Omni Channel From Brand and Agency Viewpoints takes you behind the curtain of the digital provider world. The audience of the Digital Analytics Association’s Chicago Symposium was focused on omni-channel from the point of view of how its moving parts functioned because members buy and sell media and marketing content. Brand and agency digital professionals are caught in the tidal wave of data, which is straining legacy processes and relationships to the limit.
However, “Attribution” stole the show from omni-channel—and, for a fascinating reason. The same capabilities that enable big data give ecommerce vendors the ability field solutions that “attribute” the value of each media asset to the customer purchase. Hence, attribution is a massive accounting exercise, but it is disruptive to the digital media ecosystem because it enables, in theory, far more inclusive and granular counting of digital content’s impact on ecommerce or mcommerce or even in-store purchase. This is bringing accountability to digital and advertising firms. Just think of all the media that customers see before they purchase something. Agencies and vendors are using cheap computing to try to account for each asset’s “contribution.”
I have shared the bullet points of my notes of each session before my Analysis and Conclusions.
Omni-Channel and Analytics
Keynote Address: Analytics and the Consumer Journey
Andrew Swinand, CEO, Cardinal Path
- ADA survey: 58% clients use in-house agencies, and 52% want to bring digital in-house.
- They want expertise, speed and lower cost.
- Digital agencies are failing to serve, to add value by simplifying digital for clients (brands).
- Agencies are stuck in “containers of the past”: silos for creative, media, digital, social, research and mobile, and clients want a single source.
- Agencies have to integrate data, to specialize content more and to provide more efficient distribution.
- Meanwhile, disruptors become more relevant, look at Google, which has an integrated marketing stack: YouTube, DoubleClick for Advertisers, Google Analytics, Google+, AdWords, Android. Clients can subscribe and go direct to the stack. Amazon and Adobe are others. They offer one source of the truth.
- The IBM CMO Survey revealed that only 30% respondents use data to make decisions; 70% of CMOs are unprepared to make data-driven decisions.
- The CriticalPath Online Analytics Maturity Model has six parts: Management, Objectives, Scope, Expertise, Process and Tools, and best practice is maturing all competencies roughly equally.
- Brands and agencies overspend on (technology) Tools and underinvest in Expertise (people).
- Attribution is overemphasized and boils the ocean far too often. Clean data are needed. Expertise and Process are required to activate [act, get results].
- Dark data is a problem; agencies hold that their point of difference is owning and controlling data, but open data will create more value; like the containers of the past, this is a failed strategy. Instead, deploy an integrated marketing stack. Then integrate with with the BI stack (business intelligence).
- Users only use 16% of Google Analytics’ features on average; they don’t know how. Clients don’t know what they don’t know.
- Although data must be integrated, you still need specialists and a few generalists to quarterback projects.
- Most spend is still offline!
Panel: Around the Horn on Analytics Hot Topics
Moderator: Brett Mowry, Founder, Second Act Marketing; Panelists: Ryan Wilson, Digital Sales Leader, Bizo; Matt Miller, SVP, Strategy & Analytics, Performics; Jonathan Zajicek, EVP, Chief Analytics Officer, Havas; Jennifer Faraci Colton, Digitas
- Tectonic shift in retail, omni-channel and mobile; mobile is pervasive; analytics are driving omni-channel.
- Social media is not an ecommerce channel per se, but it’s closely related; people communicate and get product information; it’s more a loyalty engine, firms are too top-line focused. Firms can hyper-segment messages and build affinity in the middle of the funnel.
- Data needs to be integrated, but you need a universal ID; yes, integration of data, but maintain the ability to study discrete data as well. Data will be mostly integrated within three years.
- As we integrate data, though, the environment is becoming more complex, so it’s harder to make simple.
- The best ways to sell analytics/digital projects are to have a business focus, show ROI and handhold as needed. Find advocates on the client side.
- Big data scares clients, but most know they need it (so, an opportunity). It’s much easier to get executive sponsorship.
