Owning a small business requires creating a strategy and hammering out specific details, but you can't do either one without having some sort of data to inform your decisions. The more data you collect, the more important it is to know how to properly analyze and interpret it. For example, when you want to see how different data points relate to each other, you create a simple line graph known as a regression analysis. Although it sounds fancy, you only need to resurrect your basic algebra skills in order to do it properly; don't panic if you failed calculus!
How to Use Regression Statistics in Business
A linear regression refers to a line graph that shows the relationship between two sets of data. One of the easiest ways to start thinking about using regression statistics in business is to consider trends over time. Time represents one of the data sets, and you can choose from any other statistic to view in correlation with time, such as product sales, total revenue, website page views, social media interactions, etc.
This linear regression would show you whether or not your business is steadily improving. It's important to show this on a graph with concrete data if you want to talk to potential investors or decide whether it's worthwhile to continue paying an advertising company, stocking certain items, etc. You cannot simply rely on your gut feelings when making business decisions. Statistics, particularly regression statistics, clearly show what's working and what's not.
Of course, you need to also know what kind of statistics to pit against each other and how to interpret the graph. Plenty of other regression statistic examples exist for business applications, but before we take a closer look at them, let's go over some helpful terminology.
Independent vs. Dependent Variables
Your line graph will have two variables: One will display horizontally while the other displays vertically. The horizontal line is called the x-axis and is known as the independent variable. The vertical line is called the y-axis and is known as the dependent variable. The dependent variable changes in response to the independent variable, but the independent variable can be adjusted to show different parameters.
For example, time is an independent variable. You can adjust the time shown on a linear regression by deciding whether you want to track hours, days, months or years. The dependent statistics, such as sales revenue, will change based on which frequency you choose.
You can glean different information by adjusting the independent variable, so feel free to play around with it. For example, you might be well aware of your monthly, quarterly and yearly sales revenues, but have you looked at hourly revenues? This could show you that you make most of your sales in the morning, afternoon or at closing. In turn, this could directly influence how you create your employee schedule.
Basic Regression Equation
The basic regression equation is y = bx + a, where "a" represents the value of y (the independent variable) if the value of x (the dependent variable) is zero, and "b" represents the slope, or rate of change. Using this formula and a set of (x,y) data points, you can plot a linear regression by hand (or use spreadsheet software to generate a graph) and note a positive or negative correlation in the data.
Advanced Regression Statistics
Among the additional statistics that can be calculated from a regression is one called R-squared. The value of R-squared demonstrates how closely correlated the two variables are. The value is between 0 and 1. A value of 1 means that the Y variable is 100% dependent on the X variable. A value closer to 0 indicates that no significant correlation exists between the data points.
If you'd much rather pay someone to go to all that trouble, you certainly can: business analysts or data analysts earn a living pulling relevant data from your records, displaying it in a linear regression and performing additional statistical analyses to help you understand what it means. They can even perform much more complex analysis with multiple variables in play simultaneously.
But you can also use technology like Microsoft Excel or Google Sheets to do simpler regression analysis yourself, even without having to dust off your calculator.
Using Excel for a Regression Analysis
To use Excel for a regression analysis, you need to first ensure you have the Analysis ToolPak enabled as an active Add-in. If you go to the Data tab and do not see an Analysis section, go to File > Options > Add-ins. Then, make sure Excel Add-ins is selected in the drop-down for "Manage Add-ins" and click Go. Check the box for Data Analysis and click OK.
Now, you should be able to see the Analysis block in the data menu. Before moving on, input your data into a sheet. Clearly label the columns and think about which variable you're trying to understand. That will be your dependent variable, because you're trying to understand how much its value depends on the other variable(s). You should only have one dependent variable, but you can have multiple independent variables analyzed at once by Excel.
Click on Data Analysis and select Regression from the list. The next menu prompts you to select your independent and dependent variables. If you select the labels for your columns, check the "label" box. Your results will appear in a new tab on your Excel sheet.
