Many organizations are challenged by the rise of the "empowered customer." Today's customers are well-informed, use other people as their primary information source, and interact with companies through multiple customer contacts (touch-points) and media. The customers also interact with companies through various channels, such as email marketing, websites, and social media. They want a superior customer experience and have visible outlets for venting frustration when they don't get what they want.
Where does your organization get the information about customers and potential customers? If the information source is not trustworthy, then no matter how good the marketing is, offers could be made to the same person multiple times (for example, if there are duplicates in the information source), or a person may not get an appropriate offer (if information associated with them is unreliable). This leads to dissatisfied customers. With Enterprise Marketing Management (EMM), there is a single place where marketing messages are authored, so the messages are consistent no matter which channel they are delivered through. Using master data as the trusted source of information lets EMM send relevant offers to the right people, increasing the chance of offers being accepted. This article explores patterns of using Master Data Management (MDM) in EMM solutions and approaches (see Resources for more information).
The terminology that is used in this article is defined in Appendix A: EMM and MDM Glossary.
If you are already familiar with EMM and MDM and why you need them, you can go straight to EMM and MDM Patterns
Marketing can be very emotive, so as architects, do you promote good feelings and try to reduce the strong negative feelings that arise around marketing, such as:
- "I got the wrong offer. In fact, it seems like the offer is for someone else!"
- "I have been asked again to join the loyalty program."
- "I got the same offer lots of times."
- "I told them not to mail me, but they keep doing it."
- "Will they stop pestering me? I told them I was not interested at the moment."
- "I said I wanted to get SMS but they are emailing me."
Many of us have experienced being contacted for something we don't want, by people who don't seem to know who we are, sometimes repeatedly, leading to frustration and irritation. Or maybe there is a new product or offer out there that you would love to buy, but you were the last to know and missed out.
Many marketing organizations can't seem to get a complete understanding of each customer—that is, creating a complete profile of their customers, which captures each customer's interactions with all the various marketing efforts. What messages has each customer been presented with? What prices have they been presented with? How did they respond? Which touch points has each customer interacted with? What happened during those interactions? Which marketing efforts were effective?
When architectural patterns and business processes are in place to collect the customer's information and activity—and are accessible by the marketer—the marketer can reliably build more effective campaigns. When marketing is done well, it can make customers:
- Feel that the company has offered them just what they wanted or needed.
- Feel helped rather than harassed.
- Feel that the company is working with them. The company understands who they are and wants to help.
- Become loyal advocates who are vocal on social media and draw others to buy the products.
If you are already familiar with EMM, then go to types of master data that EMM utilizes
If each part of the business has its own marketing department, with potentially different strategies, then this can lead to confusion and inconsistent messaging and experiences for customers and prospects.
With the EMM solutions, the marketing organization creates artifacts in a channel-agnostic way, allowing the same marketing strategies to be used across all the channels. This provides consistent marketing messages, so it appears that one person is behind all the channels guiding the dialogue and relationship. Customers then get a consistent experience from your organization. Figure 1 shows the marketer authoring marketing campaigns, which when executed can be delivered down any of the channels.
Figure 1. Marketer and EMM channels
Figure 1 also shows that customers and prospects contact your organization through multiple channels, as shown in Figure 2.
Figure 2. EMM processes
Figure 2 also shows that a person is at the center of the EMM critical processes, where:
- The decision is made about what the best offer is for an individual.
- The message is delivered and the customer's response captured.
- The entire business process and marketing budgets are managed.
- EMM collects data to augment the customer profile; this data can then be analyzed for actionable insights.
Using IBM Campaign (formally IBM Unica Campaign), the marketer designs campaigns for outbound marketing, which when run deliver relevant offers to the appropriate individuals using their preferred channels including email, mail drops, printed material, SMS, and others. Campaign has the concept of Audience Levels, such as individual or household, which can be used to target marketing campaigns to different groups of people. Audience levels hold unique references in the source data; for example, these unique references can be the primary keys of a database containing customers.
Using IBM Campaign, you can maintain a dialogue with the customer across the entire life cycle of interactions. Initially, you can create campaigns to search for new prospects, followed by brand differentiation. After the prospect has become a customer, you can use segmentation to create even more relevant marketing offers. You can create campaigns that are targeted for up-selling and cross-selling opportunities to nurture your existing customer relationships and create retention-focused campaigns to re-engage customers who haven't purchased in a while.
