Data is the mother of all solutions. It has the potential to address nearly any challenge when it is properly analyzed and interpreted. Although insights derived from data are not always perfect, most decisions are made with less-than-perfect information. Decision-makers often believe they lack sufficient data; however, the real challenge lies not in lack of data but in identifying what is relevant and then extracting, analyzing, interpreting, and applying it effectively. Often, the data you need may already exist within your own database, and when direct data is unavailable, it becomes essential to explore information that is indirectly related.
For data to become truly useful and meaningful, it must be cleaned, processed, organized, and analyzed. Once cleaned, data transforms into information, and when that information is analyzed, it becomes insight.
It is this insight that guides decision-making and, when implemented, drives action.
It is important not to dismiss data that challenges existing beliefs. Instead, such insights should be tested, as they may lead to meaningful shifts in understanding and approach.
Advancements in technology, coupled with a growing willingness among internal and external partners to share information, have significantly increased both the volume and variety of available data. Organizations today have access not only to structured data, such as point-of-sale records, orders, shipments, inventory levels, warehouse withdrawals, and customer forecasts, but also to vast amounts of unstructured data, including audio, video, images, social interactions, emails, GPS signals, sensor data, and web content.
However, access alone is not sufficient. It is crucial to determine which data is most relevant and to understand when and how to use it effectively. Equally important is the ability to identify and resolve inconsistencies or quality issues within the data.
Insights derived from data can transform a company’s marketing strategy. While human judgment remains important, it can be influenced by bias or self-interest. In many cases, data-driven decisions prove to be more reliable. The following discussion highlights four cases where data was effectively used to support sound decision-making.
Case Study Number 1: Cigarette Manufacturing Company
The Marketing Director of a cigarette manufacturing company was frustrated with low forecast sales numbers. He asked the Forecaster to use not the data of 36 months, which he normally did, but less than that, say 24 months or 12 months of data to prepare the next month forecast and then see what happens. When he did, he got much higher numbers.
This pleased the Director but raised concerns for the Forecaster, who believed that using 36 months of data would provide greater accuracy.
Since the Forecaster was not happy with these numbers, I was contacted to validate them. They have six years of data; I used the data for the first 36 months and prepared the forecast for the next month, month 37, with the same model the company had been using, and then dropped one month from the top and added one month at the bottom to prepare forecasts for the next month, in this case month 38. This way I prepared forecasts for month 38 and beyond and generated as many forecasts as the data permitted. I did the same for preparing as many forecasts as possible using 24 months and 12 months of data.
The results showed that the Mean Absolute Percentage Error (MAPE) was lowest when 36 months of data were used and highest when only 12 months of data were used, confirming that the Forecaster’s approach was more accurate. The 24-month data produced results that fell between the two.
Case Study Number 2: Steel Doors Manufacturing Company
A company manufacturing steel doors used in institutions such as prisons, hospitals, and schools observed a significant increase in sales over four consecutive months.
The company was unsure whether this increase was due to a sudden expansion in the market or simply a temporary fluctuation. This month- on-month rise seemed unusual and raised doubts. Further investigation involved analyzing industry-wide data collected by the association of manufacturers. It was found that sales had increased for nearly all companies during the same period, initially suggesting overall market growth. However, deeper analysis revealed that one major competitor had gone out of business. The increase in sales was therefore not due to market growth but because the company had captured the market share of the exiting competitor.
Case Study Number 3: Fund-Raising Company
A fund-raising company that raised money via mail found that one of the new mailing lists generated an 18% response, something unheard of.
Usually, they would be happy with a response as low as 1.5 %. The list has a total of 200,000 names. We tested another 50,000 names, and the response was still high, which was 12%. Normally, the response goes down when a larger quantity is used.
By going over the list, we realized that there are a lot of duplications in the list. If we eliminate them, we will further improve the response.
The logic made sense. For our next campaign, we mailed another 50,000 unduplicated pieces of the same list, but the return, instead of going up, went down sharply. It came down to 9%. This gave us an important insight: duplications help, not hurt. We then started capitalizing on this insight by sending an acknowledgment letter four weeks after receiving the original donations and asking again for donations. To our surprise, we got an 11% response on acknowledgement.
This not only changed the company’s core assumptions about best mailing practices but also provided another source of revenue.
Case Study Number 4: An Insurance Company
An insurance company offering a hospital policy used a two-step process to acquire customers by first generating leads and then converting them. The policy was if you go to hospital, the company will pay you $200 a day. Leads were obtained through direct mail, print advertisements, television, and radio. Initially, all leads were treated equally during the conversion process. However, questions arose regarding whether all lead sources were equally productive, but the cost of generating them varied. It was the highest when generated by direct mail and lowest by radio. By analyzing the data based on the source of leads, it was found that direct mail generated the highest conversion rates, followed by print advertisements, while radio produced the lowest. Based on these findings, the company decided to discontinue radio advertising, thereby improving the efficiency of its marketing efforts.
Lessons Learned
These examples illustrate the power of data in providing insights, even for complex problems. In many cases, the data you need is
already available within your own database; the real challenge lies in identifying and making effective use of it. It is essential to clearly define the objective before beginning any data search or analysis. The goal should be to generate practical, actionable insights rather than purely theoretical conclusions.
Results should always be validated rather than accepted at face value.
Maintaining an open mind is equally important, as findings may challenge existing assumptions and therefore must be tested before being applied.
Because relevant data is often not directly available, it is frequently necessary to rely on indirect or proxy sources.
Data-driven decisions tend to be more reliable than those based solely on intuition. Moreover, simple analytical approaches can sometimes perform as well as, or even better than, more complex models. While human judgment remains important, it must be applied objectively and with an awareness of potential bias.