Using Google Sheets for Regression Analysis
Don't have Microsoft Excel? No problem. You can use Google Sheets for free to run a regression analysis, and the process is very similar. As with Excel, you'll first need to get a free Add-on that allows you to do this kind of statistical analysis.
To do that, go to Add-ons > Get Add-ons and then browse for "Regression" to bring up appropriate analytical Add-ons. For this example, we've chosen XLMiner Analysis Toolpak. Once you've chosen your Add-on, click on it in the Add-on drop-down menu and click Start. A menu will appear on the right-hand side of the screen. Scroll down and click on Linear Regression.
Next, you'll input your y and x variables just like in Microsoft Excel: Simply highlight the appropriate columns or ranges. The major difference is Google Sheets will not automatically create a new tab with your data. Instead, you need to tell it where you want the data to go, so select an empty, out-of-the-way cell on your spreadsheet for the "Output" box. Then, click OK.
Understanding the Regression Statistics
Although each piece of information can be analyzed, for now, you should focus on two values: R-squared and P-value. Start with the R-squared value, which is given as either between 0 and 1, or between 0 and 100%. If the R-squared value is close to 1, then the variables are related. But the further it is from 1, the murkier the correlation between the variables.
For example, if you have an R-squared value of 0.96, that means 96% of your y variable depends on the x variable. If you start experimenting with the x variable, whether it be the time your shop is open or the price of your product, then you should expect the y variable to also change.
Similarly, look at the P-values. It should be less than 0.05 if the data has a valid correlation, but closer to 0.01 is even better. If it is more than 0.05, then it's likely your variables have nothing to do with each other. For example, it's unlikely that someone's affinity for French toast depends on their shoe size, and thus a regression of these data points would have a P-value of more than 0.05.
Predicting Future Outcomes
Once you've found a variable that depends on another, you can start to input hypothetical data in order to make predictions. Let's say that you have found that the number of employees on the floor during hours of peak traffic has a close correlation with the number of sales you make during that time. What might your sales look like if you added another employee to the shift? What happens if someone calls in sick and you have one less employee on hand?
You can also use a regression analysis to plan ahead about factors out of your control. Let's say you operate a food truck. How do your sales relate to the average temperature? If hardly anyone buys your spicy tacos on a 95-degree day, you can save yourself the cost of running the food truck by only setting up shop during temperatures that seem to give you more sales.
Deciding Whether You're on Track
You can also use linear regression to determine whether your current strategy is working or whether you need to make some adjustments. For example, you can track the number of website visits in relation to the number of sales made online. How many people have to visit your website in order for you to make a sale? This could help give you a web traffic target for social media, paid ad and SEO campaigns.
Speaking of social media, most platforms offer their own analyses, but sometimes you have to pay for them. Plus, they don't always give the kind of data you'd really like to drill into. Learn how to export data into a spreadsheet in order to run your own regression analysis. That way, you can look into things like whether the time of day or the type of media shared significantly impacts engagement.
Collect data on any other advertising, marketing or sales initiative that you undertake, even if you have to record it manually into a spreadsheet. Make sure you collect data long enough to get as many data points as possible, as this will help make sure your analyses are accurate. Then, get your regression statistics to determine whether the effort is worth continuing.
For Larger Businesses, Hire an Expert
If you're a quick learner and don't break out into cold sweats at the mere mention of math, then you can probably handle a regression analysis on your own. But in order to do it properly, it does take time. As a small business owner, do you really have room on your plate to dedicate to statistics? It may be a better use of your time to hire a data analyst to do it for you and present you with the results.
For a large business in particular, you could have an in-house analyst to run regular reports for various departments, ensuring accuracy and proper interpretation. For a smaller business, you can contract with an analyst on an as-needed basis. Whichever course you decide to take, don't underestimate the power of having properly analyzed and deciphered data at hand before making any major decisions.