I don’t care how many coupons you send me, I am not buying your crappy pizza.

Amusing, or depressing, as the case may be, it often seems the old marketer’s adage still holds true. And although the expression may come from a time before it was fashionable to refer to things like “customer centricity,” that is the point.

In today’s reality, if you’re dealing with the challenges of operating a subscription- or relationship-based business model, your ability to compete by being customer-centric will only be proportionate to the maturity of your data strategy.

It’s been like this for a while.

Years ago, it was savvy marketers who caught on first. They realised that data and analytics could be leveraged to drive marketing communications more effectively, and direct marketing evolved into database marketing.

This sort of successful collaboration between business functions and data sciences further evolved into customer relationship marketing and, in turn, new business philosophies emerged revolving around a far deeper, more contextual understanding of the consumer relationship.

At its core, this has always been about solving a fundamental business problem: Simply blurting out an offer doesn’t work. We need something compelling to say, and we need to say it when it’s most relevant and meaningful.

The growing issue now, though, is that there are all these potential conversations happening simultaneously, across all forms of technology and devices, and in environments businesses simply do not and cannot control.

As if that wasn’t enough, consumer data has exploded. It’s bigger, far less structured, and it’s coming at us faster than ever before. It’s the essence of Big Data, and both sides of the marketing equation have become exponentially more complex.

These modern-age challenges, and the noise surrounding it all, can easily distract marketers and data scientists alike from the core problem we’re meant to solve: to be thoughtful, timely, and relevant in what we say and how we say it.

To that end, here are some things to consider when thinking about marketing communications strategy from a Big Data perspective:

Know your baselines.

• What is the average conversion rate across each of your communication vehicles?
  
  
• What is the retention/survival rate for customers across various campaigns or promotions?
  
  
• What is the offer repeat take-rate?
  
  
• Can you measure pre- and post-campaign engagement?

Having a sound understanding of your baselines will allow you to measure success or failure.

There should only be a handful of performance indicators you measure (such as the ones listed above). These metrics will tell you all you need to know about your campaign performance.

You may want to determine your baselines in cohorts as opposed to stacking time-series data sets together to account for any seasonality-based biases.

Don’t confuse the difference between who and how.

To target and communicate your marketing messages effectively, it’s essential that you understand both who to target (propensity) as well as how to target (profile). But they are two very different things.

Predictive or exploratory models are statistical methods meant to determine an individual customer’s propensity to purchase, or churn, or do something else of interest to us as a business.

These statistical models uncover patterns in behaviour that can be used to target individuals based on their probability to respond or behave a certain way (past behaviour is analysed to understand correlations with future potential behaviours).

If, for whatever reason, predictive modeling is not an option, other analytics such as micro-segmentation and multi-dimensional analytics could help uncover segments of individuals that represent the best target candidates.

Here the process essentially involves segmenting the audience based on as many attributes as possible (creating as many segments as possible) then overlaying observed behavioural metrics across each segment to determine which ones have been most effective at realising the business outcome you’re looking to drive.

These, in turn, would be top candidates for more proactive targeting.

Regardless of methodology, the ultimate and primary goal here is to determine who to target.

Profiling target segments, on the other hand, can provide the insight necessary to effectively communicate — literally speak to — these target audiences in more meaningful ways.

The critical distinction here is that we’re not looking for predictive drivers; we’re looking for insight into the individuals that make up our target groups to help inform the marketing treatment.

For example, consumer age, time of day they engage, or the content type most often consumed may not have been correlated to an individual or segment level probability to purchase, but these factors could easily and intuitively be leveraged in marketing creative and messaging.

Data miners need to be creative here. It’s not about statistical probability at this stage but about finding ways to tell a story about the human beings beneath all the math and science.

Testing and experimentation should be an obsession.

Pricing, messaging, targeting, sequencing, colour, font, call to action — literally everything — should be tested and experimented on. There are no truths unless there are results from previous experiments to back them up.

We’ve heard about best practice organisations living in a culture of “beat the control,” and it’s a worthy thing to aspire to.

Remember, though, you’re testing and learning for a reason. Learn quickly. And when you find something that works, stick to it until you find something better.

Also, make sure your experiments are either scalable or meant to derive some more specific insight. Either way, it must have purpose, driven from analytics and insight. There can be a fine line between obsessive and academic in this pursuit.

The key imperative here is lift. Data analytics teams simply cannot have an intelligent conversation about experimentation without an advanced understanding of lift and how to explain it.

By lift, I mean the quantifiable difference between the experiment and the control (baseline) and the value it represents to the business in dollars.

This is also where data and analytics teams have the opportunity to truly demonstrate their value. Have you ever struggled to determine the ROI on your Big Data programme? Here is your opportunity to demonstrate, with tangible results, the value of applying smart data and analytical insight into the marketing conversation.

Data scientists should think like marketers — and marketers should think like data scientists.

Solving marketing and communication problems through Big Data and advanced analytical techniques requires data scientists and marketers to work together.

The end game here is an effective, meaningful conversation with customers or potential customers leveraging all the data and analytical options available. Process and cross-functional workflow can sometimes stifle the creativity and contributions on both sides.

Analytics teams are always more effective when they have the context behind what they are doing and how it will be leveraged. Marketers are often surprised at how informative and inspiring conversations with analytics teams can be. Siloes here are undoubtedly drivers of inefficiency and lost opportunity, not to mention risk.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .