Consumers’ shopping behaviors and habits have rapidly evolved, and the change has accelerated in recent years. The retail paradigm is evolving too, and with it ecommerce and relevant media platforms and strategies. In episode 9 of Xaxis’ “People of Programmatic” podcast, Brittny Schoeneman, Director of U.S. Programmatic Commerce Solutions at Xaxis, interviews Michael Schuh, VP for Media Strategy at Kroger Precision Marketing, about how a large retail company is responding to and leading the trends.

In this episode:

• Brittny Schoeneman, Director of U.S. Programmatic Commerce Solutions, Xaxis (HOST)
• Michael Schuh, VP for Media Strategy, Kroger Precision Marketing

We will cover:

• What’s changed in consumer behavior, including results of a new research report on changes in shopping behaviors and habits, accelerated by the Covid-19 pandemic.  
• How retail and advertising platforms are merging into a holistic retail media approach.
• How first-party data from retail media companies is affecting media strategies and leading to more effective outcomes.
• How search is changing, from search engines to retail sites.
• The future for retail media.

What’s changed?

Schoeneman describes the research report, delivered in two parts by the London Research Group in partnership with the Xaxis and Catalyst agencies, which was based on interviews with U.S. consumers in March 2022. The first part describes changes in shopping behaviors and habits influenced by COVID-19. The second describes how brands are responding to these changes, barriers they face and further actions they need to take. 

Research Point: 71% of U.S. consumers are shopping more online than they did before the pandemic and 68% are using more shopping apps and sites than ever before. Advances such as click and collect (buying online and picking up in-store), and contactless checkout are accelerating these consumer shifts and fueling more online shopping.

Schoeneman asks Schuh for his take on changes in consumer behavior, particularly as it affects the retail grocery market. (Kroger is the largest supermarket chain by revenue in the U.S. with annual sales of more than $121.1 billion, according to Fortune magazine.) Schuh describes a survey of Kroger customers that found 91% of households will do as much or more online shopping this year than they did last year. In response, Kroger is expanding its grocery pickup and delivery service and is opening automated fulfillment centers across the country.

Schoeneman and Schuh agree that, in addition to online shopping, there has been huge growth in hybrid shopping: customers are placing large eCommerce orders every week or two, then doing fill-in trips in-between to get fresh produce, fresh meat or something they might have forgotten in the online order. Schuh describes how most trips, whether in store or online, are influenced by digital advertising in some way, and so it’s important for brands to be present online, because the “digital aisle”— the shopping experience online — impacts purchases in-store.

“It’s no longer exclusively in-store, it’s not even exclusively online, it’s this really hybrid shopping experience that is incredibly digitally influenced.”

– Michael Schuh

Retail media: A merger of retail and advertising

Next Schoeneman and Schuh discuss how the changes are affecting Kroger’s organization and how they are working with retail media. Schuh sees “retail and advertising kind of coming together and changing each other for the better in so many ways.” As a result, retail media companies must have full-funnel media solutions and make advertising more consumer-centric, he says. In the grocery space — where customers need to feed their families, pack lunches and get inspired by new recipes — an advertiser can use branded content to address those needs, he adds.

He also describes the role of retail media in helping advertisers reduce wasted spend. For example, if a consumer buys an item that has a long purchase cycle, an advertiser doesn’t need to serve ads to that person for the next couple of weeks. Retail media companies can use first-person data to direct advertisers to redeploy spend to areas that will drive net new engagements and conversions. 

Research points: 

• More than half (58%) of companies say they invest in retail media to build awareness or extend their reach. Only two out of five (41%) say they do so to drive conversions.
• 78% of respondents agree retail media is strategically important for their company, while almost 73% have changed their operational setup to maximize the retail media opportunity.

“How do we use branded content to get in front of the right consumers that are going to inspire them, help them discover new things and ultimately drive better performance for the advertiser?”

– Schuh

Search as a single-platform journey

Schoeneman describes how the role of search engines is changing; they are no longer the default starting point for consumer journeys. Many consumers instead start searches on eCommerce or retail sites, and the shopping experience online can then be a single-platform journey.

Research Points: 

• 51% of US consumers start their shopping on eCommerce or retail sites. 
• Single-platform journeys are answering the needs of brands, retailers and shoppers. For Kroger, 13% of customers are using the platform for inspiration and discovery, 10% for research and comparison, 14% for actual purchase, and 11% for repeat purchases. 

