I had a fascinating chat earlier this week with a client who described his vision for using DemandBase to tailor messages to Web site visitors from target accounts, using Bizo to further tailor messages to individuals by title, using all this data to synch inbound and outbound campaigns in Eloqua, and eventually driving everything with predictive model scores from a tool like Lattice Engines. That could serve as a pretty complete summary of the state of the art for B2B marketing today, especially if you consider “content marketing” as implicitly included. Equally helpful to me personally, it reinforced my intention to write about Bizo and DemandBase, both of which have recently briefed me on their latest product extensions.
Let’s start with DemandBase. Astonishingly, four years have passed since I last wrote about them. In that time, they’ve continued to build applications that exploit their core technology for identifying Web site visitors by company based on their IP address. This started by providing visitor lists and real-time alerts to sales people who were interested in specific accounts. It later extended to returning visitor attributes in real time so companies could pre-fill forms and personalize Web pages to match visitor interests. The most recent expansion went beyond a company’s own Web site to the much larger world of online advertising.
To reach that market, the company had to build its own version of “data management platform” (DMP) systems that manage lists of known entities, recognizes them when they appear on an external Web site, and delivers them an appropriate advertisement. The big difference is that DemandBase entities are companies identified by IP address, while traditional DMP entities are cookies attached to browsers (and assumed to relate to individual human beings). DemandBase had to build its own engines for real time bidding (RTB) and ad serving (Demand Side Platform or DSP) to support its approach. These can integrate with Demandbase’s own network of Web publishers that will accept its ads and with other ad exchanges that connect to their own, larger publisher networks.
Data in the DemandBase DMP comes from both DemandBase and clients. The DemandBase data are the company-level attributes that DemandBase has long assembled: company name, industry, revenue, employees, technologies used, etc. Some of this, such as DUNS Number, is purchased from external sources and requires extra payment. The client data, which of course is available only to the client who provided it, could be anything but is usually attributes such as account type, buying stage, and sales territory. The system doesn’t store any information about individuals. Marketing automation, Web analytics, and Web content management systems can all access this data via API calls for analytics and as inputs to their own selection and treatment rules. Outside the DMP itself, DemandBase can store content and decision rules to guide bidding and select which ad is displayed to each account.
So much for the mechanics. The business value is that DemandBase is allowing marketers to tailor messages to target accounts even before they engage directly with the company, thereby (hopefully) luring new prospects into the top of the funnel and engaging them if they don’t respond. This is a major extension beyond traditional marketing automation, which works mostly through email to known prospects. It also goes beyond Web site personalization, which requires people to at least visit your Web site and in most cases actively provide information about themselves. As you might imagine, DemandBase offers many case studies to show how much this improves performance.
Bizo comes at Web advertising from the traditional route of building a pool of cookies and assembling them into audiences based on the attributes of the individuals they represent. The pool was originally used to target display advertising and retarget site visitors by sending them ads on other sites. The company says it has pulled data from 4,200 publishers and other sources to identify about 120 million individuals worldwide, including 85 million within the U.S. The number of actual cookies is higher still.* Profiles contain titles and business demographics such as industry, but no personally identifiable information such as names or addresses.
Like DemandBase, Bizo has found many applications for its core data asset. These now extend beyond display ads to social media advertising through Facebook and LinkedIn, Web site personalization through Adobe, Web analytics through Google Analytics and Adobe, and integration with Salesforce.com CRM, BlueKai DMP, and Eloqua marketing automation. Other partners will be added over time.
I’ll assume the Eloqua integration is most interesting to readers of this blog. Basically, it lets Bizo read audience segments created by Eloqua. Bizo then matches segment members to Bizo identities and delivers Web site, advertising or social messages tailored to each segment. Because Eloqua captures such detailed information about prospect behaviors, this allows highly tailored advertising that is tightly synchronized with prospects’ progress through buying stages and marketing automation campaigns. Since it’s driven by cookies, it can send messages to anonymous as well as identified prospects – a huge expansion in marketing automation’s reach. Bizo can even allocate advertising spend across the different media to achieve reach and frequency targets as efficiently as possible. To encourage this approach, its pricing is based on the number of unique individuals that marketers manage in its system, rather than impressions or ad budget.
