The B-Side: A RUN blog

Well, 2015 is officially in the books, and it was a remarkable 12 months. It was RUN’s first full post-acquisition year, and the old adage of “what a difference a year makes” couldn’t ring more true.

As recently as September 2014, RUN was a team of around 30 people, working relentlessly to build a best-in-class ad technology platform that delivered differentiated value to the masses. We had a handful of clients for whom we did that, working out of our humble office space south of Penn Station, and were growing soundly.

Enter: October 2014. RUN is acquired, and suddenly responsible for servicing the 60,000+ person global Publicis Groupe network. We welcomed this change with open arms, and just like when many tall orders, challenges and opportunities have presented themselves before us in the past, we buckled down and got to work – prepared to do everything in our power to not only survive, but continue to thrive.

When we look back to January and examine everything that’s happened between then and now, the most obvious sign of success is our overall growth as a business. This isn’t just in terms of revenue and staff growth, however, although both were tremendous – it’s also when we look at some of our major wins with the world’s biggest advertisers; at the way we’ve scaled our operations globally in almost no time at all; and at the way we’ve diversified our already upper-class ad technology offerings by embarking on new strategic partnerships and building all-new products and features.

The other thing we find noteworthy – we’ve managed to do all of that without losing our sense of RUN culture and spirit, which we all value just as highly than anything.

Here are the highlights:

To all RUN partners – wishing you continued success in 2016, and we look forward to building our future together!

Without the ability to de-duplicate audiences across multiple screens, the ability to cross-screen target – is, well, useless.  It’s something that cross screen companies claim they do on the “back end” but none have been able to show it.  The need for understanding how reach and frequency across device drives performance and efficiencies is the foundation of successful, measurable cross screen targeting.

When RUN de-duplicates unique users across devices, we are getting the industry at large closer to standardizing the concept of "reach and frequency" beyond just TV.  Digital and mobile have typically been excluded from R/F analysis because of an inability to measure people across all environments, but that is no longer a concern.  As a result, cross screen media plans are more powerful.

We recently redesigned a whole tab in our portal to better express person-based R/F.  Check it out below.

Understanding the Graph (cohort analysis graph)

Reach & Frequency provides users with the ability to toggle between unique people and unique devices.  This will always default to ‘By People,’ meaning that uniques are de-duplicated across multiple devices.  If the user selects ‘By Devices’ then the unit of measurement for each unique is a single device.  

Making this selection will influence the data on the entire page including both the stat boxes in the upper right hand corner and the data in the cohort analysis graph below.  If the user looks at uniques as devices then the number of uniques will be higher with a lower frequency compared to uniques as people.  

Y-axis

Displays the number of uniques (either people or devices depending on selection).  It is very important to notice that this is a logarithmic scale.  Each label on the axis increases exponentially from the label that comes before it.  

X-axis

Each bar represents a cohort of unique people or devices that have been hit with impressions at a specific frequency.

Example

If the first bar goes up to 5 million on the Y-axis and the second bar goes up to 3 million on the X-axis then that means out of the total number of uniques, 5 million of them have seen an impression once and 3 million of them have seen an impression twice.

Use Case

Users can layer KPI trend lines on top of this in order to determine which levels of frequency gives the campaign the best performance. This will default to displaying the green CTR trend line, but others will be available for selection if applicable.  If a campaign has video placements then Completion Rate will be available, and if there are conversion pixels placed then Response Rate will also be available. The user can choose to select any and all of the KPIs to layer on top of the frequency cohorts

Export Data

The Road to The Final Four is Built on Data

Today kicks off one of the greatest sporting events of the year, March Madness. The tournament is able to capture the country’s attention over the next couple of weeks as almost everyone is involved in some sort of office bracket or pool. To witness friends, colleagues and parents fill out brackets is as much fun as the games themselves. Everyone has their own method to the well madness so to speak.