Schuh adds that the consumer journey is no longer linear but rather a series of short sprints, with the ability to add to an online cart from your phone while walking down the street, or from your laptop while reading a recipe. As a result there are many unique moments when brands need to be present, he says.

He also describes a difference between grocery and “high consideration categories” such as electronics or fashion, which usually involve more research. On Kroger’s website, he says, 85% of searches are unbranded, so it is critical for brands to be present, to drive not only conversion but also long-term value.

“The really important takeaway is having that brand presence and making those products available on that digital shelf for the immediate opportunity.”

–  Brittny Schoeneman

Retail media and first-party data

Brittny and Schuh then shift gears to talk more specifically about retail media and how it is changing. 

Research points:

• Retail media can be defined as the advertising that appears within a retailer’s properties, or that uses retailer-owned data to target prospects and customers outside of those properties
• The retail media market, which was worth $31.5 billion in 2021, is forecast to reach $50 billion in 2023. Retail media (including social media) is significantly ahead of all other channels in terms of marketing budget growth. Both channels have seen over a quarter of marketers increasing their spend by more than 10%.

Schuh describes two things that have made retailers as critically important as media companies for brands, agencies and advertisers of any kind:

• Consented first-party data based on real behavior and real signals.
• Meaningful measurement both in-flight and on the back end, matching every ad exposure to online and in store sales. 

These advantages help reduce wasted spend and redeploy funds to really impactful moments and lead to important business outcomes: not CPMs or numbers of impressions, but return on ad spend (ROAS), sales, increases in lift and household penetration. Retailers have the ability to keep their customers at the center of the decisions on how brands direct their spending to make each moment most impactful for them. 

Schoeneman describes how, from an agency perspective, it's great to be able to tie media investments and campaign performance to real sales and business outcomes as Schuh described and to get a real-time feedback loop and measurement that's tied to activity at a granular level. 

“If it’s not good for the customer, it’s not going to be good for the advertiser.”

– Schuh

Looking ahead

Schoeneman asks what Schuh sees as the next major evolution of retail media. Schuh names three areas retailers should concentrate on:

• The customer: The opportunity to continue to make the customer experience better, more efficient and convenient for customers to do online shopping through personalized recommendations while promoting inspiration and discovery.  
• What brands and advertisers need: Making it easier for brands to leverage the power of retail media’s data assets, to help them reduce wasted spend and measure what matters.  
• The media ecosystem: Advances in the ecosystem that facilitate increased performance related to retail media’s first-party data. There’s also a need to create the right standards in the industry. How, for example, to better define good performance, and help advertisers invest, meet consumer needs and drive really impactful programs for their businesses?

Schoeneman wraps up the interview with a perspective from the agency side, how they are excited to see how offerings like Kroger’s continue to develop and make things easier for advertisers and clients that they support. And with so many new capabilities, partners and datasets, she will be looking for developments that help them figure out how to prioritize and where to focus. She is optimistic about the future of retail media.

Retailers are quickly becoming media companies.

– Schuh

Learn more, reach out & follow up:

• “The Retail Paradigm: The Evolution of Consumer Shopping” (research report Part 1)
• "The Retail Paradigm:How brands are investing in the new era of shopping" (research report Part 2)
• Kroger Precision Marketing
• Xaxis on Twitter

With the digital media industry facing more complexity and uncertainty than ever, brands are showing rising interest in outcome-based strategies that guarantee business results from their marketing investments. In episode 7 of our “People of Programmatic” podcast, part of our 10x10 series to herald Xaxis’ tenth anniversary, we're going behind the scenes of outcome media to help you better understand its nuances, realities, and possibilities.

Listen to this episode on: Apple Podcasts |  Spotify | Google Podcasts

In this episode:

• Henry Stokes, Managing Director, Global Accounts, Xaxis (Host)
• Michael Siewert, Global Programmatic Marketing Director, Colgate-Palmolive
• Emily Knight, Senior Director, Business Development, The Trade Desk

We will cover:

• The ways outcome media is defined.
• How brand marketers and their agencies can determine desired outcomes.
• The challenges that outcome media can help solve.
• The benefits of technology in using outcome media.
• How brands can embrace outcome media.

“Outcome based media to me is about being specific and accountable to an outcome, whether that is a media metric, or whether that's a sales figure, and having the data to do it is really what makes it different.”

- Michael Siewert

What is outcome media?

Outcome media is fundamentally about data, and about using that data across the spectrum of brand marketing and performance marketing, which have traditionally been seen as representing two sides of a divide. 