The business value offered by Bizo is similar to DemandBase: reaching prospects that haven’t yet engaged with a company directly or retargeting them when they don’t respond. The different technical approaches have their own strengths and weaknesses: IP-based identification is relatively stable but works only at the company level and doesn’t identify small businesses that lack their own stable IP address; cookies identify individuals but are often deleted, miss some people, and result in multiple, fragmented identities for others. Like the client I mentioned at the start of this article, you can probably think of them as complementary rather than competing components of a complete B2B marketing solution.
________________________________________________________________________
* Given that the total employed U.S. workforce is about 145 million,
I suspect that 85 million contains quite a few duplicates, meaning any
one profile captures just a fragment of an individual’s activity. But
that’s the nature of this sort of thing; the business question is how
well the data works even in its imperfect state.
Showing posts with label ad agencies. Show all posts
Showing posts with label ad agencies. Show all posts
Friday, February 21, 2014
Wednesday, February 03, 2010
Clarifying the Differences Between Database and Digital Marketing
Summary: Database and digital marketing are both data-driven. But they differ in plenty of other ways that make it hard for specialists in one to adapt smoothly to the other. Here's a detailed look at the differences.
Yesterday’s long (or merely long-winded?) post described the different mindsets of database and digital marketers but it was pretty short on differences between the two marketing methods themselves. Today I’ll try to be more concrete.
DB or Not DB
Database marketing is built around a marketing database that contains addressable, identifiable individuals. By “addressable”, I mean there is information such as a mailing address or phone number that lets the marketer contact the individual. By “identifiable”, I mean information is available to link data about the same individual from multiple sources. Addresses are the most common identifiable information, although there are also non-address identifiers such as Social Security Number. Addresses and identifiers are both required: a database without addresses couldn’t be used for most marketing, and a set of records that can’t be linked to other sources is just a list.
The consolidated database is the heart of the database marketing concept. Data from multiple sources lets database marketers make more effective predictions about the best treatments for each individual, and treatments across multiple channels are more effective when they are coordinated centrally. The marketing database contains attributes (age, income, location, etc.) and behaviors (promotion responses, purchases, customer service interactions, etc.). It can certainly include digital activities such as Web page views and social media comments, so long as these can be linked back to a known individual.
Digital marketing does not use a database of addressable, identifiable individuals. It may gather information from one source and even track it over time for the same entity. (Example: Web site behavior tied to a browser cookie.) But unless the entity can be linked to other sources through an identifier, the digital marketer can only make treatment decisions based on information captured in the source channel itself. This is far from useless – behavioral and contextual targeting can be quite powerful. But from a database marketing perspective, the data is frustratingly incomplete.
Addressable Media
Database marketing only works in addressable media: that is, where a message can sent to a specific individual. Addressable media include direct mail, email, outbound telemarketing, and customer service interactions. They can also include digital channels such as Web pages, mobile messages, kiosks and ATM machines, but ONLY where the recipient is known before a message is sent. Thus, a Web page that has identified me because I’ve registered and logged in (manually or via a cookie) is addressable; a Web page that I visit anonymously, even if it recognizes me as a previous visitor from a cookie, is not addressable.
Digital marketing includes many non-addressable media, including paid and organic search, Web banner advertising, social media, and anonymous forms of Web sites, kiosks, mobile (e.g., location-based messages), and the rest. These generate plenty of useful data, such as click through rates, search rankings, sentiment analysis, and page views. But this data and related analysis are quite different from what database marketers are used to.
Prediction vs Reaction
Database marketers have the rich information needed to accurately predict which offers are most appropriate for each customer. Combined with their access to customer addresses, this allows them to initiate effective outbound marketing campaigns and to define static rules for interactive dialogs. Note that in most addressable media (mail, email, outbound telemarketing), the offer must be selected before the customer is actually contacted, and making multiple offers often reduces response. So database marketers have strong reasons to work on making highly accurate predictions.
By definition, digital marketers cannot target outbound campaigns at individuals. They do have opportunities to manage interactions, but often know only what has happened during the current interaction itself. This greatly reduces their ability to make predictions. Instead, they present multiple options and react as people respond. Happily, most digital media are inherently interactive, so this is a practical approach. Since rule-based decision flows are less viable as the number of options increases, digital marketers lean more heavily on self-adjusting automated decision engines.
