<img alt="" src="https://secure.hims1nice.com/151009.png" style="display:none;">

5 Ways to Get Better Data for Better Decisions

We all want to make good business decisions. They are crucial to the success of our enterprises. The quality of our decisions stem from two factors:

1. Sound decision processes and skills. 

This is why we get MBAs and continue to learn and study all our lives.

2. Good (true) knowledge (data) of the environment and situations which command our attention.

If you make a sound decision based on incorrect or the wrong data, you will get a bad decision and possibly a bad outcome.

Several times since nuclear weapons were developed, US or Russian armed forces have been confronted with radar readings missile attacks coming at them over the North Pole from the other. You don't remember the WW3 Nuclear War? That's because calmer heads prevailed and determined that the reports were for the Aurora Borealis, the moon, an accidental training exercise, or other false alarms. The "Good" decision would have been to launch retaliation. The decision would have been sound, but the data was bad.


The ramification of a bad decision on your part might not mean worldwide destruction, but it can have a significant impact on the performance of your organization.

 

How do you ensure that you have quality data to support your decisions?

Microsoft SQL Server ways for better data and better decisions

 

1. Know what you want/need to know. Ask a question before you look for an answer.

Ever hear someone say "We need some marketing data"?

You don't just need data; you need data which leads to effective actions or positions.

Units shipped are up. That's good. Maybe.... but are they up because:
  • We are getting more repeat sales from existing customers?
  • We are getting lots of one time customers who don't come back?
  • We had an uptick in returns and replacements for defective product?
  • Wholesaler's or retailers are increasing inventory?
  • A recent coupon campaign has cut our effect price by one third?
  • Customers have started buying in bulk?


While any of these scenarios would explain an increase in units shipped, each should lead to a different plan of action. Analyze your decision processes and be sure you are depending on data which points you in the most effective direction.

 

2. Be sure it's economical.

An unrestrained quest for information can get very expensive very quickly. Subscribing to three different market surveys may seem to give us three times as much data. This might be true if each focuses on different segments, different granularity, different methods and gives us different results. The key word here is "different" if that uncovers some fact that the others miss. But if they merely represent three copies of the same report with different headings, and any one could have pointed us in the right direction, then maybe some of our budget could be spent better elsewhere.

 

3. Is it absolutely necessary to get exact numbers to make good decisions, or do we just need trends and gross numbers to make the correct decisions?

Sometimes data which is too precise and timely can even be bad. When barcode scanners were first introduced, market research firms made daily sales of products available. When a 30% increase in sales one week and a 40% decrease the next made sales managers panic, until it was discovered that it was as simple as our product being featured with a coupon in the local store's add this week and our competitor being featured the next. Over the course of a month or a quarter, everything could be viewed as normal.

 

4. Be sure it's timely enough. 

Quick data is often less than accurate and can be very expensive to obtain, but late information is no help at all. Think about how unemployment figures are released for last month, along with corrections for prior month and the one before that. Often the corrections are significant. So examine the information you are depending on to sure that either it doesn't change after you make your plan, or if it does, your plan allows for anticipated changes.

 

5. Be sure your people trust it.

Does your organization agree on the critical data? For instance, if we consider revenue, does accounting have one answer, and sales management a different one which individual salespeople almost never agree with? Does someone in marketing have a spreadsheet which he and his manager insist gives the correct number? The existence of such spreadsheets and auxiliary departmental systems are a good indication that the information your people are being supplied with is either untrustworthy or not on point for the decisions. If this is the case in your organization, it is critical that you find out why the information is not trusted and resolve the issues.

 

"If you can't get the numbers you want, consider things that correlate."

 

Sometimes it is too difficult to actually obtain the metric you want or need. Consider finding a more easily obtained data point which moves up and down in synchronization with the one you want. When looking for an early indicator of total manufacturing production, one can track the production of wooden skids used almost universally for shipping. We probably don't really care about the wooden skid market, but their use does track activity across a broad sample of manufacturing activity. If you are a bank looking for expected loan demand, this might be good enough.

 

Be careful though not to get caught by coincidental correlations. It has been observed that, historically, when an "old NFC" team wins the Super Bowl, generally the stock market goes up the next year (with exceptions), and when an "old AFC" or expansion team wins the market generally goes down. This is probably not a good strategy to depend upon for playing the market.


Executives spend their day making decisions. Maybe it's time to decide to help them make better decisions by providing better data.


Want to learn how to make better decisions based on data?

Request a FREE Data Platform Discovery Session!

Schedule Session!

Posted by Jim Hodges

Senior BI Consultant

Website

Topics: SQL Server, Power BI