“Big Data’ is poised to become the next buzz meme in digital marketing. And along with it, a new, hybrid discipline that combines data for better marketing ROI.

Now that Facebook’s stock has hit an all-time low (never having “popped” in the first place); and with some pundits calling Zynga a “fad”, the marketers who sell to marketers need the next shiny new object to polish and put into the shop window.

And it may well be “Big Data” with its attendant disciplines.

Certainly Moore’s Law (predicting accurately that we enjoy exponential computer productivity and storage gains every few months) has given us the ability to collect more data in a year than (I postulate) had been gathered in all human history until the year 2000 (in any case some very, very large set of data). And certainly marketers, always squeezed and underfunded, hunger for more and more insight in order to drive better ROI.

But Big Data is what exactly?

It really is a direct descendant of the rather venerable discipline called Business Intelligence, now repositioned for marketers and digital analysts. It represents a convergence of data sources with a presentation layer so that a “single view” of the customer base can be obtained via on line and off-line behavioral databases.

It is becoming somewhat easier these days to deploy the combination of technologies needed to make this work.

Cloud computing, where an almost unthinkable number of motherboards are churning an almost unthinkable amount of data in an almost vanishingly small amount of time, allows almost anyone with a dollar and a dream to tap into prohibitive computing power on a subscription model! The old freezer plan never offered so much meat for so little money.

Second and third generation analytics engines, combined with sophisticated data regularization techniques and ever-swifter visual interfaces are beginning to bring Big Data into the realm of the every-day, rather than the lab-coated confines of think-tanks and black ops.

There’s also the urge for differentiation. Web analytics has become digital analytics which probably has a short shelf life until we all agree that what we are really about is (wait for it): “Convergence Analytics”.

Big Data is the set of information that will power Convergence Analytics. Convergence Analytics tools will pull data from multiple sources (like BI tools) and put them together in marketer-friendly formats. They will be hard to build and easy to use. They will address every slice of the market and each will claim their own differentiator. Much as with digital analytics, the difference will probably be not so much in the claims or even the architecture, but in the interface and the customization services that go along with that.

Big Data and Convergence Analytics  will require extensive customization and implementation, even more than digital analytics tools. This is because they draw from disparate sources and must drive insight from a much more complex data landscape.

Just when folks were beginning to think we had this web analytics thing pretty well in hand, along comes Big Data and Convergence Analytics to spoil the party. But they crashed the party with invitations to an even bigger and more raucous party down the block. Soon, you won’t have to be especially acute of hearing to feel the thundering bass from the dance floor of that shindig.

Justice Potter Stewart, opining in “Jacobellis v. Ohio”,  famously said of obscenity that he could not define it but that “I know it when I see it”.

This trope may be helpful in making many determinations, but we are hoping here to make success in web analytics a bit more easy to define than that.

I think we know what failure looks like: lack of adoption, confusion, mistrust of the numbers, blaming the tool and a reversion to the age-old and rather pointless discussion about “hits” and “page views”.

Success looks very different, and here is how it does:

1. You have a strategy.

You can foresee your rollout schedule, your players and your business goals. You have identified your analyst and her role, your content agency and their role, your analytics agency and their role, and your IT department and their role; you have established rapport and touchpoints between the parties. You know what you want to get out of analytics. You have decided what your KPIs are and you know that the people you work with know how to translate those into actual reports and dashboards. You have stamped out distractions from lightweights who say “we can do it all ourselves” or “I once used Google at another job” or “let’s hold off until after we launch [we're too busy]“. You have identified your third party campaigns and made it clear those need to be tracked independently–not relying on the data coming back to you from the ad network.  And you have made it clear you are focused on accurate reporting for business purposes.

2. You pay close attention to implementation

While some fall back on blaming tools or site architecture, you have instead understood that no tool is any better than its implementation. You have chosen from among the industry leading vendors (or free tool providers); then you have  found a company that really, truly knows how to get that tool to perform for you. It may not be the vendor and it’s probably not a company that does four other things besides. It’s probably a specialist in the field. And they will have provided you with a way to turn business requirements into reporting designs. And then have built a page-tagging structure that will deliver data to those reports. They will also have worked extensively with your developers–agency, IT or whomever–to make sure the tags (javascript) have been placed correctly on the page; and that the tags are placing calls to the analysis engine in a predictable manner.

You have also made certain that the reporting layer of your analytics tool is accessible and intelligible. If you are truly successful, you will have also trained your internal clients enough so that they know what they are looking at, and can readily access reports that are important to them.

3. You use the data

Not only have you performed all the planning and technical work, you have also made sure you convene business owners within the organization to make decisions based on the data. Your organization knows analytics isn’t just “informational” but “actionable”. If a certain campaign is doing poorly, you will change it or end it. If people are accessing pages your didn’t expect them to, you will make those pages perform for you. You don’t accept excuses from your creative team that the “impact can’t be measured” because you have already measured its impact against your business goals. And if you have found the content wanting, you know how to get it changed.