• Outcome media brings a new level of sophisticated thinking to marketing.
• It allows marketers to get a clear sense of what they can expect from their media campaigns.
• It also allows marketers to measure efficacy in real time, to be accountable to a defined outcome.
• It fulfills the needs brands have to quantify results and justify their ad spend.
• It can connect media across channels, for example connected TV, shoppable media, and performance marketing.

“It comes down [to] a level of sophistication in our marketing that makes us specific to what we expect our media to deliver.”

- Siewert

The strategy depends on the brand

Outcomes should always be derived from the goals of the brand’s business, but the exact strategy depends on the brand.

• There can be upper-funnel targets, such as generating awareness, for which the metrics require a proxy to measure them.
• Further down the funnel, there may be metrics the marketer can tie directly to sales or other direct actions such as store visits.
• Determining outcomes and how to measure toward them is a case-by-case conversation, and often requires a bespoke approach to fit the brand’s needs.
• The category influences the strategy. What’s possible in consumer products goods is very different from what can be done in auto, retail, or ecommerce, for example.

“It comes down to data and what the brands have available and what their actual business objective is.”

- Emily Knight

Outcome media helps marketers focus

With all the metrics and channels they have available, marketers can, Siewert notes, “go down a rabbit hole” of different measures that don’t tell an overall story of whether a campaign is actually “winning” for the business. Our guests suggest that:

• Marketers need to think longer term, come up with a mix of metrics that prove with data and testing over time to have the best effects.
• They should focus on the goals, not the minutiae they measure to help reach those goals.
• It’s imperative to incorporate as much first-party data as possible.
• Marketers must test, test, and test toward the goal. That can range from something like performance oriented metrics to more custom solutions such as interactions with a purpose-built API.

“There are always going to be different sorts of layers of activation. In large brand campaigns, we're always going to be doing awareness [and also] we're always going to be doing search and lower funnel type things.”

- Siewert

Technology brings huge benefits

The technologies available to marketers facilitate an outcome-based approach, whether the focus is awareness, an increase in brand recognition, or more directly tied to direct actions.

• Platforms can support media buyers to help create campaigns designed to deliver on specified outcomes from the start.
• They can be used to customize throughout campaign objectives and KPIs, even to tie back to specific metrics such as sales, foot traffic, and so on.
• Media traders can work with what Knight calls “campaign wizards” who use AI to find recommended audiences that deliver on desired outcomes.

“Manage expectations and be very strategic about what you can get with a given budget for a single campaign, and across specific channels. That's where we see things be most successful.”

- Knight

Embracing outcomes can inspire transformation

Brands need to work across groups to incorporate expertise and resources across organizations at a holistic and granular level. Knight and Siewert offer the following tips to marketers:

• Define clear objectives that can align everyone involved.
• Work closely with the analytics groups and incorporate research that illustrates key metrics to which media should optimize.
• Be willing to fail and learn from that failure. 
• There needs to be a focus on what's impactful, not just what’s easy. Don’t let outcome media be a crutch that supplants sophisticated thinking.

“It really starts with organizational transformation, and then supplementing in tools that help us go faster.”

- Siewert

Outcome-based media will, our guests concur, be even more important as cookies start to diminish and first-party data become more important. Marketers can use the data that is available along with the metrics they’ve established to set hypotheses, test, analyze and continually improve to achieve their brands’ objectives.

Some Links:

• Xaxis, The outcome media company
• Colgate-Palmolive
• The Trade Desk
• Transforming media into outcomes (Xaxis)
• Is outcome-based media the future of media planning? (AdAge)
• Why outcome-based marketing is so successful (MarketingProfs)

Episode Host:

• Henry Stokes | Managing Director of Global Accounts, Xaxis
  • Henry has 19 years of experience in the digital advertising and search marketing arena and more than 10 years working in programmatic media. He is currently Managing Director of Global Accounts for Xaxis where he leads a team delivering programmatic strategy for some of the world’s largest advertisers. His career began at MediaVest and then Mindshare, where he ran the IBM and Nike digital accounts. In 2005 he set up Circus Street, a consultancy & training business working for the likes of Nestle, ITV, Guardian, and Sky. He returned to Mindshare in 2010 as Head of Digital before moving to GroupM to lead client development for Xaxis UK in 2011. After two years of rapid growth he decided to take on the challenge to do the same for Xaxis in Asia where the business was successfully rolled out across 14 markets and saw 10x growth in 5 years. Training and education remains a passion for Henry. He has delivered the digital marketing course for MsC students at the the reputable CASS Business School in London and also developed the Xaxis programmatic 101 training program in APAC.

