Photo by Claudio Schwarz on Unsplash

Creating valuable data is a force multiplier

It may come as no surprise that data is important for information technology and that information technology is fundamental to the operations of most organizations who consequently depend on data for their daily operations and need it to fuel new and innovative solutions. Unfortunately, it is often not managed with appropriate diligence resulting in everything from additional manual labor to full-on meltdowns due to data errors  such as duplicative, ambiguous, and inaccurate or missing data. 

Such issues need to be fixed urgently but in an organization there is at any given time always any number of issues that need fixing. Fixing data issues could, however, be a force multiplier compared to fixing anything else. That is because data impacts more than one domain. Solving data issues improves not only the operation of the business but also its analytics and AI capabilities.

Smooth operations

Data issues often result in manual work to correct data or enter it multiple times in different systems, if it has not been captured correctly or stored and shared in the right format. Poor data costs organizations around $15 million on average according to some estimates.

Fixing data issues will result in increased speed and quality of service because the right data is available at the right time with minimal effort. It will also improve efficiency because corrective data actions, like retyping, troubleshooting, correcting, and searching for data will be minimized. Finally, correct quality data will reduce the number of errors in business output: you no longer call the wrong customer or send the order confirmation to the wrong fax machine, 

This is done by making sure that data is entered only once or sourced from an authoritative 3rd party and then made available across the enterprise through proper integration. 

Single source of clean data

Having correct and intelligible data available will result in an increased ability to generate knowledge and business insight. For decades the main quest for BI has been to get at the single source of the truth. That quest has not been completed. The ease with which new systems are procured and developed has expanded the array of possible sources of truth significantly. It is consequently still necessary to get high quality consolidated data.

Getting high-quality from source systems reduces the time to insight, since data does not have to be cleaned before it can be used. Data scientists and BI developers spend between 20 and 60% of their time collecting and cleaning data according to various surveys. If there is a clean single source of the truth this time can be saved and spent on more value creating activities. 

Data is the mother of (artificial) intelligence 

Data is even more important for AI, since it determines functionality directly. Low quality data leads to unwanted performance in the final solution such as increased rates of false positives and false negatives and will cast doubt on the general viability of the solution. Zillow’s home buying division was eventually closed down because bad data led to inaccurate predictions of home prices. According to Forrester, 60% of AI failures are attributed to data failures. 

Good data will lead to higher quality AI solutions, which may facilitate better business decisions like procurement of products, investment timing, human resource planning etc. Correct data therefore leads to more trustworthy AI services, which means it will be used more in practice. Nothing undermines trust in AI solutions more than bad performance based on bad data.

Optimizing data is a force multiplier

Data issues cost millions, create significant amounts of additional work, and undermine business performance. It impacts three different domains but it is only necessary to solve it once. The best way to do that is to start by understanding what makes data valuable in a given context. I have developed the dala value scorecard as a simple tool to sketch out the seven critical dimensions of data value: Consumption, structure, granularity, freshness, content, validity and intelligibility. That will provide directions for where to look for improvements. Optimizing how your organization works with data is the single most impactful way to improve performance. Data is a force multiplier. 

Photo by Claudio Schwarz on Unsplash


Udgivet

i

af

da_DKDanish