Multi-Channel Attribution: Valuing the Customer Journey
Part I: Attribution Introduction
Amit Kadam, Director, Strategy & Analytics at Starcom MediaVest Group
- Starcom MediaVest Group’s attribution system assigns credit to all touchpoints [on the conversion path], so we know what drives results.
- Most agencies and clients focus on the last touch before conversion, but that’s subpar. Our conversion model has several touchpoints: Introduction, Assist1, Assist2, Assistx, Convertor and Conversion. The current state gives all credit to the convertor.
- To implement an attribution system, you need to start with strong data, and data that only focuses on what’s important.
- Modeling is key; ask yourself, how does each touchpoint relevant to conversion? To activate properly, people need to be training to act [interpret/use data].
Casey Carey, CMO, Adometry
- Data accuracy is an oxymoron; it’s very hard; data is the foundation; you need trust, which is based on clean data.
- You need to balance data complexity, business goals and the granularity of KPIs.
- Consider channels. What really matters? There are a ton of metrics/KPIs.
- Too many people look at media cost, which overlooks the fine points. They also think social is free.
- The platform you use will determine what kind of measurements you can use; they vary widely in their capabilities, and they are evolving fast.
- When you define attributes, do it carefully. Also how you do ad IDs.
- Some clients segment their funnels [only use KPIs for a part of the buying journey].
- Sources of data are mostly websites and mobile. They can include call center and POS data.
- Best practice is linking to client’s CRM system for more data on customers.
- Cost is hard because it’s very messy to assign all the people and process costs of each touchpoint; the data sits in platforms, spreadsheets at agencies, vendors, clients.
- Consider using gross margin, not revenue.
- Start with the end in mind.
Madan Bharadwaj, VP of Product Management, Visual IQ
- Madan laid out Visual IQ’s attribution process and tools. His main point is that design is critical; you will paint yourself into a corner if you don’t do it right.
- Tracking is about defining data categories, events and stimuli.
- Data integration; create a taxonomy of data types, structures; ask yourselves what’s vital.
- Define solution looks at the buying journey.
- Define report drills down into outcomes; this is often the first time users really think about it.
- User path process defines buying journey and touchpoints.
- Attribution algorithm; assign values to touchpoints.
- Report/business intelligence: users need to be able to configure dashboards themselves, to fit their needs.
- Scenario planning helps to plan future spend based on results today. You need granularity.
- Channel activation.
Part II: Attribution Drilldown
Yaakov Kimelfeld, Chief Research Officer, and Michael Perlman, SVP, Media Practice, Millward Brown Digital
- The marketing funnel is dead; now the journey is via networks; you need to understand consumers in this context.
- Mobile is eclipsing other things; mobile is 13% of all Internet traffic now.
- Path to purchase analysis, we do it differently because we identify and analyze every user individually.
- Smartphone example: analysis of people considering two brands, and we also analyze people that didn’t buy. Search and social media are most important.
- Car buying example: it’s now a 28-day decision; it used to average months. You can influence the outcome until the deal is done. There is too much emphasis on ads; they have little influence.
Jarvis Mak, SVP, Customer Success, Rocket Fuel
- AI (artificial intelligence) and big data are game-changers.
- You need a data science team to value each [marketing] asset appropriately, so you can measure individual impressions [and attribute value]; you also need robust infrastructure.
- AI demo, a flying robot with six-foot wingspan doing an infinite loop in a parking garage; it’s adaptive due to its brain and sensors.
- The key point is AI is self-learning.
- There are many liquid pools of data anyone can buy [of buying journeys]. There are ad exchanges.
- But you need models to make sense of it; models do better with more data, get better results.
- Think about the user’s (audience of messages) constraints [within buying journey]. All of them.
- Rocket Fuel’s point of difference is “advertising that learns”; based on attribution analysis, we optimize buys, and we can do it because we score each touchpoint.
Part III: Attribution Wrap (Panel)
Panelists: Casey Carey, Madan Bharadwaj, Yaakov Kimelfeld and Jarvis Mak
- AI tells you what to buy, where and when, so humans can focus on strategy, defining KPIs.