Episode Guests:

• Emily Knight | Business Development Director, The Trade Desk
  • Emily Knight is Senior Director of Business Development at The Trade Desk, the largest independent omnichannel demand side platform (DSP). In her role, Emily is responsible for working across a wide category of advertisers and their media agencies, including some of the largest brands in the world, including Colgate-Palmolive. She's held the role for over four years at the company, while serving prior roles in programmatic at LiveIntent and Mindshare. A graduate of the University of Vermont, Emily lives in the greater New York City area with her family, and outside of work you can most often find Emily chasing after her daughter.
• Michael Siewert | Global Programmatic Marketing Director, Colgate Palmolive
  • Michael is a digital media industry native with over 14 years of experience in planning & strategy, e-commerce, and capability building. He has worked with top agencies in the service of brands like Unilever and General Mills as well as on the publisher side at Vice Media, in e-commerce at Chewy.com, and with brands like Colgate-Palmolive. Currently, Michael serves as the Global Programmatic Director for Colgate where he is focused on further accelerating the adoption of programmatically traded media across the company and its ability to deliver outsized performance improvements.

Every digital media company is struggling to recruit and retain great talent. In episode 6 of our “People of Programmatic” podcast, part of our 10x10 series to herald Xaxis’ tenth anniversary, our guests explore the new landscape of work and workers, and reveal how recruiters, managers, and executives can evolve their practices to find great people and keep them on board.

Listen to this episode on: Apple Podcasts |  Spotify | Google Podcasts

In this episode, we discuss:

• The talent shortage and how to manage it.
• How skilled workers’ expectations have changed.
• Whether companies are to blame.
• Solutions including culture, mindset, and rewards structures.

“The talent shortage is real. It's not just in media. … There are millions more jobs available than there are actually bodies to fill them.”

- Juli Santiago

The world of work has changed

Early last year, as the pandemic took hold, companies changed their hiring and work practices. 

• They stopped hiring, especially at the entry level.
• They held off on promotions. 

Today, they're feeling the effects. There isn’t a junior talent pool to promote. And overwhelmed mid-level managers feel increasingly strained without the assistance they need.

Companies “didn't hire at the entry level. And then there was also a lack of promotions at that lower level. So you kind of created this ripple effect that really just caused some big pain.”

- Juli Santiago

Meanwhile, everyone developed work-from-home (WFH) practices. Some employees left urban areas and many don’t want to come back to centralized offices.

• According to PwC, workers in metropolitan areas (66%) are more likely to work in roles that could allow remote working than those who live in rural areas (44%).

There is talent out there

Applicants still far outnumber available jobs, so companies have to adapt.

• 68% of the media workforce is willing to talk to a recruiter — twice the pre-pandemic level.
• There is a “window shopper mentality.” People are willing to explore.

“If I reach out to [a candidate] they'll at least speak to me, hear what I have to say…. If what I have to say is compelling enough [they] will be very willing to make a move now.”

- Juli Santiago

Are companies the problem?

The problem of finding and keeping talent may be with the companies themselves. They need to recognize that, as we emerge from more than a year under Covid restrictions and WFH structures, workers’ expectations have changed.

• Give top talent flexibility, the ability to work from somewhere other than an office.
• Expect counter-offers, sometimes large ones, which are now common.
• Condense the time between initial contact and hiring.
• Recruit outside your immediate geographic regions.
• Also recruit outside the usual parameters.

“There's no shortage of talented people from tier-one companies, major banks, consulting firms, etc. They’ve got great skill sets that we could easily adapt, and the fault is completely with us, as an industry.”

- Michael Wright

• 51% of job-seekers would forego higher salaries in favor of flexible work conditions.
• 49% of job-seekers turned down an offer due to a bad recruiting experience, and 56% said it would cause them to discourage others from applying.

Companies must meet workers' expectations

Talented workers will take jobs that offer the things they're looking for — hybrid work models, flexibility, PTO, career advancement, supporting staff, and, above all, a sense of purpose.

• Onboarding is key. Build connections. Welcome people, fast. Even email is too slow.
• Shorten the promotion timeline.
• Expand the pool from which you’re recruiting and promoting to new industries and wider skills.
• Lay out core principles, including your philosophy on hybrid working.
• Solve for whatever is wrong in someone’s current work situation.

