Customer Data Platform Guide Reviews Tools to Build Marketing Databases

Raab Associates’ new Guide to Customer Data Platforms is now available [update: as of 2014, we no longer sell it].

You may not find that news to be fall-off-your-chair exciting. In fact, you’re more likely to wonder whether the world needs yet another report on anything at all. Fair enough. So before telling you what’s in the CDP Guide, I'll tell you why it exists.

Simply put, marketers need better databases. If you’re a working marketer, you almost surely know this from personal experience. But someone who only read industry news and vendor promotions might think all anyone had to do was to plug in the latest cool application and it would immediately be filled with fresh clean data like water from a tap. Dirty big data is our industry’s dirty little secret.


The problem isn’t new but it is getting worse. As customers interact across more channels, marketers need to not just meet them in every new location but recognize them and carry on a continuous conversation from one touchpoint to the next. Marketers can also become more effective by enriching that conversation with information from external sources such as Web pages, social media, and commercial databases. Both the carrot of better results and the stick of customer expectations are ever-more-urgently driving marketers towards building better databases.

The good news is that plenty of smart vendors have also recognized this need and are trying to help marketers on their journey. I call their systems Customer Data Platforms and define them as “a marketer-controlled system that supports external marketing execution based on persistent, cross-channel customer data.”

If there’s one absolutely critical point in that definition, it’s that CDPs put marketers in charge of building their own database. Taking control is the only way that marketers will ever get the databases they desperately need. It’s why CDPs are so important.

But too few marketers know who the CDP vendors are, what they do, and how they differ. The Guide to Customer Data Platforms is designed to provide this information. If the CDP vendors are tour guides on the path to better data, the CDP Guide is the reviews you read to decide which one you’ll hire. As far as we know, no other study serves this purpose.

Given its goal, the heart of the Guide is the vendor profiles: three to five pages on each vendor, describing capabilities for data management, predictive modeling, marketing campaigns, and message delivery, plus background on the vendor’s technology, clients, company history, and pricing. You’ll want to read those closely when you’re selecting a vendor. But first you’ll have to decide whether a Customer Data Platform is something to consider. Here is some information to help make that judgment.

- CDPs are something new. CDPs are systems that help marketers build and update customer databases, and make those databases available to support marketing programs. That may not sound very new, but most B2B marketing automation products today build very limited databases while most B2C marketing automation products rely entirely on an external data warehouse. The systems that do build databases are designed to be used by IT departments, not marketers. And many CDPs provide predictive modeling or best-treatment recommendations that go well beyond the storage functions of a basic data warehouse.

- You still can’t do this at home. CDPs may be tools for marketers, but that doesn’t mean that marketers build the databases themselves. Rather, CDP vendors provide services that build the database with varying degrees of marketer involvement. The difference is that the marketers work directly with the CDP vendors, instead of relying on IT staff that often has other priorities and an imperfect understanding of marketing needs. This makes it much easier and quicker for marketers to get the database they need.

- CDPs are an outgrowth of existing system types. Most CDP systems were created for a purpose that happened to require the same database-building capabilities as a CDP. These purposes fall into three groups which I discussed in last week’s post, so I won’t repeat them here. They’re worth understanding because vendors in each group have a different set of skills, one of which will probably come closest to your needs.

- Convergence is coming. Even though the CDP vendors started with different applications, their shared abilities for identity matching, database management, analytics, and integration will allow them to support more of the same functions over time. As marketers understand the value of their databases more clearly, CDP vendors will be able to focus on selling their data platform features rather than applications the platform supports. Of course, once the platforms themselves are common, vendors will climb the value chain by offering better predictive analytics and cross-channel treatment optimization.

- Details count. CDP features may eventually converge, but for now the systems differ in many small ways that make a big difference. To take one example, nearly every CDP creates predictive models. But some can only predict response to specific promotions, based on who has responded before. Others can do the much more sophisticated analysis needed to predict which offer will best advance a long-term goal such as becoming a new customer. And even among those that model against long-term goals, some can actually estimate the incremental impact of a specific offer and others can just see most common correlations. We found similarly subtle differences in how data is collected (via the vendor’s own Web tags or by importing from other systems), the range of data sources (just marketing automation and CRM or those plus many others), natural language processing to extract useful information from text sources such as Web pages, how much history is kept and how it’s used, program execution, and end-user control. The CDP Guide clarifies these distinctions, but it’s still up to marketers to evaluate which differences will matter in their own business.

The CDP Guide itself contains quite a bit of other useful information, including a formal definition of CDPs, detailed explanations of what to look for in a CDP, and a history of marketing databases starting with ancient Sumer (don’t worry, I skipped the boring parts). Again, the goal is to provide one package with everything you need to get started along the path of buying a CDP system.  From there, it's up to you.

