A few weeks ago, in the middle of February, I reviewed some analytics dilemmas from 2005 in a column entitled “Analytics Problems Won’t Die.”
I laughed at some, shrugged off some, and commented on the really nasty ones in light of current events. It’s time to revisit that list in light of the even more current state of affairs.
What makes this week more current than five weeks ago?
At the beginning of March, the eMetrics Marketing Optimization Summit was held in San Francisco and hundreds of people sat at round tables to discuss their contemporary “5 Top Problems in Marketing Analytics.”
Encouraged to tweet their frustrations, the participants came up with quite a list that boiled down to the following:
1. Data is much easier to spell than to use. Now that digital analysts are responsible for conjuring up deep, valuable insights out of a steady stream of amazingly disparate data sources, we’ve come face to face with the complexity of the task.
Collecting data, getting it all in one place, and matching it up with data from an alternative universe is hard work. This was described in 2005 as the problem of: data proliferation: multiple sources, reports, and integration.
The additional insult to this injury is the astonishing proliferation of data types (the data deluge) and the inflated effort necessary to transform them into a cohesive base. Data from social, mobile, video, in-store, in-vehicle, inside-your-head-via-fMRI machines are all required and more is on the way tomorrow afternoon.
My comment about this last month has not changed:
There is no end to data. There will always be a new data stream around the corner that cannot be anticipated. It has been ever thus.
2. Nobody here gets it. Previously described as political issues: management buy-in; IT vs. business unit objectives; resources and reporting vs. analysis, this is also a persistent problem.
This time around, the language included:
Lack of culture
Lack of resources
Lack of education
Lack of skilled/trained people
Lack of insight communication skills
All of the above point to the problem that spawned the first eMetrics Marketing Optimization Summit: “I can’t convince enough people to take this seriously enough to get the proper resources to make it happen.”
The solution is found in fixing the first item on that list: lack of culture.
Without a serious data-loving culture, you are simply better off moving to another company. That, or become the change agent who is willing to do everything including losing your job in order to make your company see the light.
Aside to would-be change agents: we often feel that our ideas will not be welcome – that we are not worthy. But imagine yourself in that seat of power. If somebody comes to you with what they perceive is a serious problem along with some very serious enthusiasm that cannot be squelched then, yes, they are a pest. But if they come to you with a serious problem and some serious enthusiasm along with a fully-baked plan for solving it then you would welcome them with open arms. Be not afraid. Be irrefutable. Be bold.
3. Silos, silos, silos. As before: silos: multiple resources going in different directions. This is never, never, never going to be solved as long as humans are involved. But if you do come across some Jedi mind trick that gets everybody on the same bus, singing “Kumbaya” in the same key, please let us all know at once!
4. (New one) attribution. Ah, yes – the nightmare of trying to attribute the correct percentage of success to the myriad methods of grabbing somebody’s attention. How much credit should banner ads get in the subsequent sale? Does that generic search term deserve more credit (and a bigger budget next time around) than the branded keywords? How about posters on buses?
Back in 2005, this was a topic for after-hours conversations, not the stuff of serious business priorities. Today, with e-commerce a certainty, the idea of attribution has taken root as a way to optimize promotional and advertising spend.
Practitioners, however, are starting to talk about this one as an illusion of marketecture – data integration at the brochure level – rather than a must-have. Case studies are hard to come by. Any easily-adopted frameworks are hiding under bushel baskets somewhere.
Consultants are striving mightily to find a solution, write a book, and retire. Software and systems companies are claiming that they have solutions, but all they really have is semantics.
Only a few are willing and able to take the time and spend the money on rigorous testing, rigid influence isolation, and extraordinarily complex value modeling to move attribution beyond the realm of enigma.
And even if you do all that, you are still faced with human politics. “Of course my promotional method is better than his – it’s obvious!”
5. Hodgepodge of problems. This is a clear cheat on my behalf. Each of these got mentions with none taking the prize of being listed in sufficient quantity to beat out the rest.
Generating Quality Hypotheses
What’s the art of asking really good questions? The short answer is:
A truly deep and abiding knowledge of problems to be solved.
A truly deep and abiding knowledge of available data.
A truly deep and abiding talent for communication.
A truly deep and abiding passion for advancing the art.
The rest is just trial, error, blood, sweat, tears – repeat.
Personally identifiable information is still being defined. Your Twitter account can be connected to your LinkedIn account, which can be related to your Facebook activity – naturally. And it’s no trouble at all to link all of the above to your log-on ID. But as soon as we join that up with your pre-log-in cookies and your content consumption habits on third-party sites, then the trouble begins.
Word of warning: tread very carefully lest legislation rear its ugly head and dictate your data management methods. A policy a day keeps the government away.
Understanding Visitor Intent
This one threw me for a loop. Isn’t this the goal of the entire industry?
This is what we are trying to establish. This is why we collect every search term, ad click, mouse hover, blog sentiment, shopping basket ingredient, and recency/frequency metric and match them with demographics, psychographics, credit scores, magazine subscriptions, catalog activity, and whether or not you floss.
The ultimate goal is to figure out what the heck you want. This is not a problem. This is the goal.
And it is not likely to ever, ever, ever be solved.
But if we can get that much closer to it than the competition, thar be gold in them thar hills.
The use of psychology in marketing and sales is not new, but it may be more useful than ever in an attention economy where time is precious and focus is rare. How can you tap into a demanding consumer to check whether there is an actual interest in your product?
According to a survey conducted as part of OnBrand Magazine's State of Branding Report 2017, marketers are well aware of the new technologies that are expected to be important to their brands in coming years, but the majority aren't rushing to invest in them before they're fully-baked.
Two weeks ago, Foursquare announced what could be the most important component of its data business: the Pilgrim SDK. So what does it do, and what does it mean for location-based marketing?
Combining clickstream data with machine-learning technology, behavioral analytics helps enterprises create a tailored online experience for each visitor to their web or mobile sites.