“What is the pain? What can they not fix by staying at their current situation, and then we need to help them fix that pain in a new opportunity.”

- Juli Santiago

• 65% of employees are looking for a new job.

Tackle retention: Be aware and proactive

People have spent the past year evaluating what they want and what they’re getting from work. Companies need to address those concerns and make sure to show they’re putting employees' lives at the forefront.

• Be mindful of burnout. Encourage time away from screens.
• “Ruthlessly question” whether meetings are necessary.
• Offer quick growth, training, and promotions.
• Give people a sense of purpose, the larger goals they’re helping accomplish.

“You need to give people sight of the bigger picture. … What is my work for? Is it valued? What's its purpose?”

- Michael Wright

• 72% of job-seekers said they need to understand the work culture before accepting an offer.

Stay flexible. It’s going to keep changing.

There’s a long way to go, maybe 12 more months before things “level off,” Santiago says.  Not everything will be solved, but things are getting better.  “We’re recruiting candidates, seeing them at a more rapid pace,” she adds. “Companies are starting to figure it out.”

Our guests agreed that there are exciting times ahead, and ways to apply everything that’s been learned over the past year to capitalize on the new trends. And, of course, there’s more learning and growth ahead.

Some Links:

• Xaxis Careers Website
• SearchMax, Inc Website
• “1 million more job openings than people looking for work” (CNBC)
• “U.S. job openings hit record high as employers struggle to find workers” (Reuters)
• “What’s next for America’s workforce post-COVID-19?” (PwC)
• “Hot ad tech job market creating new hiring practices” (AdAge)

Episode Host:

• Arshan Saha, CEO, Xaxis APAC

Episode Guests:

• Michael Wright, Global Head of Talent Acquisition, GroupM
• Juli Santiago, Recruiter, SearchMax Inc

This is part of 10X10, our year-long series covering the Top Ten Trends in digital media. This month's topic: Creative.

Content, context, and creativity

In a cluttered digital landscape in which cookie-driven targeting is receding, the resonance of ads depends on their conception, content, and construction.

Data-driven marketing strategies and AI-powered programmatic delivery can do a lot to get the right content in front of the right person at the right time. Also, that contextual targeting is becoming more essential than ever. But will the content that arrives then catch the eye, stay with the viewer, even get them to act?

With user engagement becoming an essential factor in effective marketing, many marketers are accelerating their investments in dynamic new formats that help capture consumers’ attention, engage them, and inspire them to action.

Here are some of the most exciting and effective new formats to deliver resonant creative messaging.

Conversational advertising

Great campaigns are informed by deep analysis of the data, built around strong creative concepts, delivered in the right contextual environments, and experienced through formats that encourage deeper engagement. The right applications in the right mix can’t help but improve marketers’ results.

Conversational advertising provides valuable engagement during a two-way exchange of useful information that helps brands meet consumers’ desires while guiding them through their customer journey.

Last mile delivery ads

Fast delivery services have been rising in prominence along with the rise of e-commerce. Now, platforms that deliver a product or service within an hour are experimenting with ads that drive consumers directly to last-mile fulfillment partners.

Unique benefits of last mile delivery ads

Last mile delivery ads can delight users with nearly instant gratification. The ability to provide targeted offers, personalize delivery options, and immediate fulfillment can serve as major boosts to ad effectiveness. They can be dynamically optimized based on the user's location and product inventory to ensure relevancy. And since marketers’ messages are delivered in a wallet environment, they can increase the purchase levels of relevant products, making them a valuable e-commerce tool.

What to know about last mile delivery ads

As users get just what they want in moments, marketers gather great data. Because the data come from purchase actions in a clear and controlled environment, it’s fully declared and not probabilistic. The data can then be mixed with data based on previous behaviors and demographic details for greatly enhanced results.

Last mile delivery application technology is still maturing and being tested in many markets around the world. Availability is highly dependent on local conditions. As we learn the technology and help it stabilize and scale, marketers will want to be judicious in using it, to learn from tests and make sure to gather data that will be of use in future campaigns.

Retail media (including last-mile delivery) 'is really an opportunity to build a relationship with the shopper, to get closer to them … where they’re in the shopper mindset, with their credit card in hand.'

David Haase, chief revenue and development officer, CitrusAd
The People of Programmatic Podcast Episode 4: A Deep Dive into Retail Media

To celebrate this milestone, we decided to take a journey through our history, examining the very beginnings of Xaxis, how we evolved over the years, why we've been able to endure in such a dynamic industry for so long, and what the last decade of Xaxis has meant to our worldwide family of Xaxians, past and present.