New Study: Three Types of Customer Data Platform Address Cross-Channel Marketing Needs

My detailed study of Customer Data Platforms should be released next week. Now that the information is assembled, I can at last pull back and get a good overview of what I’ve found.

Perhaps the most interesting discovery has been that the CDP vendors cluster into three main groups.

• B2B data enhancement. These build a large reference database of companies and employees, which they match against records imported from their clients. They generally return corrected and enhanced data and lead scores based on models built from the client’s customer files. Their reference databases are built from multiple public, commercial, and proprietary sources, and are assembled using sophisticated matching engines. Most also perform their own scans of Web sites and social networks to extract sales-relevant information such as technology use and changes that suggest buying opportunities. These vendors vary considerably in the data they return, ranging from lead scores only to recommended marketing treatments to full customer profiles. Some also provide prospect lists of companies that are not already in the client’s own database. CDP vendors in this group include Infer, Lattice Engines, Mintigo, and ReachForce.

These systems compete with non-CDP products which also add or enhance prospect records but do not maintain a database with their clients’ customers. These include Web scanning systems such as InsideView, LeadSpace, and SalesLoft, and general data compilers including NetProspex, Demandbase, Data.com, ZoomInfo, and OneSource. The predictive modeling features also compete to some degree with end-user-oriented marketing analytics and modeling software such as Birst, GoodData, Cloud9 Analytics, AutoBox, and Predixion Software. Data cleansing competitors include services from firms such as D&B, as well as data management software for technical users such as Informatica, Experian QAS, and FullContact.

• Campaigns. These systems build a multi-source marketing database from the client’s own data and either recommend marketing treatments to execution systems or execute marketing campaigns directly. These are primarily used for consumer marketing although they also have B2B clients. Most have sophisticated matching capabilities. This group includes Silverpop with its Universal Behavior feature, NICE’s Causata, AgilOne, and RedPoint.

This group competes with conventional consumer marketing automation products, which provide similar campaign management abilities but lack the CDPs' database flexibility, database management, and customer matching features.

• Audience management. These systems build a database of customers and their responses to online display advertisements. They then build models that predict the customers’ probability of responding to future advertisements and provide recommendations for how much to bid and which content to display. These systems perform the same basic functions as standard online audience management systems (Data Management Platforms, or DMPs) and provide the same very quick responses needed for real time bidding (usually under 100 milliseconds). The major difference is that they also recommend messages in other channels, such as Web site personalization or email campaigns. Like DMPs, they work primarily at the Web cookie level, can link cookies known to relate to the same customer, and can be linked to actual customer names and addresses in external systems. This group includes IgnitionOne, [x+1], and Knotice.

This group overlaps with recommendation and ad targeting engines and DMP systems. Those products provide similar functions but do not track identified individuals and are often limited to single channel executions.

Given that each group addresses a different business need, you might wonder why I think they should all be lumped together under the CDP label. Quite simply, it’s because they are all addressing a portion of the same larger problem, which is how marketers can get a complete view of their customers and use that view to coordinate treatments across channels. What marketers truly need is a combination of the features from each group: data enhancement from external sources, for consumers as well as B2B; sophisticated customer matching and treatment selection; and integration of online advertising audiences with traditional customer databases. Each of these systems has the potential to grow into a complete solution, and the normal dynamics of software industry growth will push them towards pursuing that potential. So I expect the categories to overlap increasingly over the next few years and eventually merge into complete Customer Data Platforms as I envision them.

Incidentally and tangentially related: I'll be giving a Webinar with ReachForce on October 2 on Data Quality for Hipsters, a name that started as a joke but does make the point that data quality is essential for cutting-edge marketing.  YOLO, so you might as well attend.  I'm already working on the mustache.

How Raab Associates Converted to ZohoCRM In One Weekend: a B2B CRM Success Story

Raab Associates is really two businesses: the technology consulting practice run by Yours Truly, and a marketing agency specializing in children’s books run by my beautiful and brilliant wife Susan. We keep them largely separate, but I am inevitably involved in her technology decisions. So when her ancient Goldmine CRM system finally crashed last week, we both scrambled to pick a replacement.

From my usual lofty perch in enterprise software world, Susan's requirements seem stick-figure simple: accounts, contacts, opportunities, lists, and mass emails. So our first thought was to find a system that offered those plus some cool new things like social media profiling. But a quick scan of the market showed that none of the neat new systems also offered the basic functions with with enough refinement and flexibility to meet Susan's needs.

This pushed us back to the more standard CRM options.  To my dismay, we found ourselves ruling out one after another for various. I even briefly suggested we reconsider Goldmine, an thought that was quickly rejected.  Eventually we took an unhopeful look at ZohoCRM, which I know as a popular small business system but had never considered particularly advanced. Happily, the system has a very thorough online user manual, so I was able to check it out in detail.

