The CDP Institute organizes monthly online roundtable discussions for all community members.  Here are notes from our November sessions.  Keep an eye out for email and calendar announcements of the next set of roundtables, which will be in January.

APAC Roundtable, November 27

Guest: Neeraj Pratap Sangani, CEO, Hansa Cequity

Martech results

• Mixed results from martech: some businesses feel tools help to do a great job; other businesses are dismayed have invested so much in tools but outcomes are taking too much time or not happening.
• Technology teams often lead the conversation, which is good from technology and implementation view but can harm outcomes.  Need business users to be more hands-on with tool selection and use, cannot delegate to tech or data science teams to use for them.
• In APAC, very mixed bag with clients at all different states of evolution with tool use.
• Average tenure of CMO is 2 ½ years, so are impatient to show outcomes.  This is possible with the right teams and tools in place.
• GenAI agent hype is latest shiny object, driven by stock market.  Brands are forced to pretend or take action.  Is some moderation of expectations but hype still exists.

Being customer-centric

• Different streams contribute to becoming a customer-centric enterprise: data science and behavioral science process data to show what has happened, but not why it happened; campaign, digital marketing, and service teams use the data.  Each stream requires specialized, deep skills and capabilities and it’s a large job to bring them all together.

• Dynamic personalization is now possible.  When visitor engages, he can be assigned to a specific audience type based on behavior, this can be embellished with derived data and used to target when they are near a physical store or visit the website.  Can have hundreds of customer segments with automatically -triggered, different content, offers, recommendations for each.
• Omni channel journey includes not just martech channels but also online, offline, dealers, service, customer care.  Can map out the journey on a wall, review with product, marketing, service teams to understand gaps and how CDP or martech can close them.  Need to optimize customer journey, not just count number of journeys or daily customers or click rates or open rates and call these outcomes.
• Example: customer clicks on shoe ad; real time CDP captures the interaction and adds it to the customer profile,  then applies AI to analyze the data based on past purchases by product, price, season, etc.; estimates likelihood to purchase now, likelihood to churn, etc.; generates personalized recommendation for color, size, style, price, etc.; and uses this to trigger a personalized campaign in his preferred channel.  Digital ads can promote the product or drive customer back to the website which is dynamically adjusted to show the product he has been considering and, if he has high probability of churn, to give aggressive offers such as exclusive discounts or early access to new collections.  This is real today at digital first companies, compared with ten years ago when we were just talking about it.
• Real time use cases are still aspirational for most brands.  People see value but requires mature platforms, people, organization, technology, which are not yet all in place and mature at different times.  Real time is not relevant for all businesses.  But even for highly considered purchases, real time helps to understand the context and make the right pitch. 
• A few big digital brands, the ecommerce players, are doing real time very well.  Others will learn from their experiences.

Real time CDP Applications:

• Conversion rate optimization: in the past, profiles had 200-300 variables; today, profiles can be much larger but then people lose track of what are relevant metrics and the marketer’s ability to run programs is called into question.
  • Is new focus on conversion rates, which can reach 10% to 15% in real time programs such as financial institution sending triggered campaign messages to customer when there is a jump in the stock market.  Banks have data to select the right customers to receive those messages
• Predictive maintenance and proactive service: electric cars have many fewer parts than gas-powered cars, which reduces revenue for manufacturers.  Personalized, subscription-based services using real time data will be new revenue stream.  CDP will be source of data to support this.
  • Auto industry has seven year cycle to develop new platforms and customers in Asia buy new cars every five to seven years.  AI is collapsing development cycles to three or four years.  Cars are now seen as platforms, with services sold in addition.  Need to sell correct services: BMW had misfire when tried to sell heaters in seats as service, and customers rebelled.  Services will be big revenue stream, can be added through dashboard without visiting service center.  Services can include entertainment, purchases, experiences, maintenance, charging locations.  May be opportunity for dynamic pricing in some markets.  Cars already have several hundred sensors throwing off “zillions” of data points; marketers must understand which data points are important to them and to serve the customers.  CDP helps to make this data available and to take advantage.  Car itself can be an activation channel.  
  • Example of using real time data: in 2016, was with friend who had Tesla.  Car had problem and immediately received call from Tesla which reported the problem and offered to fix it.  This is real, high-value personalization, far beyond writing customer’s name or sending birthday greeting.
  • Now see personalized offers on websites, generated by AI trained on CDP data.  Use templates to ensure brand consistency when AI generates content.  Brands are aware of risk of making mistakes and use frameworks to control this.  Has to mature but seeing many POCs.
• Acquisition: CDP can add valuable insights about early stage customers, even when anonymous.  Can look at behaviors on website and social media, help to improve lead management processes.  Can tell when kind of people are visiting so can target on digital, what kinds of products they are viewing, how long they stay, whether they go to the shopping cart.  Many brands are capturing this and acting on it.  Depends on the focus of the marketing teams.
  • Example: auto buyer books test drive, visits dealer, salesperson adds information to DMS about test drive, model viewed, financing options discussed, when customer expects to buy, other brands he is considering.  Data goes from DMS to CDP, which recognizes customer is a hot prospect and can select best offer to close the sales.  This sort of proof of concept is common, used to show value and convince management to make the investment in effort and expense.
  • APAC markets are growing quickly, e.g. 7% per year in India as high as 12% in some industries.  This leads to stress on acquisition, not retention, which is where CDP is most helpful.  This will change as APAC markets mature and become more like US and European markets, where are growing more slowly and are more interested in extracting value from current customers.  Will change quickly in APAC.
• Fraud detection: fraud is very big problem, hurts both customers and businesses.  Real time data helps with fraud detection.
• Analytics: predictive analytics, churn prediction, forecasting, marketing results measurement require CDP for real time data.