Message Control
Database marketers directly control the messages they send to each customer. This is yet another factor that helps to justify the costs of building a comprehensive database, running sophisticated predictive models and precisely customizing each message.
Digital marketers have vastly less control over who sees what. Much of their messaging is blind to the audience who will see it, or can only be targeted on limited information about behavior or context. Indeed, some of the most effective and intriguing digital marketing techniques, such as viral campaigns and shareable widgets, rely on distribution that's totally beyond the marketer's control. Social media provide even less control, since the messages themselves are composed outside the company. The net result of all this is to reduce the degree of individual targeting that digital marketers can execute.
Response Measurement
Database marketers can typically capture response to a promotion directly, with a coupon, telephone call or Web click. Even when they can’t, their database still ultimately tells them who bought what, so they can correlate the promotions they’ve addressed to an individual with that individual’s subsequent behavior. The ability to do precise response measurement is yet another factor that lets database marketers fine-tune their programs.
Digital marketers can also measure who clicks on a Web ad, and sometimes can track that person further into the buying cycle. But they don’t know what other promotions or social media that person saw, what else they purchased, who else saw the same promotion but didn’t respond, or who responded through some other channel. All these uncertainties leave digital marketers reliant on indirect measures, such as consumer panels and surveys, which are more typical of conventional mass media. These are approaches that most database marketers would find almost laughably imprecise.
What’s It All Mean?
Database marketers and digital marketers both have plenty of data and the good ones are highly analytical. Both can apply advanced statistical techniques and rigorous testing methods. Both can work to integrate their data and their customer strategies across channels. To some extent, they even work with the same media: in particular, a Web site can support both digital (anonymous) and database-driven (addressable) marketing programs.
Yet despite these similarities and interactions, the two groups work in largely different media, use different techniques and have different priorities. Database marketing is inherently more controlled and precise; digital marketing is more fluid. Good marketers will learn to apply both. But individuals who have specialized in any one area will find it hard to adjust to the other. At a minimum, they’ll need to be conscious that the old rules don’t apply.
Adjustment is even harder for organizations, who will have invested in specialized systems, processes and people to support one technique or the other. This, in my opinion, is why the leading database marketing vendors have not been the leading digital marketing vendors. Which, if you’ll recall, was where I started this discussion.
One final point: there's no reason the same organization or individual can't master both database and digital marketing. That is, although there are major differences between the two, there is no fundamental conflict. My point in these articles is simply that it will take conscious effort to address the differences and fill the gaps that they imply.
Yesterday’s long (or merely long-winded?) post described the different mindsets of database and digital marketers but it was pretty short on differences between the two marketing methods themselves. Today I’ll try to be more concrete.
DB or Not DB
Database marketing is built around a marketing database that contains addressable, identifiable individuals. By “addressable”, I mean there is information such as a mailing address or phone number that lets the marketer contact the individual. By “identifiable”, I mean information is available to link data about the same individual from multiple sources. Addresses are the most common identifiable information, although there are also non-address identifiers such as Social Security Number. Addresses and identifiers are both required: a database without addresses couldn’t be used for most marketing, and a set of records that can’t be linked to other sources is just a list.
The consolidated database is the heart of the database marketing concept. Data from multiple sources lets database marketers make more effective predictions about the best treatments for each individual, and treatments across multiple channels are more effective when they are coordinated centrally. The marketing database contains attributes (age, income, location, etc.) and behaviors (promotion responses, purchases, customer service interactions, etc.). It can certainly include digital activities such as Web page views and social media comments, so long as these can be linked back to a known individual.
Digital marketing does not use a database of addressable, identifiable individuals. It may gather information from one source and even track it over time for the same entity. (Example: Web site behavior tied to a browser cookie.) But unless the entity can be linked to other sources through an identifier, the digital marketer can only make treatment decisions based on information captured in the source channel itself. This is far from useless – behavioral and contextual targeting can be quite powerful. But from a database marketing perspective, the data is frustratingly incomplete.
