Artificial intelligence, automation, self-learning models and digitalization of the oil field are all terms that have been bandied around the oil and gas industry over the last couple of years. There is rarely a conference or workshop that does not include a session somewhat related to these terms. The marketing departments of oil companies and service companies alike trumpet the latest advances without many, or any, concrete examples—or an actual definition of what they really mean.

There is much discussion on how digital twins and Big Data analytics will improve safety and efficiency, but the digital twin is pretty useless without its real-world counterpart. Data analytics without the right data is just a big guess.

Data are the fundamental building blocks of any model or system. Without data, there is no analysis. It is just opinion. As Sherlock Holmes uttered in the 2009 movie by the same name, “Data! Data! Data! I can’t make bricks without clay.”

Therefore, what is striking in the industry is the lack of the right data in many areas of operations, particularly in the highly complex deepwater environment. In these scenarios, risk and cost are both heightened. Crucial decisions about downhole conditions often are derived from models based on measurements and data from the surface, which may not adequately reflect what is happening 6,096 m to 9,144 m (20,000 ft to 30,000 ft) away in the wellbore.

Recent industry papers from multinational operators indicate that real-time downhole data are available for 15% of total well construction time and that this is primarily limited to on-bottom drilling while circulating. Tripping in and out of the hole, casing, cementing and completion activities are essentially performed blind and reliant on models that may not have the right data to inform the correct decision.

Albert Einstein once said, “Not everything that can be counted counts, and not everything that counts can be counted.” Data are important, but the right data are essential.

Before blindly accepting the output or analysis of models, we have to fundamentally understand the limits and the assumptions inherent in those models and whether those assumptions are adequate to resolve the question we are asking. As drilling, casing, cementing and completion windows become tighter and as well complexity increases and reservoirs deplete, the need for the right downhole data to resolve these questions becomes paramount. We should constantly be exploring the best way to acquire the right data to resolve industry problems.

I truly believe that Big Data, data analytics and the other current buzzword terms will have a major influence on how the oil industry operates in the future. However, this initiative will fail if we are not acquiring the right data at the right time and from the right place to make it work.