CDP Deployment

• Data integration is greatest complexity in CDP.  Platforms are largely integrated with data science and analytics, but is still room to put in place real-time optimization using modeling, predictions, and trend forecasting.  This supports not just customer experience but also internal operations such as inventory management.  Are many complex systems to integrate, including legacy systems, as well as DMS (Dealer Management System, for automotive), CRM, CDP, website, customer service, third party providers.  Need to ensure rich data quality and standardization.  All adds to costs.  Must balance cost vs efficiency.  Unifying data science, behavioral science, digital experience, and real time capabilities is needed to create a great customer journey.  Won’t be seamless but can work in unison towards one destination.
• Today, CDP usually doesn’t have real time connection with CRM; more often 24 hour or weekly update.  Makes real time personalization impossible.
• CDP is taking over the role of building single customer view.
• Some companies store data outside the CDP to reduce costs, but this dilutes the value of the CDP and makes it harder to add value.
• Marketers need to take more interest in data-driven strategies and in genAI.   Need to build business cases to justify new investments.  GenAI is expensive, especially in APAC.  Did cost analysis in India and found human labor was still cheaper than genAI, because of expensive skills needed.  Expect costs to come down as systems mature.

US/EMEA Roundtable, November 27

Guest: Joaquín Rihuete Cortíjo, Martech Strategist, Omega CRM, a Merkle Company

Curated Ad Audiences

• can no longer use first party data to do customer match or look alike models with Google in Europe
• instead, use curated audiences based on private marketplaces and curated packages to generate quality audiences; are just starting to test this.  Curation uses shared ID such as ID5, Liveramp, Utiq, Merkury ID to combine data from different sources, such as CDP and publishers, in the SSP; this is a challenge.  Curation requires advertiser to have large first party audience (such as Verizon in US with 100 million customers) to match against the other sources (i.e., publishers); hard to make it work for advertisers with small audiences.  Being pushed by industry to find alternatives to walled garden audiences.  Curation also allows publishers to sell their first party data when are allowed to.  Is also benefit for SSP, which gains more quality, and for advertiser, who gets better results.

Identity Platforms

• Merkury is global identity platform from Merkle/Dentsu; has panelist data in Europe; has more than 10,000 data attributes per profile; can connect to CDP and run machine learning or AI look alike models within Merkury; can also code CDP profiles with advertising ID to match against other audiences.  Merkury has demographic attributes; in the US, also has geolocation, email, phone numbers, interests, other directly-gathered data.  In Europe, is more algorithmic data.  No business data such as industry, job role, demonstrated interests.                             
• Utiq in Europe has high very match rate, since data comes from major telcos and publishers and is fully compliant with local privacy regulations. In Spain, claims 100% match rate, are working with main publishers and have new way to collect consent which is based on explicit opt-in.  Consent may be limited to 90 days but not sure. 
• Curation vs walled gardens: Google Audiences are very large and powerful from the start and show higher conversion, while first party audiences are smaller but create higher engagement.  Audience is usually credited (or blamed) for engagement but it is also influenced by other factors such as website conversion optimization.

Impact of AI

• AI in advertising for lookalike is best run in advertising/identity platform rather than CDP, because ad platform has the most volume.  Most machine learning or AI in CDP is used to build segments and make predictions. 
• Haven’t seen use of AI for attribution; is almost no demand.  Very big companies can do multi-touch attribution, others use last click from ad server or Google Analytics.  GenAI gets most of the attention today while attribution would use other statistical techniques to build conversion models.  Relatively easy when you have a short sales cycle and lots of conversions; much harder if you have a long sales cycle.  Is big investment to build and, in the past, clients didn’t use it even when it was available because was too complex.  Clients use last click or maybe simple fractional attribution.
• One effect of GenAI is to force companies to clean up their customer data, which creates better first party customer databases as a by-product.  This is a role for CDP, supplemented by an identity platform and advertising platform to add addressable IDs for advertising. 
• GenAI also creates need for access to documents with unstructured data, such as files, PDFs, audios, video, etc.  Once that data is made available, can be used for advertising, service, sales, owned media, own channels, etc.   Vendors such as Salesforce are building for that but few companies have actually invested in this.
• Salesforce is making the Data Cloud CDP their foundation for everything, and pricing it to be affordable for smaller companies as well as enterprise.  Agent Force also draws on the Data Cloud CDP. 
• Companies are very open to genAI because the interface is very easy.  Research finds that testing is very widespread but production deployment is more limited.  Putting AI into production requires quality control and supervision to ensure mistakes don’t happen, which slows adoption beyond the test stage.  Slow adoption is common to other new technologies but what’s unique about AI is the scope of potential applications is so broad.