"Marketing analytics can have substantial impact on a company’s growth. But if companies cannot figure out how to make the best use of it, in the end, it’s just another expense."–McKinsey
As most of you know by now, I'm particularly interested in helping companies create indispensable data and analytics products–you know, the kind that actually save time, solve problems, or deliver insights with minimal tooling and effort. One of the topics I try to follow is how well (or not) all of this data out there is actually helping inform consumers and businesses on a practical level.
This article from HBC/McKinsey on marketing analytics was interesting. In this survey about the adoption of marketing analytics within companies, "only 10% of respondents believed they were very effective at feeding insights into customer behaviors back into the organization to improve performance."
It makes you wonder: is the problem the quality of the information, or is the problem in the people involved with feeding the insights?
My guess is that not infrequently, the problem is either that we have data pretending to be information, or we haven't designed that information to be useful (or served at the right time in the right context). The value is perceived as low, given the cost to put the recommendations in play.
This stems in part from the fact that a second finding in the survey article said that "developing and delivering both analysis and insights require an often complex series of hand-offs and communications from business users at one end to hardcore data crunchers on the other. That leaves ample room for miscommunication, misunderstanding, and wasted effort. Companies should consider hiring analytics translators to bridge the gaps. These are people who have enough knowledge across at least two functional areas so they can communicate effectively between them both. Analytics consultant, for example, is a role that is specifically tasked to help bridge the gap between analytics professionals and business decision makers."
It seems possible to me that these findings might also apply in the realm of B2B and B2C software data products. However, I am not sure you need an "analytics" translator. You might just need some design strategy, and coming up with one might be something you can do yourself if you have the time:
- First, you need to think about the end results you want to deliver, how the people involved are going to use and apply the findings, and what the possible value is that you can extract within the bits and bytes. Work backwards by talking one-on-one with the real customers/stakeholders, and then interweave technical and time constraints to figure out what is feasible to deliver. I like to talk to both the stakeholders who will consume and utilize the information (your software's customers, or your internal clients) and the people who know "what's possible" within the data that is available. Since the latter is usually somebody with an engineering background, I usually seek out an empathetic, business-minded software architect or principal who is able to put aside implementation details and help imagine "what would amazing look like?"
- Second, sketch some mockups or prototype the product/service before you invest in technology and code. You run those by your stakeholders, and technical people, and all, "Does this provide value? Is it possible to create? How would you use this?" Make changes. Iterate. You may indeed need some help with designing software interfaces at a pixel/visual level, but not every problem that is solvable with data and analytics necessitates a high-fidelity user interface. (Admittedly, many do, and remember that aesthetics have a lot to do with perceived value.)
- When you find enough practical value you can deliver to your customer, start to implement.
Sometimes we UX folk call this upstream process "design thinking," but forget the buzz words. The skills and process are what matter, and these steps might be all you need to avoid the results found in this particular survey. While I'm sure there are technical problems that are solved by an "analytics translator," you have to discover what is needed, usable, and desirable first.
Save yourself time and money, and get the design right before you start talking about libraries, analytics tools, data lakes, machine learning, and AI. By the time I publish this, there will be another medium to add to that last sentence, but alone, it's not going to save your product or service.