Teams are chosen by uniform colors, if you have had a good time visiting, lucky numbers and a host of hokey suspicions and beliefs. However, things operate a little differently here at RUN as we look at most problems as largely being a function of data and the tournament is no exception.

Methodology

To rank the teams for the tournament the overall body of work from the 351 division 1 teams. Each team plays roughly 30-35 games leading to about 5000 games to take into consideration. For example, Kentucky beat Florida 3 times so it is safe to say Kentucky should be ranked ahead of Florida. The goal is to rank teams that have not yet matched up against the other by collectively using the results of each individual game between similar opponents.

By constructing a matrix of dimensions 5000x351 (lets call this A), all of the results can be summarized nicely by assigning each game and team an index with the game outcomes as a 5000x1 vector (+1 for a win, -1 for a loss b) and setting this up as a system of equations of the form Ax = b. By using a technique called Singular Value Decomposition it is possible to solve a system of linear equations while minimizing the the sum of the residual squares. This is actually a relatively simple operation; (ATA)-1ATb minimizes this outcome vector.

Selecting the proper feature set for these games is a whole other challenge in itself. Visiting teams perform worse than home teams, the distance a team has traveled may also make a difference, whether certain players were injured during a crucial game this year, if a team is suspect to hot or cold spells during a game, the point margin of a game, and many others could possibly be considered as important to a model. One way to test this is to split the games into 2 subsets, one as training and one as a test set. If adding the additional feature improves the predictive ability in the test set, using that feature may be warranted in the model.

Picks

Without further ado, the rankings for the 64 remaining teams in the tournament is shown below. A value of 23.5 for Kentucky means they would be favored by at least 1.5 against every other team. As there is some variation in teams performance (a standard deviation of about 5 points per team is the norm), this model doesn't show that no team has a shot at all to beat Kentucky, but rather Kentucky has a little over a 60% chance of winning against Arizona at a purely neutral location for example, or 99+% chance of winning the opening round match up against Hampton.

Team Name (Points Rank)            

Kentucky (23.4853)

Arizona (21.9274)

Villanova (20.7868)

Wisconsin (20.4317)

Duke (19.0242)

Gonzaga (18.3545)

Virginia (17.9946)

Utah (17.4055)

North Carolina (16.3667)

Oklahoma (16.0151)

Kansas (15.3647)

Baylor (15.2212)

Iowa State (15.17)

Notre Dame (15.0928)

Ohio State (13.9861)

Michigan State (13.8107)

Louisville (12.9536)

Wichita State (12.8848)

Butler (12.8416)

Georgetown (12.6951)

Texas (12.6082)

West Virginia (12.5249)

Iowa (11.9778)

Xavier 11.6809

Providence (11.5781)

Davidson (11.4933)

Arkansas (11.4836)

Northern Iowa (10.8693)

Southern Methodist (10.7042)

Maryland (10.5401)

Virginia Commonwealth (10.1163)

North Carolina State (9.86253)

Oklahoma State (9.81573)

Stephen F. Austin (9.74219)

San Diego State (9.65263)

Georgia (9.58844)

Mississippi (9.47138)

Purdue (9.09989)

UCLA (9.09053)

Louisiana State (9.0651)

St. John's (NY) (8.94679)

Dayton (8.7051)

Indiana (8.67195)

Buffalo (8.61914)

Oregon (8.5047)

Cincinnati  (8.02717)

Georgia State (6.68439)

Valparaiso (6.45939)

Harvard (5.33481)

Wofford (4.26732)

New Mexico State (4.05838)

California-Irvine (3.3264)

Wyoming (2.90762)

Northeastern (2.29004)

Belmont (2.1971)

Eastern Washington (2.01769)

Albany, NY (2.007)

Alabama-Birmingham (0.988353)

Coastal Carolina (0.847573)

Robert Morris (-0.576433)

North Dakota State (-0.59635)

Lafayette (-1.42122)

Texas Southern (-2.44106)

Reaching potential brand consumers based on their geographic location – also known as “geofencing” or “geotargeting” audiences – has become a core competency of the mobile ad industry. Geolocation stands as arguably the most valuable targeting asset that mobile advertising has going for it – not too many other internet-enabled devices travel with every person to every single place they go – which makes up for the screen’s creative limitations and then some.

