Better decisioning in the age of data overload
In the world of people-based marketing, the magnitude of data collection has become so great that Jeff Bezos has a service dedicated to shipping data in semi-trucks. International Data Corporation (IDC) estimates that compound annual growth rates (CAGR) of spending on AI systems will reach 37.3% during the period from 2017 to 2022. This is supported by Adobe’s survey indicating 82% of marketing executives will adopt AI for their personalization strategies. Despite this enormous growth of investment, the need for human marketers to ultimately make the decisions will not diminish anytime soon.
International Data Corporation (IDC) estimates that compound annual growth rates (CAGR) of spending on AI systems will reach 37.3% during the period from 2017 to 2022.
In many aspects, business decision making will remain a human process. Recent news and research have exposed issues with bias inherent in AI systems, which can result in misidentification. Facial recognition has become a target for some critics. Marketing decision makers will not retain this function by becoming AI data crunchers, learning advanced ML techniques and their limitations, or implanting a super computer in their heads. Data visualization will always be vital to simplify results, enabling humans to override decisions made by a machine.
To be sure, we haven’t reached the full potential of the AI world of tomorrow, but we are already seeing the problems of an ever-expanding world of data. Over the past 10 years, marketers have been hearing about the mandate to break down organizational “data silos” and merge this data together to provide analysis for marketing decisions. And for 10 years, we’ve had surprisingly little success. Once the silos are overcome, a problem always emerges when bridging these disparate sources: Human decision makers typically want too much data, and they are often reluctant to update legacy reporting processes, such as leveraging an Excel table.
The idea that more data correlates to better decisions is woefully rampant in the marketing realm. Unfortunately, dashboards have begun to resemble Excel spreadsheets, as brands demand more data to be shown. The notion that this greater volume leads to better understanding and decisioning is unequivocally flawed. In fact, more choices, more KPIs, and more granular data, obfuscate decision making.
Solving this issue requires redefining how the business measures itself and how custom KPIs can be used to simplify understanding. Dashboards can then be leveraged to help marketers draw proper conclusions or identify where a deep-dive analysis is required. Below are some steps to take, with the goal of culling the data to a handful of elements that simplify the decision process while still providing insights needed to make informed decisions.
1: Avoid Excel table visuals
Ditch the desire to view data in tabular form. Review each of your tabular reports each month and circle the metrics you use to make a decision. Place an X near those that you like to reference as “good to know.” After three reports, consider removing all but those metrics you circled.
2: Get comfortable making decisions with less data
The world won’t come to an end when a decision is made with less than ideal information. We rarely have the time necessary to become 100% positive that our decisions are correct. If we ever meet the 100% confidence threshold, it’s likely the opportunity has passed. Hindsight is no match for decision.
3: Experiment with creating custom KPIs
KPIs are often mistaken as a single metric or simple calculation of metrics. However, KPIs can encompass many metrics in a way that provides insight into performance. While not all will impart the confidence to make a decision, they will help to narrow down where to focus those precious analyst hours. For example: To understand how clicks translate to visits, create a ratio of site visits to ad clicks and ditch the click-through rate (CTR) from the media companies. This ratio can be used to evaluate publishers and request additional impressions or refunds.
Evaluating large data sets just because we have the data does not lead to stronger or more accurate decisions. Many times, it can overcomplicate and slow the decision process down until a poorly informed decision must be made. Review your reporting processes and identify the key metrics you use to make decisions, then save the others for when you have free time. Or just leave the office early.