Moving beyond the Data Hype – SAP XM

When we started thinking about SAP XM as the advertising platform of the future – Beyond Programmatic, one of the critical aspects for us to consider was the question of Data:-

• Type of Data
• Quantity of Data
• Quality of Data
• Significance of Data
• Etc.

We were exposed to the various Data Management Platforms out there already doing a great job:-

• Advertiser DMP
• Publisher DMP
• Independent DMP
• Hybrid DMP
• Etc.

It was mind boggling!

I wondered, when ‘techies’ like us find it difficult, what would be the situation with the customers?  Hence, we decided to approach the need for managing the Data on our platform without any preconceived definitions in our mind and completely focus on what our customers and partners were telling us as their main challenges:-

• Challenge 1 – “I know we have a lot of data but I don’t know what it can do for me. ”
• Challenge 2 – “I want to reach my consumers at the right place at the right time”
• Challenge 3 – “I want to be closer to my customers”
• Challenge 4 – “ I already am working with a DMP – I don’t want to change my relationship”
• Challenge 5 – “Is my data safe?”
• Challenge 6 – “How can data help me improve my advertising RoI?"
• Etc.

The above challenges provided us with some base constraints:-

SAP XM needs to handle data generation across all channels i.e. cookie based, ID based, segments, collections etc.

• SAP XM needs to consider multi-dimensional data sets
• SAP XM needs to provide a secure and trusted platform to the customers
• SAP XM needs to be an extension of the customers’ enterprise landscape
• SAP XM needs to handle comprehensive data sources – 1st party, 2nd party and 3rd party
• SAP XM should allow for customers to onboard their existing relationships onto the platform – provide the choice to customers
• SAP XM must leverage the machine learning based capabilities to find intelligent matches
• SAP XM needs to exploit the best in class in-memory technology i.e. HANA

Our intention has always been to achieve the mythical ’segmentation of one’ – the ability to reach that single individual with the information that person needs at the exact moment she needs it. Yes we want to do the traditional use cases i.e. intelligent targeting, retargeting, look-alikes, cross channel tracking but also achieve some fancy things:-

• Sentiment analysis and allow our customers the ability to create personalized content which allows them to be really close to their consumers
• Intelligent Simulation allowing our customers to “test flight” campaigns before they go productive to get the best value for their bucks
• Fill in gaps in our customers’ data making it richer and hence allowing them more intelligence
• Use the power of machine learning to predict persona swings in order to increase the relevance of the message our customers are creating for their users
• Track life transitions  in order to not create advertising but trusted relationships

We have started off with classifying the dimensions we have access to on well known standards –

• 1st Party Data
• 2nd Party Data
• 3rd Party Data

Let’s take the opportunity and define what each of the above means in our context -

1st Party Data generated by SAP XM– the data which we generate by users working directly with our platform would be 1st party data. I would classify the following under this-

1. Registered accounts – advertisers, publishers, individuals
2. Registered users through website or the portal - individuals
3. The ad requests we receive from specific users (dependent on Data Privacy guidelines) – individuals
4. The bid requests we receive from SSPs
5. Advertising-specific events – view, clicks, win/los notices, conversions, etc.
6. Logs – context of individuals
7. Etc.

1st and 2nd Party Data available in the corporate systems – somebody else’s 1st party data utilized by us will be 2nd party data. We have an inherent advantage here because we have lots of customer data within SAP system and if the customers agree we could utilize this intelligence pursuant to the constraints defined by the customers. Some of the systems which could provide us with this data are –

1. CEC systems

• yProfile – contextual information across multi-dimensions, multi-account; account specific information - individuals
• yMarketing – is it still onPremise? If yes, then customer specific information - advertisers
• yCommerce – user buying patterns from specific eShops / accounts – retailers/advertisers, publishers, users

1. SAP backbone – account information, customer purchasing history etc. – advertisers, publishers
2. SAP Cloud Solutions & onPremise Solutions run by the advertisers or publishers – S/4HANA, SAP CRM, SAP C4C, SAP CAR, etc.
3. Networked offerings – advertisers, publishers

3rd Party Data – curated data sets which are available as packaged offerings to buy or sell would be classified by 3rd party data. This is data which is provided by data aggregators who are focusing on mining offline and online user stores and creating relationships using various theoretical and practical algorithms. Data aggregators correlate the information on various matrices to create widely different and distinct patterns from same set of information. Usual examples would be

1. Demographic
2. Contextual
3. Behavioral
4. Etc. 

Approach

Typically, we would utilize 3rd party data to supplement our 1st party and 2nd party data sets in order to offer better value to our customers. However, if I have to put a prioritization on the 3 broad categories I would believe that the 3rd party data would bring us the least amount of differentiation. At the same time, it’s probably the most well-known data set in the context of advertising and hence even though I believe it’s a commodity it needs to be supported by us already in the 1st iteration.

One of the biggest USPs of Google, FB and other Walled-Gardens is the fact that they have closed user ecosystems. A goldmine of information of individuals directly interacting with them. Huge 1st party data!

SAP XM has the unique situation of potentially also having access to similar amount of 1st party data as well as 2nd party data but where we have an unfair advantage is that a lot of the data we have access to is business / transactional data.

Our open approach allows us to create the connections between the various data dimensions in order to create correlations and patterns and use the data intelligence. We have already established some interesting connections. The initial results of the intelligence we are able to derive in the context of rich & holistic profiles look very promising.

We obviously still have additional challenges:

1.    Some identifiers change over time (more or less often), for some identifiers we will have multiple valid values over time (e.g. home-IP, business-IP)

2.    The quality of data needs to be good and how do we ensure that

3.    Which is the best algorithm to be used to create perfect matches

4.    Identification of the user needs to be extremely fast and not constrained by traversal complexity

5.    Move beyond cookies and even traditional IDs or support adhoc type of IDs

The one thng i am absolutely clear about is that our open approach allows us to work with multiple partners and that also adds to the value we bring to our customers.

Our work is cut out for us – stay tuned and I will let you know how we go along…J