Imagine this scenario: The management team at a manufacturing company wants to make some capital expenditures. The staff analyst uses historical information from an old data collection system to see if they had a significant ROI the last time they made such an expenditure. They proceed with the new investment if the answer is yes, not knowing that there was a more profitable solution available.
Now imagine a different scenario: The same management team works with different functions at the company, using analytical optimization technology. People in each function are able to run different scenarios from multiple angles and come up with multiple solutions. For instance, the data tells them that instead of making capital expenditures, it’s more profitable to optimize the existing production process.
The bottom line is simple: Bad decisions can be made even when companies have an abundance of data. The key is knowing your current capabilities and what you want your data to do for you.
Piles of data don't equal actionable insights
Too much data and too few ways to use it meaningfully is a major pain point for businesses across all industries. Our Supply Chain Ingenuity survey has found that most distribution companies have the technology to collect business intelligence; however, many are unable to turn it into actionable information to support their business decisions.
There are several reasons for this disconnect, including poor configuration and user interface, lack of user training, data latency and inconsistent data. The good news is that these issues can be fixed if you know where your company is on the analytics maturity curve and why. In other words, you need to map your data analytics strategy to your current capabilities and business plan.
The analytics maturity curve: Challenges and solutions for every stage
Business intelligence helps you translate your initial investment and strategy into sales, market share and, ultimately, higher margins. The science of business intelligence has progressed along an upward curve over time, and there is a big goal in sight — to consistently and accurately predict and shape future market events.
Step 1: Descriptive analytics — what happened?
Descriptive analytics is the building block for the development of all business intelligence. It starts with buying and implementing the backbone technology that automates data collection, report writing and visualization. Descriptive analytics uses historical data to understand what happened in the past, including information about sales, customers, operations and finances.
- Lack of pull-through because the business is built around silo- and function-based departments that don’t communicate with each other. This silo structure prevents companies from quickly collecting and analyzing data, which are critical to an effective analytics program.
- Not knowing how the data will be used in a business’ management strategy. If the right questions aren’t being asked and the right data isn’t being collected, then the exercise is largely irrelevant.
- Tracking the wrong metrics due to lack of communication between departments, clear business goals or training.
- Reports are not designed correctly, which leads to pulling the wrong data or drawing incomplete conclusions.
- Incorrect system configuration that doesn’t meet business requirements (i.e., data is updated weekly instead of daily).
How to make it work: Descriptive analytics work best when you try to understand the past and build on it. Begin by assembling a cross-functional planning team (e.g., finance, operations, and marketing and sales). Set clear, measurable goals for your business strategy and make sure you have a transparent governance process to keep track of progress. Work with your IT department to configure your data systems based on your goals and train multiple people on how to use the tools.
Client case study: Proper configuration renews trust in analytics system
A global manufacturer wanted to understand their customer profitability based on existing contracts. They invested heavily in a new business intelligence system. Once people started to use the system, they realized the data was so different than what they knew about their customers that they stopped using and trusting it.
We found that the system wasn’t configured correctly. We redesigned the data architecture by fine-tuning the extract, transfer and load (ETL) process, and adding more logic into the data flow. The last critical step was to retrain people on the new configuration, restoring trust in the technology.
Step 2: Diagnostic analytics — why did it happen?
The next logical step along the maturity curve is to determine why events occurred to make better decisions in the future. These root-cause drivers of performance require the ability to segment and analyze the data in ways that answer specific questions. Understanding the user’s needs across all functions of the business is critical to success.
- Siloed business functions that don’t share goals, learning and plans
- Lack of visibility into root-cause drivers and how they impact each other across all functions of the business (sales, operations, finance)
- Wrong technology configuration, which leads inaccurate snapshots into the drivers
How to make it work: Effective diagnosis starts with bringing the right people to the same table. Operations, sales, finance and marketing need to start talking to each other about goals, and then start designing cross-functional planning processes and discuss the data you have collected and why. For instance, if you are a manufacturer, use your sales and operations planning (S&OP) tool monthly. Integrate finance, analytics and business intelligence into the S&OP process to have a cross-functional dialogue based on a shared understanding of root-cause drivers.
Client case study: From taming data errors to root-cause analysis
A manufacturer and distributor of supplies for the energy industry wanted to understand why they had budget variances and to what extent (e.g., prices were going down, but raw materials costs were up). Their initial approach was to make educated guesses using Excel-based tools.
When guessing proved ineffective, the company worked with Grant Thornton to perform a diagnostic study. We analyzed variables like product mix, availability, people, processes, time and lack of visibility to help understand the budget variances and develop corrective action. We found that much of the time was spent on low-value activities like explaining why budget errors happened, instead of high-level work like getting to the root cause.
Our solution was to create a data-based dynamic model that gave the client information about points in the supply chain (i.e., inventory levels), as well as across the chain for particular products (i.e., profitability by product). The model gave the company an unprecedented view into root causes, which helped them answer long-standing budgeting and pricing questions.
Step 3: Predictive analytics — what will happen?
Companies that have mastered descriptive and diagnostic analytics can start asking questions about the future thanks to predictive analytics. They can find out with a high degree of confidence how likely they are to meet a particular goal.