IBM Interact (formerly IBM Unica Interact) allows you to do inbound marketing. When a user comes to your organization's website, call center, Point of Sale (POS), or other inbound channel, Interact finds the most relevant offers for that person in real time, then presents them to the user. IBM Interact shares components with IBM Campaign and can access the same individuals as Campaign.
Figure 3. IBM Interact
Figure 3 depicts a web channel with an area within the page that Interact delivers content into. Interact works by:
- Real-time customer interaction uses content from the customer profile and data from real-time calls to create a list of candidate offers and scores using segmentation and rules.
- The offers are adjusted using blacklists and whitelists and suppression rules. Blacklists are lists of offers to be excluded from the candidate list of offers. White lists are the opposite and are lists of offers that should be included in the mix—either globally, by customer segment, or at the individual customer level.
- The offer score determines the final "winning" offer to be presented to the customer. This score can be determined by the marketer or by using the self-learning algorithm, which uses the "wisdom of the crowds" over time to build a score value for an offer.
Interact records whether the offer was accepted or not, and this information can be factored into future batch and online campaigns.
IBM Campaign uses information that is already held in the enterprise. This source data (including information about the people that could get marketing offers) can be located in one or more databases, which are mapped into Campaign and used to target customers and assign them offers.
Figure 4. EMM source data
Figure 4 shows some of the types of data that can be used by IBM Campaign. You can group the types of data into four groupings:
- Person descriptive data:
- Customer ID that uniquely identifies the customer.
- Customer demographics, which is the information that the marketer uses to segment.
- Contact details so you know how the customer likes to be contacted. Respecting the contact preferences lets you ensure that you send offers down the right channels.
- Existing contracts, which gives you details of the contracts (sometimes known as accounts) that you have with this person
- Product information is needed to understand what the previous buying or browsing behavior was and the product that is tied to the offer when it is made.
- Person activity: Product purchasing history can give an indication of likely future purchases. Transactions show you the previous interactions with the organization. Other activities may include social media information.
- Precalculated insightful data includes scores from predictive analytics and may include previous activity data.
With precalculated predictive scores, analytics are used to calculate insightful information in advance (for example, whether the customer will leave the company).
Precalculated social sentiment analyses are metrics that are calculated in advance using analytics and show whether a customer is happy or unhappy based on their social media interactions. This information can be used to influence what offers are presented to the customer.
Relationships show ties to other people, including whether the person is in a household or family. This allows the marketer to target offers at a household or family level.
If you are already familiar with master data, then go to The business value of combining MDM and EMM.
Data often resides in different silos, only being available to a small part of the enterprise. For some information that is core to the working of the business, there is often duplicated, mismatching, and incorrect versions in each of the silos. Where this data is of high value (for example, customers, products, and accounts) it should be accessible to all parts of the enterprise.
Master Data Management (MDM) solves this problem by creating a consolidated version of this data (in what is called the master data hub) and provides common ways of accessing it that the whole enterprise can use. MDM has processes that allow the data to be normalized, not duplicated, and synchronized with other copies of the same data. For example, rather than a marketing system getting the data about people from its own database, it gets it from the master data hub. In doing so, the marketing system gets the same trusted information about people from the same source as other parts of the enterprise. So, not only is master data trusted, due to the processes around it, it opens up the data so that all parts of the enterprise can see it.
There are different approaches to implementing master data.
Virtual information registry uses read-only copies of the data in the source systems to dynamically create a consolidated view of the data. This approach is useful when assembling a single, read-only, tailored view of master data. The data is still updated in the source systems.
Physical (persistent) master repository stores a consolidated version of the master data. Information sources are fed into the repository. There is a single place where the master data can be managed, including creating, updating, viewing, and deleting the data. This is an ideal approach when you want a centralized place to manage your master data.
Key IBM InfoSphere MDM capabilities that EMM can capitalize on are:
- Single view of trusted master data
- Collaboratively create and manage complex product catalogs
- Accurately identify unique customers/prospects
- Manage household and relationship views
- Works with all industries
- Works with all styles
IBM InfoSphere MDM has a number of editions. The base functions are in the Standard Edition, and the other editions introduce valuable additional capabilities including other MDM styles and domains.