As we wrap 2014, advertisers are warming up to making mobile a larger part of overall media plans, and beyond that, seriously considering how mobile-specific targeting capabilities including geotargeting can impact their brand’s bottom-line. However, the story at digital media agencies remains more or less unchanged when it comes to geotargeting specifically. That story goes a little something like: pick a mobile geolocation vendor, pass along some zip codes or addresses and see how things go.

Usually, the measurement plan that comes along with a geotargeted mobile ad buy includes delivery of impressions, reach, and frequency, and some kind of engagement metric – CTR at the uninteresting end of the spectrum, all the way to sales lift or ROI at the most interesting end – compared to a more untargeted mobile tactic as a benchmark.

If an advertiser is working with a partner who prides themselves on performance and an attribution/ROI story as RUN does, this insight can be valuable, no doubt. But what if there was a way to make geolocation data valuable BEYOND just its first instance? What if storing geolocation data, anonymously and in a PII-compliant manner, could be made useful for future campaign objectives or tactics?

That is our goal with assigning behaviors to location.

Imagine if every lat/long point or “box” available for targeting also had corresponding businesses, hobbies or behaviors – ultimately, audiences – attached to it. With an advanced tech stack and DMP, it is possible to build unique segments of people, like “Gym Rats,” “Household Shoppers” or “Sports Nuts,” and open them up to advertisers. While brands are likely already identifying “types of people” with their standard geotargeting tactics – “Auto Intenders” who visit car dealerships, for example – it is the ability to reach those same Auto Intenders AFTER their visit to the dealership, all the while still KNOWING they were there in the last two weeks, that is an all new way of thinking about leveraging geo signals.

     It is with great pleasure that we announce our acquisition by Publicis Groupe, a purchase that signifies a colossal paradigm shift. Specifically, RUN will be aligned with Starcom MediaVest Group (SMG), an organization that specializes in human experience, investment, content creation and digital in 130 offices worldwide. We’ve always prided ourselves on our ability to create disruption and from those breaking points, push forward. This acquisition embodies that spirit, an ethos shared by both RUN and Publicis.      At RUN, we began pushing boundaries when we offered complete, real-time transparency to advertisers during a time when nobody else would or could. Today, we charge past definitions of “DSP,” “DMP,” and even “Platform” to become an enterprise advertising technology and an agency, and a direct extension of all the brands we represent.

"Agile technology products and expertise and the resources and multi-channel demands of a large agency holding company are no longer mutually exclusive."

     This acquisition reveals that agile technology products and expertise and the resources and multi-channel demands of a large agency holding company are no longer mutually exclusive. This is great news for marketers who have expressed a desire for a central resource for their audience data and activation needs, and it is also great for the digital ecosystem. We’ve gotten too comfortable with the dispersal of different IDs and because of this, the number of different partners who offer slightly varied versions of the same thing.

     Holding companies, by their very nature, work with everyone. Within that structure, RUN will be a home for Publicis’ brand data regardless of its origin. We will not only bridge devices, as we have continued to do with Device Connect™, but we will aim to knock out a few bricks from the “walled gardens” and start to create depth in understanding consumer behavior across phones, tablets, music players, connected TVs and soon, smart watches and cars

"Create depth in understanding consumer behavior across phones, tablets, music players, connected TVs and soon, smart watches and cars."

As the industry gets closer and closer to universal metrics (Nielsen's OCR, comScore vCE, Viewability, etc), media plans will start to blur the good 'ole fashion allocation "bucket." This is inevitable, as nobody cares if you're consuming content, and therefore a brand message, on a laptop, smartphone, tablet, CTV or whatever other video-enabled device is coming down the pike. There will always be media partner buckets (ie - publisher direct, vertical network, audience, FEP, short-form, etc), but again, the screen shouldn't matter. That said, it's challenging for vendors to decouple themselves from a core-competency as agencies still like to buy based on their roots.