We invited a few of those Xaxians to share their thoughts and memories on video.  We ended up with a powerhouse line-up of current superstars, veteran teammates, former executives, current leadership, founding members, industry partners, and more.

What makes Xaxis unique?

Our people.

Thanks to all the people who appeared in the video, to all our clients and partners, and to all the Xaxians all over the world for helping make Xaxis what it is today.

We invite you to hear from them in the video above. When you're done, give us a shout out on social with the hashtag #HappyBirthdayXaxis!

Our 10x10 series explores a new essential topic in digital media each month. This month, we dive into one of the most exciting and important elements of programmatic media: Artificial Intelligence. 

We’ve asked the people behind our proprietary AI technology, Copilot, to share their behind-the-scenes insights about data visualization, data science, academic research, analytical insights, and more.

So far this month, Tobias Sutters, a Copilot Senior Product Manager, shared a behind-the-scenes story into the team’s creation of a new AI insights package and Katherine Mello, a Copilot Software Engineer, showed us how her team integrated the principles of color theory into a new approach to data visualization.

Now, in a special audio and video episode of The People of Programmatic Podcast, we hear from five Copilot crew members about the academic research, review, and writing that goes on behind the scenes as part of their ongoing efforts to stay on the leading edge of AI technology in digital media. 

The discussion also covers the team’s experiences in data science, data visualization, and organizing journal clubs, as well as which modern examples of AI in action have inspired them.

To date, the Copilot team has published three journals:

1. Dynamic Bidding Strategies with Multivariate Feedback Control for Multiple Goals in Display Advertising by Michael Tashman, Jiayi Xie, John Hoffman, Lee Winikor, Rouzbeh Gerami
2. Online and Scalable Model Selection with Multi-Armed Bandit by Jiayi Xie, Michael Tashman, John Hoffman, Lee Winikor, Rouzbeh Gerami
3. Techniques Toward Optimizing Viewability in RTB Ad Campaigns Using Reinforcement Learning by Michael Tashman, Jiayi Xie, John Hoffman, Lee Winikor, Rouzbeh Gerami, Atefeh Morsali, and Fengdan Ye

And they have one patent pending titled “Double Blind Learning Insight Interface Apparatus, Methods and Systems” which describes the unique new way that the Copilot Insights platform delivers real-time updates and insights through a user-friendly interface guided by machine learning.

Featured in this episode:

• Jiayi Xie, Senior Data Scientist
• Rouzbeh Gerami, Head of Data Science
• Sambhav Jain, Senior Designer, Data Visualization
• Katherine Mello, Software Engineer, Visualizations
• HOST - Kelly Mergenthaler, Senior Product Marketing Manager

Listen to episode five of The People of Programmatic Podcast, “The Academics Behind AI,” below:

Watch the video version of episode five of The People of Programmatic Podcast, “The Academics Behind AI,” here:

A lot of people believe that the Copilot team spends its days deeply entrenched in data and algorithms. It’s true, we do. But that doesn’t just mean we’re mired in spreadsheets and equations all day. Actually, we invest a lot of time into brainstorming, experimenting, and exploring creative new ways to explain the power of AI.
Testing and learning is in our DNA. Every mistake is embraced as an opportunity for improvement. Our commitment to constant learning is very much reflected in our ever-changing color palette, which has evolved over the last year from a handful of default Tableau colors to a custom palette built around the needs of our users.
A lot of thought goes into making sure we use the right colors to communicate the story of our AI — it’s much more than just a marketing team adding a new hex code to the brand deck. Here’s the story of how the Copilot color palette evolved to its current state, and what we learned along the way.

Where We Started 

Before we had a dedicated data visualization team, we still had visualization needs, of course (like any team that has its hands on data and wants to use it to make decisions).  

At that time we were mostly building dashboards in Tableau for our internal needs, such as monitoring or testing features, but we also had some simple visualizations in our UI to show delivery and performance across the campaign.

Most of our visualizations were time-series based; lots of bar charts, line graphs, and area graphs — the bread and butter of data viz. For the visualization colors we primarily relied on Tableau defaults or, when we had the option, the color palette of our parent brand, Xaxis. Since our dashboards were for our own internal trader use, we put more emphasis on simplicity than aesthetics.