- Standard is a virtual master registry that has a flexible and extensible Data Model, which is built and optimized for MDM.
- Advanced has a virtual master registry or physical (persistent) master repository, which has pre-built and extensible data models for any domain. It includes a physical and logical data model that supports cross-industry, multi-domain (party, product, account, and so on) requirements and is enhanced by industry-specific templates.
- Collaborative contains collaborative processes and an extendable authoring UI for creating and updating master data.
- Enterprise supports all MDM implementation styles: virtual master registry, centralized physical master repository, and collaborative authoring. It also has support for all master data domains: party, product, account, financial, location, and custom domains.
This article describes approaches that can be used with EMM and a single view of master data with these MDM editions and talks mostly about Person data. It is also possible to manage account and product data using InfoSphere MDM.
With EMM, you have a single place where the marketer makes decisions about what offers are presented to the customer, but it is not all-knowing. It needs to be connected to high-quality data sources.
The Party domain in a master data hub creates trusted views of the customer, including the customer's identity and demographic information and contact methods. The customer segmentation that is used by marketers typically involves customer demographics and previous customer activity. Getting reliable customer information, including contact methods, means you can be confident that offers will be presented to the right customer and honoring their contact preferences.
The key business benefits of using EMM and MDM are:
- Better targeting of offers, leading to higher response rates, net revenue, and number of products/bundles sold per offer
- Reduction of ineffective marketing
- Improved accuracy of segmentation and forecasting behavior
- More accurate product/bundle information in offers/campaigns
- Identification, retention, and growth of profitable customers, which reduces churn
Getting the data right for EMM with master data describes master data and The business value of combining MDM and EMM describes the business value of populating your EMM solution with data from a master data hub. Now you will look at how EMM can use this master data to underpin the EMM solution.
From Figure 4, the Person data should be managed in the master data hub.
IBM campaign, at a minimum, needs:
- Customer contact methods
- Customer identifier
- Customer name
Marketing campaign definitions are likely to require addition information about the customer, such as their demographic profile. This depends on how the marketer makes the decision and what information influences the decisions on what offers to present.
IBM Interact uses the same profile information, offers, and contact history that IBM Campaign uses. Interact can also fold in additional real-time customer data that it can acquire through a web service call to the master data hub. All of this data will help arbitrate which offer should be presented to the customer.
Information about individuals and how to contact them is picked up by EMM from an external data source that is called the marketing repository. This marketing repository should be populated using data from the master data hub. It is technically possible for EMM to use the master data hub directly. However, it is recommended that you have a separate copy of the master data so that large queries on the marketing repository by EMM do not affect the performance of the operation master data services.
In today's world, marketers may have many concurrent marketing campaigns, and it can be difficult to limit contact fatigue, prevent conflicting offers, and meet channel or inventory capacity limitations. Within the EMM suite, IBM Contact Optimization (formerly IBM Unica Optimize) enables marketers to determine the optimal contact strategy for each customer over time by looking across the proposed offers and channels across multiple marketing campaigns. The optimized contacts are created by using information such as customer contact preferences, contact history, business rules, and constraints. Again, this data about the customer can be populated using data from the master data hub.
The general patterns for distributing master data to the marketing repository are shown in Figure 5.
There are several approaches to populating the marketing repository from the master data hub. Figure 5 shows the EMM and MDM architecture (see larger image).
Figure 5. EMM and MDM architecture
- ETL (Figure 6): Data relevant to EMM is
extracted from the master data hub repository and transformed into the flat
simple objects, which are then loaded into the marketing repository. This approach
can be used for the initial bulk load of the marketing repository and also for
subsequent delta updates. This approach is ideal when there is a lot of data to
be moved in batch.
Figure 6. ETL process
- Messaging (Figure 7): Messaging is a
trickle feed approach, where the master data hub is configured to produce events
when the master data changes. Events can be issued every time an
element in the
master data changes or more selectively. The event kicks off its processing in an
Enterprise Service Bus, which maps the master data into the marketing
repository. This approach is a trickle feed and is ideal when updates must be
replicated to the marketing repository as quickly as possible.
Figure 7. Messaging process
- Change Data Capture (CDC) (Figure 8):
When the master data hub makes changes to the master data, these changes are
recorded in its database log. Changes to master data are identified from the
log and immediately passed to the marketing repository. CDC also has simple
mapping routines to perform the correct conversion. This approach is
appropriate when the master data repository is heavily loaded and changes need
to be replicated with minimum impact on its performance. CDC can be used in
combination with ETL, with ETL doing the transformations.
