A Data Scientist’s Wish List
I believe that marketing has become the most technology-dependent function in business since 2014. And if marketing is now so dependent on technology, then it makes sense that technology should be given a larger share of the marketing budget.
That being the case, let’s consider what a data scientist’s wish list might look like for the years ahead.
When I think about the common obstacles analysts struggle with that could be addressed with some additional spending, there are three data collection challenges that come to mind. These include:
- Point-of-sale data
- Getting a salesforce to record any data at all
- Untangling the overlap of digital data within a single transaction record
Merriam-Webster defines a wish list as “a list of desired, but often realistically unobtainable items
<a wish list of useful changes>.” The data scientist in me focuses a lot more on “useful” changes than “realistically obtainable” changes.I think we can address all three of the data collection challenges with one technology budget line: voice recognition technology — which is already very, very real. Just ask Alexa!
Point-of-Sale Data Challenge
When collecting data from a sale taking place at a cash register, the main constraint is time. While data is collected during some transactions, it takes additional time to ask the customer for additional useful information, and then actually type that data.
One example of a point-of-sale data collection question is about purchase influence: “Is it for personal use or for a gift?”
Another example is the classic, “How did you hear about us?” From a customer service perspective, we are not easily able to capture this information. And during cash transactions, data is not collected at all. This creates a total void for analysts trying to do their jobs.
What if every cash register had a microphone and voice recognition software that could record the conversation around each transaction? This data could be translated to a written data set that would be available for query and subsequent data appends. There is ample time during a simple cash register process to ask a couple of questions or even capture names and addresses if they agree to be added to a mailing list of some type. Voice recognition software will eliminate the time restraint on the front-end of the transaction, while time spent recording data on the back-end can be automated.
The information technology necessary for such a scheme exists now. But I can hear a few of you already saying, “After data scientists get more budget dollars, there will be a privacy concern to overcome.” As this is a data wish list, we will let the law and regulations professionals solve the privacy challenge.
Salespeople Data Challenge
While there are ample platforms out there to capture data from sales professionals, the main constraint here is — once again — time. It takes a lot of time to enter all that data, and sales people prefer to spend their time selling, rather than typing endless bits of information into a platform. In many cases, they simply will not enter any data, or they make it up, keeping these interactions brief to maximize their sell-time and commissions.
But what if every platform supporting salesforce data had voice recognition built into it? Talking takes much less time than typing, and this system would vastly increase the data sets available. If analysts had the budget to capture voice-recorded data and then funnel it into an environment for query and reporting, it would be a dream come true for data scientists.
Larger budgets could allow for appending captured data to customer and sales records. These appends could be used as lead generation and conversion intelligence. The lead generation element is necessary in the design of a sales funnel that allows us to understand where and why conversations end. The conversion element would be focused on what happens when a sale is achieved. This intelligence would be instrumental for improving conversion rates.
Digital Data Overlap Challenge
All analysts are familiar with the challenge in any campaign of reconciling digital data to completed orders. Web analytics attribute orders one way and specialized digital platforms — paid search, for example — attribute orders a different way. The delta is usually the window that each uses to allow for an order to get assigned to a reference code.
For example, Google Analytics may report 608 orders while paid search, email and social media platforms individually report a combined 800 orders, as seen above. The challenge for the data scientist is figuring out which media really influenced the sale.
Voice recognition can once again help with this conundrum. A link could be added to a shopping cart that prompts a buyer to answer a simple question as they send for their order: “What influenced you to choose this product or service?” A promotional offer of some type would ensure an acceptable response rate. This link would enable the microphone on the buyer’s computer and capture the answer on the database within the Web platform. Those results could then be appended to that transaction during a response analysis. There are quite a few details to be worked out on this one, but this is certainly within the reality of wish-list scheming!
The integration of information technology and marketing is accelerating into the years ahead and inviting data scientists to be bold with their wish lists. Be brave in asking for larger budgets to test ideas. It’s a terrific time to be a data scientist!