Addressable Media
Database marketing only works in addressable media: that is, where a message can sent to a specific individual. Addressable media include direct mail, email, outbound telemarketing, and customer service interactions. They can also include digital channels such as Web pages, mobile messages, kiosks and ATM machines, but ONLY where the recipient is known before a message is sent. Thus, a Web page that has identified me because I’ve registered and logged in (manually or via a cookie) is addressable; a Web page that I visit anonymously, even if it recognizes me as a previous visitor from a cookie, is not addressable.
Digital marketing includes many non-addressable media, including paid and organic search, Web banner advertising, social media, and anonymous forms of Web sites, kiosks, mobile (e.g., location-based messages), and the rest. These generate plenty of useful data, such as click through rates, search rankings, sentiment analysis, and page views. But this data and related analysis are quite different from what database marketers are used to.
Prediction vs Reaction
Database marketers have the rich information needed to accurately predict which offers are most appropriate for each customer. Combined with their access to customer addresses, this allows them to initiate effective outbound marketing campaigns and to define static rules for interactive dialogs. Note that in most addressable media (mail, email, outbound telemarketing), the offer must be selected before the customer is actually contacted, and making multiple offers often reduces response. So database marketers have strong reasons to work on making highly accurate predictions.
By definition, digital marketers cannot target outbound campaigns at individuals. They do have opportunities to manage interactions, but often know only what has happened during the current interaction itself. This greatly reduces their ability to make predictions. Instead, they present multiple options and react as people respond. Happily, most digital media are inherently interactive, so this is a practical approach. Since rule-based decision flows are less viable as the number of options increases, digital marketers lean more heavily on self-adjusting automated decision engines.
Message Control
Database marketers directly control the messages they send to each customer. This is yet another factor that helps to justify the costs of building a comprehensive database, running sophisticated predictive models and precisely customizing each message.
Digital marketers have vastly less control over who sees what. Much of their messaging is blind to the audience who will see it, or can only be targeted on limited information about behavior or context. Indeed, some of the most effective and intriguing digital marketing techniques, such as viral campaigns and shareable widgets, rely on distribution that's totally beyond the marketer's control. Social media provide even less control, since the messages themselves are composed outside the company. The net result of all this is to reduce the degree of individual targeting that digital marketers can execute.
Response Measurement
Database marketers can typically capture response to a promotion directly, with a coupon, telephone call or Web click. Even when they can’t, their database still ultimately tells them who bought what, so they can correlate the promotions they’ve addressed to an individual with that individual’s subsequent behavior. The ability to do precise response measurement is yet another factor that lets database marketers fine-tune their programs.
Digital marketers can also measure who clicks on a Web ad, and sometimes can track that person further into the buying cycle. But they don’t know what other promotions or social media that person saw, what else they purchased, who else saw the same promotion but didn’t respond, or who responded through some other channel. All these uncertainties leave digital marketers reliant on indirect measures, such as consumer panels and surveys, which are more typical of conventional mass media. These are approaches that most database marketers would find almost laughably imprecise.
What’s It All Mean?
Database marketers and digital marketers both have plenty of data and the good ones are highly analytical. Both can apply advanced statistical techniques and rigorous testing methods. Both can work to integrate their data and their customer strategies across channels. To some extent, they even work with the same media: in particular, a Web site can support both digital (anonymous) and database-driven (addressable) marketing programs.
Yet despite these similarities and interactions, the two groups work in largely different media, use different techniques and have different priorities. Database marketing is inherently more controlled and precise; digital marketing is more fluid. Good marketers will learn to apply both. But individuals who have specialized in any one area will find it hard to adjust to the other. At a minimum, they’ll need to be conscious that the old rules don’t apply.
Adjustment is even harder for organizations, who will have invested in specialized systems, processes and people to support one technique or the other. This, in my opinion, is why the leading database marketing vendors have not been the leading digital marketing vendors. Which, if you’ll recall, was where I started this discussion.
One final point: there's no reason the same organization or individual can't master both database and digital marketing. That is, although there are major differences between the two, there is no fundamental conflict. My point in these articles is simply that it will take conscious effort to address the differences and fill the gaps that they imply.
Subscribe to:
Posts (Atom)