The Rise of the Product Leaders While the role of Product Leaders has always been critically important in ad tech's rise to prominence, it is only recently that agencies and brands have really started supporting these roles as well.  Over the last two years or so, brands and agencies have hired their own product and tech savvy specialists that sit within their trading desks and planning teams.  Additionally, certain agencies are starting to build their own tech stack.  As all of this is happening many of these product leads on the agency-side do not want to speak with sellers, who have limited technical chops, as many of these agency product leads were on the sales side at some point, so they can see right through a salesperson trying to fake it.

Moving Forward Keeping the above in mind as we think about engaging Product Leaders in four specific ways:

It’s becoming increasingly important to understand the technical needs of a client upon meeting or speaking with them.  Understanding their challenges and providing a specific solution during meetings is pertinent when meeting with these newly minted product leaders. Saving them from some of the song and dance that sellers still approach the market with, they find extremely refreshing and typically engage in a much more meaningful way.

We have also noticed that agency and brand product leaders understand that the space is still evolving so it’s important to be even more forthcoming on methodologies that are being used.  Being able to show how these products/solutions are visualized in your own platform, along with showing how the people in your organization use these functions for campaigns is tremendously important.  Less storytelling, more showing.

Honesty is the best policy. If there is something that your system can’t yet do, be honest with them.  There is always something that comes up in a meeting that your system can’t provide but instead of saying yes, selling the deal and then running back to the engineering team and making them figure it out in a crunched time frame, let them see that you can say no too.

The biggest theme from the “Mobile, Mobile Everywhere” article by Allison Shiff is the lack of mobile measurement prohibiting increased spend in the space, especially among the larger brands. It’s something we agree WAS the biggest barrier from a brands’ perspective but we are finding that to no longer be the case.

After beta testing our Device Connect product and releasing it to the public in May we have seen a major uptick in spend from not only our current clients but also new clients, including many Fortune 500 brands. Device Connect facilitates real-time conversion tracking (post view & post click) across all mobile platforms - even iOS Safari while also solving some of the major challenges the mobile industry faces today.

Mobile Challenges

• 90% of all mobile conversion go unrecorded

• Conversion tracking in IOs

• Availability of Deterministic App to Mobile Attribution Systems

• Accuracy of status quo Industry Cross screen targeting platforms

• No 3rd Party ad servers can track mobile conversions in all environments

Device Connect Solution

Accomplished with the use of TRUE ID through our exclusive partnership w/ Verizon Precision Marketing Insights Group

• Measurement across all mobile devices and operating systems

• Uses a myriad of identifying data points most significantly True ID

• Ability to track conversions from in app to mobile web (100% accuracy)

• Cross Screen Measurement

• All Data points stored in our DMP allowing Brands to segment and target their loyal customer base now and in the future

• Visualized through a transparent platform in real time

Behaviorally targeting consumers based on demographic, online behavior or historical product purchases is a very common practice in today’s desktop digital marketing landscape. The ecosystem consists of a wide marketplace of data providers, which often specialize in different targeting capabilities, methodologies and even specific vertical strengths (B2B, Automotive, CPG, Retail, etc.). However, demographic targeting is perhaps the most important as advertisers are most concerned about reaching their target audience at scale. In no environment is this truer than in mobile. 

Currently, the biggest challenge in mobile is scale. Even the largest aggregated Behavioral Data exchanges have limited Mobile Device ID targeting data which only represents a fraction of the active smartphone and tablet devices in the marketplace. This poses a problem for mobile marketers looking to justify buying mobile display and video media while still being efficient and accurately hitting the target audience. Fortunately RUN is able to use two different methods to solve for this - Nielsen Mobile Online Campaign Ratings and Verizon Precision ID Demographics.

