Data Analytics is Now Accessible to Small Businesses Like Never Before
For the better part of the last couple of decades, the field of data analytics has allowed large corporations to grow, scale, and optimize while leaving small businesses in the dust. Corporate profits are at all-time highs, and it can often feel like an impossible uphill climb for entrepreneurs and business owners to keep up. The monopoly on data certainly was a large part of the growing gap between the haves and have-nots.
The good news is that unlocking data is now easier than ever before- even for the most cash-poor and resource-limited organizations. But first off, what is data analytics? Data analytics refers to the statistical analysis of raw data in order to make conclusions about that information. In other words, every sale, every customer, even every Facebook “like” is a piece of data that is being collected somewhere. Data analytics turns these seemingly inconsequential data points into a collective of meaningful information that can drive decision making.
Within data analytics, there is a sub-category called data science. Traditional data analytics focuses more on viewing the historical data in context while data science focuses more on predicting the future. Data science is a multi-disciplinary blend that uses machine learning and predictive modeling to solve analytically complex problems.
Why Data Matters for Small Businesses
Being a data-driven organization is not something to do because it is trendy or because you want to appear high-tech. Small businesses need to realize that being a data-driven organization works. Understanding and acting on your data will impact your bottom line. Savvy business leaders invest in data because better data leads to better information, and better information leads to better understanding of your business. A better understanding of your business leads to better decision making. Making better decisions leads to better outcomes. These outcomes come in the form of increased sales, improved operational efficiency, reduced churn, increased return on investment, and ultimately higher profits.
While the current state is that large businesses are still disproportionately benefiting from the current data landscape, the tides are turning. This is primarily for two reasons. The first reason is the trend of democratization of data. Accessing not just data- but tools and technologies that help you make the most of data- is easier and more affordable than ever before. Don’t believe it? Take into consideration all of the affordable tools that are now available for a small business or non-profit:
Google Analytics and Google Cloud Platform are available for free to analyze your website traffic data.
You can code in Python (the top coding language for data analytics) in environments such as Anaconda or PyCharm for completely free.
BigQuery is a data environment where you can run scripts and store data for less than $1/month.
Zapier allows you to move data from a source to a destination, and has a free version with substantial functionality.
You can build reporting dashboards that will allow you to visualize and track trends in your business for completely free using tools like Google Data Studio or Tableau Public.
All of your social media, marketing, and operational platforms are collecting data about your performance that you can access and extract for free.
In addition to the tools that are available for little to nothing, companies like Render Analytics that specialize in servicing SMBs and nonprofits are affordable, value-adding resources. These data analytics agencies cater to smaller organizations instead of massive corporations not just because of good will, but because there is an underserved market and a legitimate business opportunity.
As a result of the rising availability of data to smaller businesses, business owners have an opportunity like never before to use technologies and techniques that allow them to improve their decision-making (and therefore their profits). This article will review some use cases of data analytics and data science that small businesses with limited resources can leverage. For years, these use cases were traditionally only available to organizations with more resources and scalability, but now even the local barber or restaurant can take advantage!
Increasing Website Leads & Sales through A/B Testing (Data Analytics)
A/B testing is a data analytics technique that compares two versions of the same element against one another to see which one performs better. In other words, A/B testing is like running an experiment! Let’s say you are running an A/B test on your website home page. On one version of your home page, the primary call to action says “Call Now”. On the second version, the primary call to action says “Book an Appointment”. If you sent half of your website traffic randomly to version A and the other half to version B, you could see if one of the experiments generates more phone calls than the other! This is an example of how to run an A/B test.
Why perform A/B testing on your website? Depending on the nature of your business, your website might serve a certain purpose for you. If you are an e-commerce company, you likely sell your products directly through your website. If you are a brick and mortar business, you might use your website to generate phone calls or leads in some other capacity. No matter what a successful lead through your website looks like to you, you can use A/B testing to improve the percentage of website visitors who take that desired action. In other words, A/B testing can improve your conversion rate online!