The above is what analytics looks like when it’s really doing what it is supposed to do in the organization. It’s planned well, it’s executed well, and its implications are clearly understood and accepted by those who need to do so.

If you don’t see these characteristics in your organization, start with some research into who can help get your analytics program back on track and headed in the right direction. If you do see all these characteristics, and you are not very secretive, you may want to start speaking publicly about your success, as, likely, it is rare.

A recent Pew survey found that 68% of respondents “would not be okay” with targeted advertising, citing it as an objectionable invasion of privacy. A recent GigaOm article said that the number of users deleting cookies has nearly doubled in the past year.

Signs are growing that users are getting a little bit freaked out by the sense they are being watched. And for the analytics industry, this has to be an inflection moment.

For “big data” it may be even more of a problem, because the goal of big data is to try and know everything about the customer–but let’s stick to the more common form of digital analytics and see if this holds any lessons.

Naturally I think it does.

I think its about not chasing rainbows.

I think its about going back to understanding user trending behavior and improving your site based on observed behavior. Even with high cookie-deletion rate, you can still get a sense of what repeat visitors are looking at. Fact is, even small samples will reveal important patterns. What pages do they keep coming back to? How did they find you? What pages did they leave from? Where did they leave the funnel?

Microtargeting has two problems as I see it. One is, it alienates the consumer (or so a number of consumers say). Two is, it really isn’t all that efficient, at least not on the web. Consider that I know someone who has gotten “targeted” ads for braces, lingerie, boxing, and the Mittster. Consider also that the only reaction these ads got were howls of laughter. Wonderful!

The “dream” scenario where advertising is targeted to the individual  in space, time and temperament really may turn out to be a nightmare after all. People may not like it so much–in fact, people may actively loathe it. And, its going to be really, really costly to get anywhere close to non-hilarious targeting as suggested above.

Of course it’s possible that the ROI may be justifiable due to the fact that for those users who are okay with it, the return may be massive. But that is pure conjecture at this point.

What I see is a rising tide of consumer resentment–not so much against being tracked, but being targeted.

Tracking and learning from user behavior is only going to be more important. We still haven’t gotten nearly as good at that as we think we have, and there’s lots of room for improvement before we’ve exhausted its possibilities.

Targeting may be a bridge too far. And a bridge unnecessary.

Broad targeting works quite well it seems–or has display advertising really been a bust for about a hundred years?

What is the ROI on big data? It might be a minus: expending lots of time and dollars, only to annoy your customer.

Or, targeting needs to get so good, and so sophisticated, that it stops being annoying. And who knows how much that might cost?

Is Digital Analytics more like a Deli counter or more like Whole Foods?

The paradigm is an elegant one: by properly setting up profiles and dashboards for different constituents in the organization, digital analytics data is readily available in a consumable form to anyone who gets credentials to do so.

It’s streamlined, efficient and always current on data.

It should be the undisputed standard of web analytics delivery.

So howcome it’s about as common as a sunny day in Portlandia?

The answer is: people either don’t really want it, or don’t have the time and money to set it up.

My suspicion is the real driver here is that people don’t really want it. They don’t want to be told that their analytics is all right there waiting for them. Largely this is because, without sufficient training in analytics data, methods, and statistics, most marketers and executives will look at their dashboards and say “meh” because they are not by trade equipped with the knowledge base to drill down and find the important nuggets of information.

That’s why the prevalent mode of analytics delivery is the one-report-at-a-time, when-I-need-to-see-it-only paradigm. There really is no problem with this, except that it is, on the face of it, inefficient.

But is it really less efficient than an efficiency engine?

People continue to drive decisions. And when they need to make decisions, they need information. And when they need information, they want to go to an expert and have that expert prepare information so they can look at it and rapidly see the relevant patterns. It requires specific analytics technical and business experts to be involved; as well as data analysts.

The self-service model does not necessarily provide this. It assumes the tool is able to create a display layer good enough to answer nearly every question a user might need answered. And while that’s a great idea, the tool has not yet been created that actually does it. In fact, far from it. And think of all the money that got spent putting together the amazing self-service engine, only to see that money wasted because folks don’t particularly want to use it.

What they really want to do is call “the analtyics person” and say “I need this report”. And then they really want to get that report with commentary and context.

That report would not be possible without a well-engineered and business-aligned analytics platform. But it can certainly be accomplished without anything like a self-service environment.

Self-service is, it seems, too risky in a couple of ways: first, you might not get the environment right. Second, you can’t foresee emerging requirements and will have to supply ad hoc reporting anyway.

I think we have established what is wrong here: vendors would love for the tool to do all the work. But in the real world, the tool is just that: it needs to be in the hands of a skilled expert. Much the way cars don’t yet drive down the street and go pick up groceries on their own, neither does analytics run on auto-pilot. All you need is a few random requests that require your experts to scramble for information, and the “savings” you wanted to realize by building everything into the environment starts to unravel.

My guess is that the current system of tool/business/technical/analyst/consumer will persist. This is because analytics consumers want it that way (they want to call a person and get what they want); and because no automated system has yet become expert enough to anticipate the unknown.