- Measure “return on marketing investment,” but each channel needs different metrics.
- You can use attribution solutions to measure what content is most effective on non-ecommerce websites, like brand home pages. Also, internal portals and social networks.
- You can correct for competitor actions—and collaborative brand actions (owned by same holding company).
- To succeed in analytics, you need to appreciate the business context, know how to tell a good story [with the data] and have a passion for the field. Don’t be afraid of open ended questions; they’re often most valuable. Also ask uncomfortable questions when things are going well.
- Data professionals can transition from “reporter” roles to “analyst” when they add more value, ask the right questions, don’t hide in the data. When they focus on outcomes and make clients look like heroes.
Moderator: Tony Bombacino, CMO, BrightTag; Panelists: Shelby Saville, EVP, Managing Director at Spark Communications; Kristin Haarlow, VP, Director at Spark Communications
- Since the 90s, it’s the same old thing: “Right person, right message, right offer, right time.”
- Clients already think omni-channel, in terms of cause/effect in the ecosystem; consumers have access to more content but it’s still in silos.
- Is big data new? Well, it’s very useful to us because it’s on top executives’ radar; “data” used to reside in the bowels of digital.
- Analytics impact all decisions now; at Spark, we reorged the whole firm to make it integrated.
- The buy cycle for cars is now a month; it used to be six months, and that’s because consumers have more information, faster; to get ahead of this, you need data.
- Brands must be available when consumers need them; they don’t control the purchase cycle or process.
- Always on? Search and social media, yes, also display.
- Dynamic creative is disrupting; platforms enable us to test and iterate content/assets to see what works, real-time. Creative agencies are disrupted; their model has to change [because it’s based on a static world] where they are paid by the creative.
- Whoever has the data rules, but even more important is making it actionable; to do that, you need people to interpret it, so you can take action.
- Another issue is mixing our (agency) data with client data; it’s critical to establish trust between client and agency data teams.
Adam Greco, Senior Partner, Web Analytics Demystified
Adam shared his observations and offered some questions for the audience to ponder:
- Attribution test got a 10x higher “credits” than reality; double- and triple-counting is rife.
- Clients can’t even optimize their own websites, and they want to bring digital in-house? That won’t work, but they aspire to it.
- An easy way to test your attribution model is to take a channel and cut its impressions by half for a week, note the effect, and change channels.
- Threats: platforms Google, Adobe and Amazon are invading, and clients want to bring digital in-house.
- Is AI the future? More data requires machine-manipulation; it’s beyond humans’ ability.
- Is omni-channel the big bang, or is it better to optimize each channel? In either case, don’t boil the ocean.
- The “360 degree view” of the customer was the old CRM concept; now it’s impossible, data is expanding too fast.
- Is omni-channel all or nothing?
Analysis and Conclusions
- The same computing power that makes big data practical is bringing transparency and disruption to digital and offline advertising and marketing. The famous quote, “We know that half of advertising is wasted, we just don’t know which half,” is less true every week.
- Attribution means that every function of a website that enables viewers to interact (click) can be accounted for, in theory. It’s the accounting of the results of all the agency-created “stimuli” customers encounter on their buying journeys.
- Attribution systems, because they detail where customers click and when, enable brands and agencies to buy media more efficiently.
- Agencies are overwhelmed and clients are frustrated; based on speaker remarks it looks like many clients will bring digital in-house but don’t have the capabilities to manage much better than agencies.
- The digital media and advertising ecosystem is headed for an even higher level of disruption. It is focused on compelling customers to “convert,” which is legacy thinking in itself because it imposes the brand’s agenda on customers. As amply detailed in these pages, serving customers is far more potent in building trust, preference, commitment and profit.
- No one all day spoke about creating relationships with customers.
- Agencies as described are fragile because Google et al are organized in the cloud, so they enable clients to “buy by the drink,” a classic disruption to a legacy industry. Of course, large brands will continue to want to consolidate their buys, and agencies have extensive expertise. It will be interesting to watch.
- For a customer-focused view of big data opportunity, see the Big Data & Analytics Competency Center.