Additionally, you can A/B test your website to reduce bounce rate. Bounce rate measures the percentage of website visitors who enter and exit your website without taking any action. These are people who visited your website but did not engage with your content in any way. You’ll never get a bounce rate of 0%, but if you can lower your bounce rate by 5-10%, that becomes 5-10% more potential customers who are actively interacting with your brand.
So let’s see A/B testing in action. One of Render Analytics’ clients- a spa chain in central Florida- has a services page on their website for their different massage therapy offerings. There are around 8-10 massage-related services that this spa is offering at any given time. This massage therapy page is the second-most viewed page on their website behind only the home page. The problem is, the exit rate for this page was historically very high, and very few people who hit this page ended up calling to book an appointment or making any other purchasing decision.
This sounded like a perfect opportunity for an A/B test! So Render Analytics created a second version of this page. This new experimental version still had all of the same main content, but had a slightly more modern design, and perhaps more importantly, included a link to become a member and a link to book an appointment next to each and every service.
The result? A double-digit percentage increase in membership sign-ups and phone calls to book appointments from this page! But this experiment might feel a little easy- someone could look at both designs and intuitively guess which one performed better. Let’s look at an A/B test that isn’t so cut and dry.
This A/B test of a lead generation form illustrates how wrong our assumptions can be. Most people would assume adding an eTrust image would improve form completions. After all, it makes the potential leads feel like there is an added layer of security!
The results of this test would challenge that assumption. Version B ended up outperforming Version A by about 13% in terms of form completions. Excluding the certified privacy image actually improved the performance of the form! As it turns out, images of this type are usually associated with payments and credit card information. As a result, many users assumed that they were about to pay for something and decided not to complete the form. This is a perfect example of why A/B testing is important! The data often reveals information that our instinct does not.
If you are a small business wanting to try A/B testing on your own site, you have a couple options! While most A/B testing software is expensive, there are two offerings out there that are very affordable for small businesses. The first is Google Optimize, which is completely free. While the affordability is great, it can be a little technically complicated and will require you to continually add and tweak code on your website. You might want to consider hiring a data analytics agency or freelancer to help you if you choose this route. AnalyticsBox- a website platform provider that competes with companies like Shopify and Wix- is the other affordable alternative. Just like other website platforms, you might have to pay a couple of dozen dollars per month to host your website, but A/B testing is completely built into the platform and included for no additional cost! This makes the coding process much easier, and you do not have to be as technical to run A/B tests!
A/B testing is not limited to only websites. In fact, A/B testing is not even limited to the digital space. For example, you can A/B test the pricing of your products to see if your customers have price sensitivity. If you can get away with raising pricing in an A/B test, you can position your business to make more money per sale. Another offline use case for A/B testing is direct mailing. Direct mailing has been making a comeback in terms of return on investment, and a big part of that is the rise of A/B testing and data analytics. To learn more about A/B testing for your small business, give this article a read!
Reducing Customer Acquisition Costs through Pattern Detection (Data Analytics)
Analyzing and understanding historical data to identify patterns or trends is a helpful strategy when making future business decisions. Too often, business decision makers trust “their gut” or drive their business based on what they believe are best practices. Frequently, our intuitions and concepts we consider “common knowledge” turn out to be misleading or misinformed. By truly understanding historical customer acquisition data, business owners can grow their business and generate more customers at a lower cost per acquisition.
Data-related pattern detection techniques can be applied in two primary use cases in regards to customer acquisition. The first use case is choosing the best channels to focus money and time on acquisition efforts to get the optimal return on investment. The second use case is to determine within individual sales or marketing channels the most profitable target audiences and focus resources on the low-hanging fruit.
Years ago, I worked for an e-commerce business in the digital image licensing business. They were interested in retaining more first time customers as ongoing return business because they realized that the amount of lifetime value they got from returning customers was significantly more than a one-time purchase. Even just a second purchase was enough of an indicator that any given customer was likely to generate consistent returning business. So the questions the business had were: Can we determine if and when a customer is likely to return for a second purchase, and what can we do to get a higher return rate at these key touchpoints?