Predictive analytics is a very powerful tool that is driven by big data. It requires building the proper security, data flow and workflow to make the data usable and easy to understand. Predictive analytics works best when it measures a single outcome and how external factors impact that outcome.
- Shortage of advanced skills, which is inherent to turning from data plumbers into data scientists
- Limited technology that can’t be reconfigured to run predictive models
- Additional types of data that you don’t normally collect
- Not knowing when or how often to run analyses that give you relevant data
How to make it work: Chances are your team has the knowledge you need to run predictive analyses, but they need proper training. Identify the specific skills that you need and invest in thorough training. You also can take a closer look at who runs the analysis and uses the output, and set expectations for both parts. For instance, the user needs to ask the right questions and the analytics scientist needs to know how to get it in an understandable format.
Investing in new, more advanced technologies to properly run predictive analyses may seem discouraging, but the investment will pay for itself many times over. Being able to consistently choose the most profitable scenarios in a highly competitive market is invaluable.
5 ways predictive analytics can grow your business
- Build a specific customer view — Analytical customer relationship management (CRM) stems from predictive analytics and is the tool you need to understand your customers. CRM can tell you who buys your products or services and why. It also can help you build buyer personas, which you can use to improve your offerings.
- Cross-sell for higher profitability —You can use predictive analytics to see how your buyer personas may perform in different scenarios. This is particularly useful if you want to expand your product or service range.
- Increase customer retention by detecting silent attrition — Predictive models can tell you how likely a customer is to leave you by examining their past service usage, service performance, spending and other behavior patterns.
- Nurture leads through strategic direct marketing — Predictive analytics can find the most effective combination of product versions, marketing material, communication channels and timing that you can use to target customers.
- Manage risk — Predictive analytics can help you spot patterns in business activity that may lead to fraudulent behavior. It can also help you assess the outcomes of entering new markets, launching new products or switching vendors.
Step 4: Prescriptive analytics — how can we make it happen?
Prescriptive analytics will tell you how to perfectly synch up business functions and optimize variables to achieve your goals. It’s all about adjusting your tactics based on analyzing the forecasts and constraints.
Prescriptive technology — the pinnacle of business intelligence done right — is very new, and few companies are using it at full potential. It allows you to adjust your objectives in real time to stay ahead of everyone else, which gives unprecedented flexibility and competitive edge to companies across all industries.
- Strong and steady leadership support to fully embrace the shift to holistic planning processes
- Difficulties related to creating a true collaborative environment where all teams and functions are communicating effectively
- Competition for a very limited pool of highly skilled data analysts
How to make it work: If your business wants to embark on the path to prescriptive analytics, start with making sure it’s possible to coordinate efforts across business functions and at all levels (strategic, policy/tactical and operational). These capabilities require breaking down the silos and collaborating to support better decision-making.
Also, in order to dilute any risks and investment costs, break the prescriptive analytics implementation into pieces. A good place to start may be with finance, using proven processes and tools, and then move on to sales and operational planning. This way, baby steps can result in major gains, especially in cases where operations and finance have not had direct communication channels previously.
4 prescriptive analytics applications that can benefit your business
- Fully integrated business modeling can help you outline the demand, supply and full financial model into a single representation.
- Holistic optimization allows you to identify the financially optimal alignment of resources and assets to achieve a desired result while maximizing critical business metrics (like profitability or quality).
- Forward-looking costing helps you understand forward-looking root- cause drivers of the business (as opposed to descriptive and diagnostic analytics, which are backward-looking).
- Marginal economics help you understand the economic value of products, resources and bottlenecks in your business.
Here are some questions you need to ask in order to advance your analytics program:
5 questions to assess your business intelligence program
Most companies are using one or more of the analytics methods described above. But are they using them fully or effectively? In our Supply Chain Ingenuity survey, just over half (54%) of all respondents rate their company’s data as mostly or completely actionable, but only 19% say their data is completely actionable. This leaves 81% with room for improvement.
- Do I have leadership buy-in? Shift your corporate mindset to encourage the understanding and use of analytics. Make a strong case for how profitability would improve if business decisions were based on strategic, forward-looking insights from the data you’re already collecting.
- Do I know what my business intelligence capabilities are? Start by finding out what resources you already have and then fill the gaps from the ground up. Building a solid foundation will result in a robust advanced analytics capability that you can leverage across the organization and continue to expand upon.
- Do I need to hire analysts? One key issue is finding staff that is experienced and capable of producing relevant, timely analytics. Few businesses have the personnel to execute more advanced projects, which may be the biggest overall hurdle to successfully implementing an analytics program. Many companies are getting help from consultants and software to develop the infrastructure they need to use analytics for true competitive advantage.
- Do I have the right technology? Excel spreadsheets can only take you so far. You may need to upgrade or purchase new software and hardware, depending on the size of your database and the type of analyses you want to run. If you want to grow your business intelligence program over time, purchase a platform that you can upgrade incrementally.
- Do I know what questions I need to ask? Powerful technology and a team of analysts are not enough to get you the right answers. You need:
- Quantifiable goals – "I want to increase my on-time delivery rate by 10%."
- Hypotheses – "Switching to a just-in-time (JIT) delivery model will help improve the overall delivery rate by 10%."
- Questions – "What variables should I measure to see if JIT deliveries will increase my on-time performance."