Even more happily, the answers all came back positive as I imagined working through Susan’s basic business processes in Zoho. Build contact lists, check. Mass emails, check. Opportunities linked to campaigns, check. Pull-down status list and callback date on opportunities, check. Custom filters across all field types, check. End-user report writer, check. Multi-field search, check. A bunch of other details that I no longer recall, check check check. Reasonable cost, double check: we would have grudgingly paid a couple hundred dollars a month for a solution, but Zoho’s mid-tier Professional edition costs all of $20 per month with no limits on database size (Susan has about 14,000 contact records – well above the minimum for many small business systems). We may even splurge for $35 per month enterprise edition, which provides some advanced automation features but is probably overkill for most small businesses.  Just call me Diamond Jim.

At this point, we were ready to sign up for the free trial account, which was a simple process and didn’t ask for a credit card. Let me point out that I purposely hadn’t signed up sooner because I didn’t want to waste time exploring a system that I wasn’t pretty confident would meet my needs. Diving in too soon is a classic mistake among software buyers – and, in this instance at least, I actually followed my own advice.  (While I'm patting myself on the back, I'll also point out that we evaluated the software against our actual business process, not an arbitrary feature checklist.  That's another best practice that too few buyers follow.)

We now pulled a small set of test records from Goldmine to test the import function. The online manual guided me through the exact steps necessary, complete with a handy checklist of preparatory tasks.  When I went to load the file itself, I got the first of many delightful surprises: Zoho took a guess at mapping the input fields, based on their names, and got about half right. That’s a pretty sophisticated function and a big time-saver. It’s the sort of refinement you don’t see in a new system because it’s not essential to get the product into market, but gets added after enough users request it and the developers have some breathing room. Zoho has actually been around since 1996 (although CRM came later), so they’ve had time to add a lot of those little helpers.

In any event, the test import worked perfectly the first time out, which was a great feeling of accomplishment. Susan and I played with the system a bit more now that we had some real data in it, and found all sorts of nice little options, like being able to rename objects (she calls an opportunity a “pending record”), rearrange the fields on each screen, change the order of sections, and move fields from one section to another.  Again, none of these is cutting edge, but they’re not always available and make a big difference in making the system more usable.  The interface itself was also highly intuitive – lots of nice dragging to move the fields around, for example. There were plenty of other unexpected goodies that I would have otherwise needed to configure or live without, like automatically listing the associated contacts when you view an account record, and listing the associated opportunities – I mean, pending records – when you look at a contact. And, oh yes, you can control which fields are displayed on those related records.

At this point we were feeling pretty good about actually pulling off the conversion, so I spent all day Sunday manually cleansing those 14,000 contact records to ensure the critical data was populated. Even Zoho couldn’t help with that one. I finished around midnight and had a moment of panic when I saw that Zoho would only import 5,000 records at a time.  But it turned out to accept all three batches without waiting for the first batch to finish, so I was able to submit them and get some sleep.

I woke up bright and early (well, actually, late and cranky), feeling pleased that Susan could start using the system without missing a business day.  Alas, we found that somehow there were twice as many account records as expected. A quick call to Zoho support pointed us to a rollback function that should have cleaned up the problem in a few seconds. Sadly, it rolled back one set of records but not the other (remember, there had only been one import).  I spoke again with Zoho support, who promised to look into it but hadn’t accomplished anything several hours later.  At that point, I realized – duh – that it would take about two minutes to delete the records manually (you can only delete 100 at a time, but it’s three keystrokes for each batch, so you can probably do about 50 batches per minute). Once I figured that out, I cleaned out the old records and reimported everything, and we had a clean set of data.

Susan has been working with the system for the past two days, and I’ve been peeking over her shoulder and poking around a bit myself.  ZohoCRM is certainly not perfect – there are bunch of little things she would like to do, such as preview a template-based email with the variables populated. There are also some oddities like two unrelated sets of email templates, a vestige of Zoho's earlier separate systems for CRM and mass mailings. Those quirks take a bit of getting used to but are far from show-stoppers. There are some other tasks that cumbersome at the moment, but I suspect we’ll be able to automate once we have time to explore those functions. And, yes, there are some things it doesn’t do that Susan would like, such as associating multiple email addresses with the same contact. I wouldn’t exactly say they’re trivial – certainly not to Susan – but she can live with them.

We're generally satisfied with customer support: phone calls aren’t always answered immediately, but after about a minute on hold, a very nice lady picks up the line and offers to take a message. I appreciate the human touch, and, more important, the opportunity to get immediate help if something is truly urgent. We do get callbacks in an hour or two and the agents have been pleasant and helpful, which is about all I can ask. There’s a “how’d we do?” email after each interaction, which is a good sign that Zoho is trying to do a good job.

Bottom line: We’re still in the honeymoon period, so I may find Zoho isn’t really as great as I think.  On the other hand, I proposed to Susan almost immediately after meeting her and that's worked out just fine.  So I'd say ZohoCRM is worth a close look for small business CRM, even for people who think it may be too simple for their needs.