Figure 8. CDC process
- Export: Where the master data repository has a built-in export process that allows master data to be extracted in different formats.
- ETL implementation: IBM InfoSphere DataStage is ideally suited to doing high performance ETLs (with built-in data lineage), for the initial load of the marketing repository. InfoSphere DataStage also allows you to do batch delta updates. Currently, it is common practice for the marketing repository to be updated nightly (when the system is quiet), so running DataStage jobs nightly to keep the marketing repository up to date with customer information is a good fit.
Messaging Trickle feed implementation: Using the IBM InfoSphere
MDM notification framework, as master data changes events are produced that put
messages onto an IBM MQ queue with the details of the changes. These messages
drive IBM's ESB - WebSphere Message Broker (WMB). The WMB mapping node is then
used to map the notification changes to the marketing repository.
Figure 9 shows a WMB flow with a mapping node in it.
Figure 9. Messaging broker nodes
- With change data capture: IBM InfoSphere Change Data Capture is a database log-based approach, so as master data is created and updated these changes are replicated in real time to the marketing repository. CDC can be used to feed InfoSphere DataStage ETL jobs.
- Export: With IBM InfoSphere Collaborative Edition, you can directly export its master data to a file or database.
If you want to jump directly to how you can evolve this solution for the future go to What next?.
Deciding which attributes are required in the EMM source is a business decision that may involve the chief marketing officer, the marketer, the business analyst, and the information architect. Typically, details of the products being sold and the customer, including their preferred contact method and historical transactions, are in the marketing repository.
Though the basic customer information is the same across industries, additional industry-specific attributes are useful for EMM. For example, for a telecom provider, the number of dropped calls is important but is unlikely to be useful for other industries.
The products that are sold and the historical transactions format are also industry-specific.
IBM has the Industry Models, which are standard attributes for particular industry verticals. It would make sense to use a subset of these attributes as the basis for your marketing repository. Alternatively, a custom database physical model could be created for the marketing repository schema.
Up to now, this article has discussed how IBM Campaign can use master data as a source of master data. After IBM Campaign has run, the customers have been segmented. The options on where to store the contact and responses are:
- If required for offline processing, then it can be put into a Data Warehouse or a Data Mart.
- If required for online processing, then it needs to be in an operational database. One option is to store some of the results as attributes in the master data hub. MDM Advanced Edition has the Campaign feature for this purpose
This section looks at the considerations around introducing master data when your existing enterprise already has a marketing platform.
An existing EMM solution may not have data sources that have been created with master data in mind. These data sources may need to be amended so that their content can be correlated with master data. You need to decide for these existing data sources whether existing data will be either deleted and the repository reconstructed from scratch or used as the starting point for updates.
An approach that can be used to update the marketing repository in place is to:
- Identify the data that needs to be replaced.
- Extract it.
- Cleanse it (including de-duplication). Untrusted data can be imported into the master data hub then cleansed using MDM processes; this process of cleansing the data in place is called ever-greening.
- Write back the cleansed data to the marketing repository.
Figure 10. Cleansing the marketing repository
Figure 10 shows the data being extracted from the marketing repository. The process shows the data being standardized and de-duplicated, then reloaded into the marketing repository.
With EMM, you are looking to have the marketer construct the campaigns and make the decisions around offers in a single place so that customers are treated consistently. For this reason, if your existing master data already stores offers and campaigns, then you need to ensure that these are sourced from EMM.
For IBM Campaign, it is possible to derive data for the marketing repository from your data warehouse, either because you do not have a master data hub, or you need historical values about the customer and the customer's buying history.
If you are taking data both from the data warehouse and the master data hub, it is important that these repositories are synchronized together so their information correlates.
To answer this question, you need to know which attributes in your marketing repository are changing, how fast they are changing, and what the effect is on the offers that are presented.
Current practice is to update the marketing repository every night. This may be good enough if you run batch campaigns or if the master data is not changing very much in a day.
As marketing organizations look forward and use real-time marketing, updating the marketing repository daily may not be good enough. Using the messaging or CDC approaches allows more flexibility around keeping the marketing repository up to date.