In order to answer these questions, we collected customer purchase data including individual customer IDs, customer first purchase date, and if applicable, second purchase date. Once this data was collected, we simply put the data on a line graph to see the average length of time between first and second purchase. As it turned out, customers who decided to make a second purchase (again- an indicator of long-term customer retention) ended up doing so within 14 days of the first purchase. This was extremely valuable information! If we were going to get a higher customer retention rate, we needed to get our second transaction within 14 days, or we would likely lose that customer permanently.
This was actionable information! As a result of our finding, the team decided that for every customer who did not make a second purchase within 10 days of their first purchase, we would send them an email offering a discount for their next purchase. This was an attempt to get some of these soon-to-be-churned customers to make one more transaction before the likeliest point of drop-off. The result of this simple analysis and corresponding email strategy was a 5% total increase in return business, which ended up netting the company a significant increase in revenue and profits. Understanding the behavior of past customers helped to inform a strategy to retain more future customers.
For a small business who wants the option to explore their own customer trends, the easiest place to start is building a simple reporting dashboard! There are free dashboard tools like Tableau Public and Google Data Studio that allow users to build full reporting suites of charts, graphs, and tables in a way that non-technical users (like business owners!) can explore and interpret trends. The great thing about dashboards is that they are very interactive; you can build a dashboard with interactive filters such as date, customer location, product, price, or any other dimension you can get your hands on! Simply playing around with filters and dimension splits, even the least data-savvy business decision makers can find impactful insights.
Once a business is looking to take the next level of investment into being data-driven, there are several data analysis types that can be particularly helpful in identifying opportunities to increase customer acquisition. One example is data segmentation, which is simply slicing and dicing your data by all of the dimensions you have available about your customers and sales (i.e. geo, date & time, demographics, return vs new, price, acquisition source). By splitting up your data by each dimension individually, as well as combinations of dimensions, trends tend to emerge. For example, if you split your sales data by acquisition source, you will be able to see which sources generate the most business or the highest return on investment. This type of analysis can be done at its most rudimentary level in Excel pivot tables, but to be done effectively, you might want to consider hiring a freelancer or agency to help!
Proactively Predicting Customer Churn Through Regression Modeling (Data Science)
Creating a consistent flow of new customers is important for a healthy business, but perhaps just as important- if not more important- is retaining existing customers and maintaining a steady flow of return business. We discussed earlier that data science is a sub-category of data analytics that focuses on the ability to predict the future and understand complex relationships. Perhaps the most common data science use case in the business world today is regression modeling.
A regression model provides a function that describes the relationship between one or more independent variables and a dependent variable. What does that mean? In simpler terms, regression modeling helps you identify the most important factors in predicting the outcome of your most important business metrics. If we know the numbers for X and Y, can we accurately predict Z?
Regression modeling can be particularly useful for businesses looking to increase customer retention. By identifying factors that can predict churn and factors that lower the risk of churn, business decision makers can proactively predict when a customer is likely to leave and take actions that reduce the likelihood and frequency of that outcome.
One of the largest insurance companies in the United States hired Render Analytics with a similar problem. They wanted to understand the factors that contributed towards a customer leaving them for a competitor, and determine what could be done to reduce this risk. Specifically, the analysis explored many metrics related to cost, customer experience, location, and plan type. We collected hundreds of variables related to these topics, and they were all collected with the goal of predicting a single binary output: retained or churned.
Through the regression analysis, we were able to identify that some variables like plan type actually had no significant impact, and therefore strategies like trying to upsell to more expensive plans did not have an impact on retention. The insurance company was particularly curious about customer experience- they wanted to know whether positive customer experience or feeling about the brand had an impact on churn.
