If your business builds strategies from your data, you should be sure it’s telling you the truth. Your business needs knowledge to build successful strategies. Plato defined knowledge as justified, true, belief. While the inclusion of belief in this definition has been debated for centuries, truth is a firm criteria for knowledge. Finding truth may require looking beyond the numbers on the screen. Labs8 would like to offer you a few suggestions for keeping your data honest.
Look beyond your customer data
We capture all kinds of information about customers. We know where they live, what they do online, where they shop, where they work … the list goes on. We use this data to make assumptions about buyer’s motivations and personas. To fully know your customers, you need to look deep to find who they truly are. Customer interviews and conversations are still enormously valuable in our technology driven world. When I managed an airline loyalty program, we had a member advisory group made up of some of our most frequent fliers. We held meetings several times a year and the members were encouraged to call our team directly and let us know about their travel experiences. I had many conversations with these customers, not all were pleasant, but every one was valuable in helping us know and serve our best customers.
- Set up a Customer Advisory Group.
- Conduct quarterly customer interviews.
- Provide your Customer Service team questions to ask customers.
- Give Sales questions to ask when talking with prospects and customers.
Make data integration a priority
Each functional area of your business captures different sets of customer statistics. Without combining this information, you only have partial knowledge of your customers and prospects. If Sales is tracking in excel and Marketing is using a CRM, your business may not be getting the full picture needed to make good decisions. Data integration can be messy, especially if you are applying it to established systems, but it is critical to knowing how to expand and retain your customer base. Here are steps you can follow when implementing or updating your integration system.
Step for data integration:
- Identify what currently exists and where it is located.
- Determine what you need. For example:
- Contact info. – what is their name, their email, where do they work?
- Source – their first interaction with your business (website, email, trade show)
- Sales data – are they a prospect, a lead, or a qualified lead?
- TouchPoints – where do they engage with your business? (website, email, downloads, customer service, trade shows)
- Customer data – what do they buy, how often do they buy, how much do they spend?
- Choose software(s) that can be easily integrated with existing systems.
- Develop a uniform legend for tracking and be sure all functions use it consistently.
- Assign one person to be responsible for managing data quality and. consistency. This person will also need to oversee access controls.
- Conduct periodic reviews to be sure the software(s) and system you are using meet your changing business needs.
Prevent bias from impacting your data
This may be the most challenging aspect of finding truth. Humans invite bias into the equation when they determine hypotheses and set up experiments to test hypotheses. Biases may come from past experience, personal opinions and other considerations. These biases can get in the way of pulling knowledge from any set of information. Even with the proliferation of AI/ML, which minimizes human involvement, bias can creep in. We train the machines and determine the AI/ML algorithms. We decide what data to use and what to ignore. This wave of new technology reinforces the importance of asking the right questions before you begin collecting the data. Here are a few questions to help minimize bias.
Questions to reduce bias in the data:
- Is the sample representative of the targeted population?
- Have I allowed enough time for patterns and trends to emerge?
- Am I collecting data about the right features/attributes?
- What data am I leaving out?
- When a data set is incomplete, how do I determine whether to include or exclude it?
- What are my pre-existing biases that may influence my interpretation of the results?
Know how to identify valid data
Marketers have more tools than ever to measure the effectiveness of campaigns. Most email and ad platforms offer a/b testing to optimize performance of your ads and offers. These platforms create impressive campaign reports, but unless the results are statistically sound, they may lead your business in the wrong direction. Sampling errors are probably the most common challenge in obtaining valid results. Small samples may offer directional value, but should not be relied on for making strategic decisions for your business. While being quick to act is critical to success, acting quickly based on invalid data can put your business at risk. Here are a few steps to take to be sure your data is statistically valid.
Steps to determine data validity:
- Be sure your sample size is large enough and representative of your targeted population
- Set-up your hypothesis before you begin collecting the data
- Determine the confidence level you want to achieve (95% is commonly used)
- Use an online tool to calculate results. Microsoft Excel has a built-in formula to calculate statistical significance
- Keep in mind, results that don’t meet your statistical significance criteria, may still have directional value for your business
The team at Labs8 has extensive experience in working with data collection and integration. We would be honored to help your business fine tune the process of transforming data into knowledge.