The structure of products, including how their hierarchical, category, cross sell, up sell, and bundle relationships, are stored in the product master data. So analyzing the effectiveness of marketing campaigns are major considerations when authoring your products in IBM MDM Collaborative Edition. As with customer master data, product master data is in the marketing repository and, as such, is mapped to IBM Campaign. The product is what will be sold if the marketing offer is accepted by the customer.
You have MDM and EMM existing together with your business reaping the benefit. So, what next?
A reasonable next place to look is the IBM Smarter Analytics Signature Solution Next Best Action. This is a pattern that is centered on a 360-degree view of the customer, captures customer activity (activity in the widest sense including social networking content), and coordinates analytics management (predictive analytics) with real-time decisions to optimize how to best treat your customers. Next Best Action includes master data and a marketing platform, which can be implemented using IBM InfoSphere MDM and IBM's EMM solution. Figure 11 shows that the EMM gets its data from the Master Information and System of Record database.
Figure 11. Next Best Action pattern showing data sources
Using EMM you can create consistent marketing across your channels. Underpinning EMM with trusted data, including master data, means that you truly know who your customers are. This can lead to increased customer satisfaction, higher marketing response rates, and reduced churn.
If you're unsure whether you need a master data management initiative to complement your EMM initiative, ask yourself if any of the following statements are true in your organization:
- Inconsistent or incomplete data about your customers and prospects prevents you from accurately segmenting customers to the level of granularity you desire
- The same customer or prospect is sometimes duplicated in your marketing repository
- Your marketing data doesn't reflect your customers' activity across all your channels, only within specific channels
- Mapping your marketing initiatives back to your customer or prospect data is complicated because the data model is different for every application that holds customer or prospect data
- You don't have a single place where you can track enterprise-wide (channel-agnostic) information about your customers, such as privacy preferences, contact preferences, or third-party-provided demographics
- Your current sources of customer and prospect data are a blend of person-centric data and household-centric data, but you don't know how to combine them properly so they can be used as a solid base for your marketing efforts.
After you have a single view of the trusted master data for EMM, you can leverage this source of trusted data across your enterprise. Master data then becomes the trusted foundation of your Smarter Analytics and your Smarter Commerce solutions.
I would also like to thank the following people for their help:
- Mandy Chessell - IBM Software Group, Information Management, IBM Distinguished Engineer, Master Inventor, Chief Architect for InfoSphere Solutions
- Bhavani Eshwar - IBM Software Group, Information Management STSM, InfoSphere MDM
- Lena Woolf - IBM Software Group, Information Management STSM - InfoSphere MDM
- Nigel Jones - IBM Software Group, Information Management InfoSphere Solutions
- Vanessa Melaragno - IBM Software Group, Worldwide Sales Senior Client Solution
- Jon Case - IBM Software Group, Information Management Product Management, MDM Enterprise Industry Solutions
- Mike Cobbett - IBM Software Group, Information Management Master Data Management Tools Architect
- For more information on Smarter Commerce, go to Smarter Commerce.
- For more information on Smarter Analytics, go to Smarter Analytics.
- Find everything EMM at Enterprise
Marketing Management (EMM).
- Find everything about IBM MDM at Master Data
- Get terminology and definitions in the EMM and MDM glossary.
- The red guide that is mentioned in this paper is available
Analytics: Driving Customer Interactions with the IBM Next Best Action
- Stay current with
developerWorks technical events and webcasts focused on a variety of IBM
products and IT industry topics.
- Attend a free
developerWorks Live! briefing to get up-to-speed quickly on IBM products and
tools as well as IT industry trends.
- Follow developerWorks on
- Watch developerWorks
on-demand demos ranging from product installation and setup demos for
beginners, to advanced functionality for experienced developers.
- Get involved in the My developerWorks
community. Connect with other developerWorks users while exploring the
developer-driven blogs, forums, groups, and wikis.
David was the technical lead for the development of WebSphere Product Center 5.3, now MDM Collaborative Edition. He also worked in MDM Workbench development for 5 years, which involved creating simple model-based user interfaces to generate customized MDM Server implementations. In his current solution architecture role, David is promoting and developing trusted predictive analytics and EMM solutions by putting master data at their heart. In 2012 David presented at the Information On Demand conference on EMM and Master Data and demonstrated the next best action solution.