it,” he concluded.

Finding Solutions for Advertisers

Many companies are tackling multi touch attribution head-on, striving for a real time model that might even incorporate media mix modeling, drive minute by minute allocation decisions and increase advertising agility. But there are many, many considerations that may vary by brand and by company.

Charlie Hilton, AVP Advanced Analytics for AT&T, noted that her company has been, “doing attribution for nine years. We are now shifting gears to true attribution and near real time optimization.” Her attribution methodology needs to take into account how all of the different media tactics interact with each other, how to quantify the precise combination of different touch points along the way and if advertising for one product impacts other products.

Kristina Kaganer, Director of Global Market Strategy, Coty, has product lines that span from discount to high end.  “Fighting for shelf space at Walmart involves different competition than when you are competing at Sephora,” she noted. Scalable attribution is her company’s goal and that requires the collection and control of pertinent data, technology that functions on the local level (which can differ depending on locality), the standardization of processes that can cross local markets, data storage and analytics. 

Finding Solutions for Media Companies

There are five keys to driving sales, according to David Poltrack, CBS Corporation Chief Research Officer and President, CBS Vision and they are Reach, Targeting, Recency, Creative and Context. “Few TV campaigns are tapping the full potential of TV,” he asserted. From untested creative, a disregard to context, a cost efficiency rather than a reach orientation and little attention paid to recency, advertisers are not maximizing the true value of television in a buy. “Digital is not going to replace TV,” he stated. “Advertisers and marketers must focus on getting TV right. TV is a reach medium and a combination of TV and digital is important. Understanding how to get that combination right is what attribution is all about,” he added.

Summarizing the results of a study with CBS, Leslie Wood, Chief Research Officer, Nielsen Catalina Solutions noted that the fundamentals for advertising are as follows: Creative is the most Important factor, although media is playing an increasingly larger role. Reach is critical because only consumers who are reached can respond. Targeting has to focus on buyers of the product or service and Recency is vital because the value of exposure decays over time.

Finding Solutions for Measurement and Data Companies

Nielsen has always owned its own data. But in a world of first party data and big datasets, Nielsen is becoming more flexible and philosophical in how they acquire data and partner with other companies in the space. Matthew Krepsik, Global Head of Analytics for Nielsen, spoke of his company’s journey working across walled gardens. “We have the algorithm. We are good at the math. But it is not about building a better algorithm. The most important thing we can do now is to seek out better data.”

It is not a simple journey because, he noted, “every touch point can drive conversion. Shoppers are increasingly changing their behavior. They are becoming omni-channel digital natives and engaging brands directly.” Different products have very different purchasing cycles which make them multi-dimensional, multi-faceted and difficult for attribution. Nielsen is striving to tackle the de-duplication of devices so as to better analyze ROI, data hygiene and privacy compliance so as to insure an attribution model that better allocates marketing dollars

Attribution “is an elusive quest,” he concluded.  So too for us all.

This article first appeared in www.MediaVillage.com