Mobile OCR is Nielsen’s migration of their desktop demographic campaign ratings system, which scores every impression delivered to a user. Utilizing Nielsen’s API with Facebook, desktop cookies are scored with corresponding Age and Gender designations. The output is Reach, Frequency and GRP levels achieved against the Adult 13+ Audience, as well as GRP levels achieved in each demo segment.

Mobile OCR takes this approach to the mobile ecosystem and allows a media buyer to buy impressions against only the target demographic segment they intended to reach, with confidence that the audience data was validated against Facebook data.

The second approach of accurate demographic audience targeting comes from another newly available deterministically derived dataset – the Verizon Precision ID. This straightforward approach leverages Verizon subscriber data, a proprietary carrier ID parameter, allowing a media buy to accurately target Age, Gender, Income and Language Preference segments. 

We are delighted to announce that we’ve integrated with LiveRamp, the best in class offline to online solutions provider in market.

This integration allows RUN to offer clients a personalized 1:1 experience with their existing customer base on desktop and mobile. Clients can now ignite online sales from customers who they know have made an offline purchase.

True to our mobile first approach, RUN is one of the first companies to work with LiveRamp to execute an offline to mobile solution. We now offer the ability to ingest CRM data via LiveRamp and target existing customers on mobile. This fits perfectly into our existing mobile product suite and is a great compliment to Device Connect which is a true deterministic mobile attribution solution.

We are thrilled to announce our newest product, Device Connect – the first of its kind deterministic mobile attribution solution. With this launch, we are not only solving for in-app to mobile web attribution, but also powering our platform with robust cross-screen measurement tools. We’ve been listening when our clients have told us what they need to bring confidence to their mobile advertising investments, and have developed a solution.

With our revolutionary new approach, we offer an absolute method of tracking mobile conversions, cross-device reach and frequency reporting, and the ability to target persistent mobile consumer profiles. We’ve worked closely and tirelessly with Verizon’s Precision Marketing Insights Group to bring Device Connect to its current evolution.

Prior to this announcement, we released Cookie Translator, an always-on technology that transforms desktop audiences into mobile audiences. This advanced system can identify a desktop user on his or her mobile device with 70-80% accuracy. But we know statistical models for user identification are only as good as their accuracy, so we couldn’t stop thinking about how to improve ours. We kept asking ourselves the following question:

“Wouldn't it be great to know with absolute certainty that the person you advertised to while they were playing a game on their phone is the same user who visited your mobile website an hour later – and the same user who made a purchase the next day on their laptop? How could we do that?”

________________________________________________________ From there, we defined two methods:

Access publisher login data (i.e Facebook), which only a handful of companies can activate at scale.

Think bigger, more high level – beyond the page and app. Enter: The Carrier-level ID.

_________________________________________________________ RUN developed a suite of mobile and cross-device advertising products utilizing Verizon's carrier-level ID. With this ID, we match desktop site visitors to the other devices they use. We can also track all mobile conversion events with complete certainty, regardless of the browser or operating system! 

This is monumental in the arena of mobile, still largely viewed as an upper-funnel channel because even the major third-party ad servers lack the technical chops to track events in any cookieless environment. With this new release, we continue pushing ourselves to deliver nothing but the most innovative and forward-thinking solutions. We will always be transparent about our methodologies and proud of our footprint because we have the technology to back it up.

Our “Always On” Cookie Translator was unveiled by our co-founder and COO, Dan Schwartz, at the 4A’s Transformation conference a little over a month ago. It was received with great fanfare and continues to be a distinguished point of differentiation for RUN in the commoditized marketplace of DSPs and Ad tech companies. To continue to build on this momentum we thought it was about time to dive a little deeper into not only how it works but how it’s visualized in RUN Portal.