New Color Opportunities  

As Copilot’s technology became more sophisticated and complex, so too did our visualization needs. We realized we needed a data visualization team that could fully focus on translating these complexities into visual formats. This new data visualization team would have two main goals:

1. Translate our clients' data to actionable insights. 
2. Visualize our algorithms so that both our traders and clients could understand what happens “behind the scenes.” 

One of the first big projects we tackled was a visualization of a complex clustering model that grows over time. The account teams had gotten pretty good at describing the model, but we knew that words could only go so far. Adding color and shape to the explanation was the best way to really show what’s happening.

Similarly, we built an A/B Insights feature to visually compare two flights and see how they differ in performance and delivery.
The greater complexity of these new visualizations brought with them some new color challenges. We needed additional colors to represent:

• Differences between two different flights ("A" vs "B"). 
• Scenarios where there is no data, i.e. data vs. no data colors.
• A range of performance values from “good” to “bad.” 

Let’s focus on that last point for now — encoding performance. To understand the challenges of visualizing that data, look at this color-coded cost per click (CPC) performance map of the US.

Our First Stab  

With the marketing team’s color palette as a starting point, we decided on a continuous green → yellow → red scale for encoding performance.

The strongest aspect of this color scale is that it is familiar. This type of green-is-good, red-is bad, and yellow-and-orange-are-somewhere-between encoding is used in many places, from traffic lights to the stock market.

But the major drawback to this choice is that it is not accessible to everyone; this visualization could be hard to read for individuals with certain types of vision deficiencies.

Ultimately, we decided to build a new color palette from the ground up. But we weren’t going to approach it arbitrarily or base our output solely on aesthetics — we needed to dig into the fundamentals of how color is used to communicate data and determine how to best use those principles to serve our needs and our clients’ needs.

Back to the Drawing Board: Find the Right TYPE of Color Scale  

One of the first steps in landing on the right colors is deciding what kind of scale should be used. Color scales used in data visualizations typically fall into one of three buckets — divergent, sequential, or qualitative. You can read more here, but here’s a quick overview of how the three scales are used:

1. Qualitative scales use a range of distinct colors to represent unique variables that have no inherent order. As an example, think of a pie chart that shows people’s favorite ice cream flavors. Mint Chip and Chocolate will have unique colors on the chart but have no specific relation to each other.
2.  Sequential scales typically have a single hue which varies in saturation and/or luminosity. These are used when the values for each color are inherently ordered or are numeric, such as a population density map where darker hues represent highly populated regions and lighter hues represent the sparser areas.

3. Diverging scales are used only when there is a meaningful central value in your range of data, like an average or mean. Typically, that center value would have a neutral color, and the values to one side (representing data points below that point) would share one hue, while the values to the other side (data points above the central value) would have another distinct hue, to help distinguish both ends of the spectrum. These are useful for showing how a range of results compares against a central value — for example, a bar graph that displays customer satisfaction ratings relative to a specific benchmark. 

In our case, we can immediately throw out qualitative because we’re dealing with ordered, numerical data. That leaves us with either sequential or diverging.

If we were to use a sequential scale, only one “end” of our data values would be highlighted (e.g. if we made it so that values with a higher CPC were the darkest shade, then those DMAs would stand out, but not the DMAs with the lowest CPC). For our purposes, we care about both the high and low values — the good and bad CPC results — so a sequential scale isn’t exactly right.

In most cases, we tend to have a “middle value” that is important to us, which all other data should be compared to (such as a mean CPC value). So a diverging scale makes the most sense for our visualization.

Consider Colorblindness  

On the Copilot team, we believe in making advertising better for everyone. That means everyone. 

Red-yellow-green color scales are not accessible to many people with color vision deficiencies (CVD). According to colorblindawareness.org, “CVD affects approximately 1 in 12 men (8%) and 1 in 200 women in the world,” with red-green being the most common. 

During our color research, we came across a tool called Colorblindly, an extension that simulates the experience of colorblindness for developers so they can build websites that work better for people with color vision deficiencies.

The two maps below show how the red-to-green CPA visualization would look for users with trichromacy (“normal” color vision), followed by a simulation of how it might look to users with CVD. 

It was clear we needed to update the hues at each end of the scale. 

We leaned into an app called ColorBrewer to make our map palette more accessible and ended up with a diverging color scale with pink and green at each end.

 
Finally, we’d picked the perfect palette. 

NOT! 

Adjust as Needed 

As I mentioned, building a color scale has been a process.

While we were successful in highlighting good and bad performing objects with our pink and green color palette, we neglected to take the visualization context into account. We noticed that objects with values closer to the mean are hard to distinguish because of the low contrast between the light colors and our white UI background. Furthermore, at the point where the color changes from pink to green, it appeared as if there was a big change in value, even when the colors were very close to the mean.

