It sounds like the kind of marketing metaphor that sells products: You can casually dump your unstructured, unfiltered customer data into a kind of reservoir, called a data lake, and your analytics service can perform the magic of making sense of it all in an instant, at the moment you need it to.

There’s a “magic wand” aspect of this metaphor that may appeal to the marketing instinct in all of us. And with the added benefit of a lake, can the sword Excalibur be far behind?

It’s one of a handful of fundamentally different approaches to the task of managing the rapid influx of customer data, made feasible by cloud dynamics. The marketing data lake (MDL) is being marketed by Informatica, the integration services provider, to marketers rather than to data analysts or software developers.

As a result, as CMSWire has reported recently, a survey shows one marketing professional in five has yet to even hear about it.

“Marketers have ever more applications at their fingertips,” said Franz Aman, Informatica’s senior vice president of marketing, speaking with CMSWire. “And they all have individual analysis capabilities — reporting, dashboards and what have you — that by themselves do a reasonably good job telling you what goes on in that particular marketing silo, both from an app and an organization perspective.”

Software Defines Marketing

We’ve talked before in CMSWire about marketing jobs whose functions have evolved mainly to support the software used by the people in those jobs. We sometimes talk about Salesforce, Marketo and Google Analytics as though they are skills in themselves, on a par with market analysis, audience research and customer awareness.

It’s this unwritten contract between organizations and their technology suppliers that has helped traditional data warehouses maintain their beachheads.

Aman provided this example: When a digital media team becomes bound to key performance indicators (KPIs) as defined by the software they use, their goals become tied to the bottom-line numbers: increase volume of leads, reduce cost-per-lead. The efficiency of software in not only calculating those numbers, but defining them, leads to software defining the practices of marketing, rather than vice versa.

“CPL down, good; CPL up, bad. And that, by itself for that digital media team, may sound like a perfect definition of KPIs in their world,” he continued.

“But if you now look at end-to-end in marketing, where leads are turning into revenue, the answer may be something very different. If you get leads that are too inexpensive, you may not get the quality that converts into revenue.”

The need to drive down costs compels marketing departments to extend the service life of the software they use. That, in turn, prolongs the databases that drive that software — databases based on schemas that pre-date the era of social media.

It’s this calcification of marketing practices that forms the substance from which silos are built.

The Customer as an Account

Collapsing those silos, in Aman’s view, may require a similarly organic approach — as he suggests, a kind of collaboration between marketing and sales that erodes those barriers through both personal interaction and data sharing.

In an e-book published by Informatica and co-authored by Anish Jariwala [registration required], Aman tells the story of his own company’s efforts to tear away the silo walls between its marketing and sales divisions, so that both could make proper use of the same data stores. He wasn’t afraid to name names, praising Marketo for having built a more robust, flexible database that was easy enough to integrate with Salesforce, in his opinion.

Aman perceives his methodology for integrating departments as a “big data approach” — as an opportunity made feasible through technology. Specifically, the more access that marketing may be given to the data surrounding sales, including the specific transactions that take place in individual accounts, the more that marketing will be able to predict opportunities to target products and services to the accounts they’ve already accumulated.

“The interactions between sales and customers,” said Aman, “are good predictors to whether something is going to turn into revenue or not. The predictive model we now have for what turns into revenue is more accurate than the forecasts I’m getting from sales.”

The methodology Aman’s speaking about is account-based marketing (ABM). While it sounds like something that’s already been in existence, in practice, it’s the use of past and present accounts (structured data, by definition, stored in databases with refined schemas) as input variables for analytics and research (often unstructured data, stored in constructs such as data lakes).

Marketing Data Lakes to the Rescue

Such a methodology, Aman argues, is only achievable through both collaboration on the person-to-person side and integration on the machine side.

The analogy he uses involves what he calls a “California closet” — a pristine arrangement of goods where everything is in its place (unlike, I interjected, an “Indiana closet”). If such a closet suddenly needed to store a surfboard, the homeowner might have to call a carpenter.

Likewise, simple integrations between data stores and databases tend to require re-architecture. But in an environment where, for instance, it would be nice for a marketer picking up account data from sales to attach the Marketo ID for that customer to the record for next time, to put it mildly, there isn’t always a carpenter on hand when you need one.

Aman’s point is that an MDL approach, from the very beginning, eliminates the issue. But by calling it a “marketing data lake,” don’t we run the risk of constructing a shiny, new silo on the remains of the old one?

“I think marketing is a good fit for a couple of reasons,” he responded. “One is, marketing does deal with a lot of semi-structured data, [which] is a pain to move into a traditional, California closet environment.

“Marketing is one of the organizations that will benefit the quickest from the approach. And you get agility that you just don’t have with more traditional approaches. So going to that data lake model gives you so much more flexibility and experimentation capability. Marketers, who ask new questions all the time, will probably benefit the fastest from that new approach.”

For More Information: