Seems a bit like an oxymoron, no? Well, that’s exactly what analytics have become these days: an oxymoron. A real conundrum. On one hand, data helps us predict change and plan for the future. On the other, that data can be wrong or misleading and, therefore, really screw things up. So, I say, take it all in, but then let (most of) it go.There’s an ongoing debate regarding the roles of data and entrepreneurship. In particular, the increased availability of analytics data and tools is making planning, scheduling, and analysis much simpler and more accurate. Amazon is one of the best examples of using analytics to improve logistics (i.e. more one-day shipping).In contrast, the argument stands that these tools are less effective than originally expected. The most significant instances are incorrect data, method, and change. If the data is wrong, access to more data does not improve analysis. Mistakes like Boeing, Afghanistan, WE WORK, G.E. and retail stores represent diverse examples where people simply focused on wrong information. The existence and use of the phrase “alternative facts” supports the unnerving idea that it’s easier to make up lies than it is to refute those lies. That alone does not bode well for analytics and data.Data can also be misleading when a dramatic change occurs. Disrupters like E-Commerce, ride share apps, and food delivery dramatically affected markets and parameters. Consequently, significant shifts in culture, politics, and buying habits also make economic forecasting much less reliable.Additionally, analysis is dependent on using the right tools and methods. Many assumptions and approaches may not be appropriate. For example, investment advisors frequently tout their individual excellence while changes in the overall market are usually the largest factor in investment success. Mathematics shows that the more history one has on a topic, the more accurate the analysis. However, if parameters change, history may become irrelevant.This is why we take it all in. Think on it. Absorb it. Let it all sit for a bit. And then throw most of it out the window.You should absolutely consider what they teach on the first day of a statistics course (Validity, Reliability, and Accuracy) rather than ignore it.A recap in case you need a refresher: Validity is simply focusing on whether your methods are valid. While sampling, correlation, and other tools can improve performance, the analysis must be valid. For example, many of us predict that our team will win. However, the odds in most professional leagues are that about 3% of approximately 30 teams will actually win. Reliability is the repeatability of results. Differing results in political polls or verifying results of medical tests are examples of reliability issues. Accuracy is just the correctness of the measurement process. The most violated rule of accuracy is that you are only as accurate as your least accurate number. There is a famous story about a museum guard answering a child’s question about how old a dinosaur was. He said 280 million years plus 39 years and 20 days. When asked where the number came from, he said, ”When I started, they told me it was about 280 million years old. I have been here 39 years and 20 days.” While this number certainly seems precise , it probably isn’t very accurate .I would add a fourth factor to this list, which is probably the most important: Bias. On one hand, bias is a complex mathematical term correlated with sampling, randomness, analysis, and other things. On the other, it is how our culture, background, gender, age, and preconceptions etc. affect our attitudes and decisions. For example, many studies have shown that we form an opinion about a presentation within 90 seconds of it starting. I highly recommend that, in dealing with bias, you manage its existence rather than trying to deny it.Finally, tools as well as methods of reporting are dramatically changing. A colleague of mine recently challenged my website saying it was “too dependent on PowerPoint and Excel.” While these are both great tools and are the most dominant analytical and presentational methodologies, they can have many limitations: The information can be old, longitudinal analytics is frequently lacking, they are not interactive, they are not visual enough, and they can be very boring and/or misleading. Nothing is worse than being forced to sit through a PowerPoint presentation that is too long and loaded with endless Excel sheets.In summary, analytical tools offer great potential for success, but they need to be utilized properly and in conjunction with intuition to be effective. So, gather all that data and pay close attention to it, but don’t be afraid to toss it all out.