To address these issues, we decided to:  

1. Add a neutral ninth grey color in the center to make the value transition smoother. 
2. Reduce the overall brightness of the palette to make the neutral colors more visible.

We changed our divergent scale from:  
To: 

 
Before  
 
After  

The last step will be to build out a set of data visualization style guidelines which would include best practices on accessibility, diverging vs sequential scales, and more. Documenting our standards and learnings will help us build more consistency across our visualizations, whether they live in the app, on Tableau, or within a slide deck.

Our users are giving us very positive feedback on these new visualizations so far, but we’re never done trying to improve it. As Copilot continues to integrate more and more data and data-driven insights into the platform, we’ll continue to refresh our approach to representing all of that information through clear, compelling, appealing visualizations.

Copilot Insights: The Art and Science Behind Our AI Visualizations

The digital media and advertising industry has witnessed an explosion of innovation that has embedded artificial intelligence across a suite of buying and measurement technologies.

While these developments have driven undeniable value for brands, we still get many questions about “what the AI is actually doing” or “why the technology made certain decisions.” Deconstructing machine learning and big data remains a formidable challenge in an industry where millions of buying decisions are made each second.

Recently, the Copilot team invested in bridging this divide to better answer “What is Copilot doing?” and other frequently asked questions for our clients.

Our Vision

In October of 2020 the Copilot team embarked on a journey of reimagining insights through the lens of machine learning and data visualization. It was important for our team to find innovative solutions to showcase the value of the Copilot decisions being made behind the scenes. With a team of Data Scientists, Engineers, and visualization experts, we looked at the common threads from which different algorithmic components derive decisions. Our goal was to compartmentalize the various ways campaigns are optimized and visualize them in artistic and relatable representations.

Our Challenge

The challenge was to communicate how an ad campaign performs and delivers and explain how our technology, Copilot, adds incremental value with machine learning optimization. Adding to the challenge, the project took place during COVID-19 and required collaboration across various teams during an extended period of remote working.

Our Method

To design Insights, we first connected with Copilot users across many countries to identify and organize pertinent campaign information and analytics. Through this process we discovered three key categories of questions that surface when running any Copilot campaign:

- What did Copilot do?
- How did Copilot perform?
- What can I learn about my audience?

Next, our product and visualization teams organized “sketch parties” to engage a diverse group of experts and crowdsource hand-drawn ideas. Our own Katherine Mello walks through the processes and thinking behind this virtual brainstorm here.

Each sketch party presented a specific design challenge to the team and allowed them five minutes to sketch their ideas. After short presentations and discussions, those ideas were collated, evaluated, and merged into low fidelity mocks for further input from our users.

Copilot Sketch Party, examples from Data Science and Marketing

Through ongoing user interviews in Asia, Latin America, Europe, and North America, we refined those low fidelity mocks into high fidelity solutions. During this process we consolidated concepts such as the color scale that communicates goal performance on every slide and the widgets that actively translate data into human digestible language. Our aim was to make visual processing of the package more streamlined and understandable.

Low Fidelity Mocks, bringing ideas and concepts to life

The Result

Copilot Insights was launched in March, 2021. It provides an automated, in-depth analysis of every Copilot campaign. The package is available across all major DSPs to compare and contrast campaign performance in a unified view.

The Copilot Insight package contains ten bespoke views split across three sections:

• How a campaign performed.
• The optimization that took place.
• The characteristics of a campaign that drive performance.

Table of Contents, Copilot Insights

Machine learning is embedded in our analytics to predict potential uplift generated by individual optimization components. These models actively deconstruct Copilot optimization to estimate uplift on a given goal type. Uplift models are embedded in Copilot Insight visualizations and highlighted across the package.

Estimated Uplift in CPCV, Copilot Activation

After reviewing insights available in the industry, we concluded that similar analyses often lacked simple human interpretability. To improve the experience, we embedded widgets that translate data into simplified language and offer key takeaways in every view. The additional takeaways succinctly communicate salient campaign information and make it easier for our clients to digest and understand campaign management.

Example Widgets, Sites & Applications

It was important for our design specialists to provide analyses in divergent, engaging, and aesthetic formats. By pulling together various analytics streams, we aimed to simplify the complexity of applying AI to media buying and audience targeting.

Copilot Targeting, analysis of features on Copilot campaigns