As it turns out, not so much. The only time customer experience had a statistically significant impact on churn was during negative call center experiences. There was a small difference in retention between positive call center and negative call center experiences (based on survey results). Our regression work revealed that the number one most impactful factor in maintaining customer retention for this insurance company was price. As it turned out, most people just wanted to go with the cheapest plan that covered their basic needs. Using that information, the insurance company invested resources into enhanced call center training and staff retention, and also began experimenting with lowering prices to match or beat local competition. The result? An increase in customer retention!
So how should a small business without the infrastructure of a national brand use regression to predict and improve customer retention? As always, it starts with collecting the right data. Similarly to pattern detection, some of the most basic regression techniques can actually be done in widely available tools like Excel, but to do a more extensive analysis, you will likely want to bring in a freelance data scientist or data analytics agency like Render Analytics to do a thorough job. This is not as expensive as one might think, and you can outsource this type of work for a few thousand, and sometimes only even hundreds of dollars.
Hire (and Retain) Better Employees with Clustering Analysis (Data Science)
Another common data science technique is clustering analysis, which is often an alternative to regression. Clustering is the task of grouping a set of objects in such a way that objects in the same group are more similar to each other than to those in other groups. This is helpful because common traits are not always obvious, and it is not always clear which traits, characteristics, or behaviors are highly correlated or connected with one another.
In the hiring process, clustering analysis can be particularly useful! By grouping current employees as well as active job applicants, a hiring manager can identify traits that are potential indicators of early flight risks. This is helpful to avoid training and investing in an employee who will likely quit only a couple of months into the job. Clustering can also help identify “positive” groups who have traits or behaviors that are indicators of productive, high-quality employees.
When it comes to hiring and retaining quality employees, some characteristics from a clustering analysis might be unique to your business, but some might be relevant across the employment landscape. One study using clustering analysis found that new hires who ask an excessive number of questions early on in their hiring are actually more likely to quit very quickly. There is of course a line, and asking a certain number of questions might just be a sign of eagerness to work. But if a new employee is constantly asking questions (and especially questions about their own benefits, pay, etc as opposed to questions about their responsibilities), this is an indicator across all industries of an early flight risk. This observation was discovered as a result of a clustering analysis across hundreds of companies!
There are so many data points your business can collect to use clustering analysis in the hiring process! When interviewing candidates, take note of a few data points. What questions are the candidates asking (if any)? Over time, you might find patterns with certain questions that you wouldn’t have otherwise noticed without recording and tracking over time. Data related to years of experience, technical skills, and other relevant background information is worth recording as well. Also record where your candidates are applying from; where did they find your job application? Over time, you’ll get some high-quality data, especially if you are in a business that hires a lot of people!
Like regression, rudimentary forms of cluster analysis can be done in more simplistic tools like Excel, but for a fully-formed analysis, using Python packages or hiring data science specialists is recommended. The goal of a clustering analysis is ultimately to find common characteristics of positive and negative “clusters” that you were not previously aware of in order to take advantage of that information in future decision making.
Wrap-Up: Getting Started with Data for Your Business
Two of the four use cases we covered here were related to data science, but before getting into that level of depth, remember that the first step is to identify the business questions you want answered and begin collecting the data related to those use cases. It might take some time to collect an adequate amount of data, but everyone has to start somewhere!
Start small by collecting and reporting your basic numbers. Choose the key performance indicators that are most relevant to your business, and build a reporting dashboard in a cheap or free tool that will help you keep track of those KPIs and explore trends. Once it is time to go a level deeper, remember that many of these techniques and analyses are best suited to be done in coding languages like Python, and tools like Excel can only get you so far.
There are freelancers on platforms like UpWork and agencies like Render Analytics that specifically focus on small businesses. Contrary to what you may believe, these solutions are affordable for businesses on a budget, and ultimately are an investment in your future. Click here if you’d like to book a free consultation with Render Analytics to learn more! If you’d like to learn about the future trends on the internet for businesses to stay in front of the curve, give this article a read!