Building User Profiles

Our goal with the Cookie Translator has always been to create a transformative solution to identify and build user profiles from desktop display campaigns. As each company’s dependence on cookies wanes and more advertising spend gets allocated into the mobile space, the ability to identify users across screens has never been more important.

Our Cookie Translator was developed with this in mind and it constantly translates desktop campaign cookies to mobile device ID’s while the campaign is in progress. This constant churn of mobile device ID’s is segmented and stored within RUN’s DMP. This allows advertisers to begin targeting their audience across mobile devices during the campaign or target their audience with a subsequent mobile message once the desktop campaign has ended.

Portal Visualization

It is important for us to not only capture the Cookie Translator at work but also make sure that it is transparently visualized in RUN Portal in real time. With that in mind we made sure that every campaign has its own tab that shows the percentage of the plan with the following:

• Percent of Cookies Translated into Mobile Device IDs • Number of Devices Available for Targeting Overall • Number of Devices Available for Targeting Per Placement • *Number of Devices Available for Targeting Per Web Page (i.e   Conversion Page)

*This is good way for businesses to upsell existing clients on their mobile phones and tablets.

These data points are also warehoused in our DMP and accessed whenever the advertiser needs them. As we target mobile IDs we report against them in real time through the RUN Analytics Portal. 

The New York City – Southern New York Chapter of the National MS Society today announced a new engagement with the RUN Fund, which includes a cause-related digital advertising initiative to support the Chapter’s Walk MS 2014 fundraising event. With the support of RUN, the leader in mobile-focused programmatic advertising technology, the RUN Fund is helping reach the Chapter’s potential donors through a robust pro-bono digital advertising campaign.

“With the generous support of the RUN Fund, we are breaking new ground to engage with our donors and raise awareness for Walk MS, our signature fundraising event,” said Robin Einbinder, chapter president. “I have no doubt that this digital advertising campaign will galvanize the community to help raise more funds for local programs and services for people affected by MS.”

RUN is proud to announce our integration with HasOffers, a leader in attribution analytics for performance advertising. 

Like HasOffers, RUN understands that better attribution technology creates the best value for advertisers. We strive to measure and manage advertising based on scalability, accurate attribution, and real-time reporting. Integrating with HasOffers was a top priority for RUN because we put a premium on seamlessly integrating our clients’ preferred attribution systems with our own analytics suite. In this case, we believe the whole is greater than the sum of its parts.

PlowShare, one of the nation’s leading distributors of public service advertising, has appointed RUN as its exclusive advertising technology partner. The leader in mobile-focused programmatic advertising technology will assist PlowShare in developing a new, state of the art trading desk – Harvest RTG – which will provide unparalleled capabilities in analyzing nonprofit donor data to its world-class partners.

"PlowShare’s extensive nonprofit client list – from the Girl Scouts and World Wildlife Fund to the American Red Cross and Wounded Warrior Project – will greatly benefit from this highly measurable approach," said Seth Hittman, CEO and Co-Founder of RUN. “With Harvest RTG, PlowShare will support major nonprofits in accessing extensive data to best understand and reach their target audiences. We admire the work PlowShare does, and we look forward to helping them further enhance that work through our strategic relationship.”

Harvest RTG will deploy RUN’s proprietary DMP to allow nonprofits to create deep and multifaceted audience databases for cross-screen retargeting. This will enable advertisers to reach likely donors at home, in-office or on the go with relevant messaging, at the right times. In addition, the nation’s largest nonprofits will have access to all of RUN’s global programmatic inventory.

“RUN’s unique proficiency aligns with PlowShare’s history of developing and distributing successful PSA campaigns for many of the nation's best-known nonprofits and government agencies, and we look forward to extending the relationship.”

"Our core values are anchored in integrity, creativity and accountability, which clearly match everything RUN stands for," said Jeff Boal, founder and president of PlowShare. “RUN’s unique proficiency aligns with PlowShare’s history of developing and distributing successful PSA campaigns for many of the nation's best-known nonprofits and government agencies, and we look forward to extending the relationship.”