Big Data Part 2: A Step-by-Step Data Collection Plan
Posted by Nazz Baksh
In my last post I explored the reasons and understanding behind data collection and analysis for your organization. In this post, I outline step by step how to create your plan and start collecting the most relevant data for your use.
Here are a few considerations when planning your data collection needs:
1) Know your audience
Who are you collecting the data for? Will it be used to determine the strategic direction of the organization, or to track expenditures? Identify your audience and needs first.
Now that you’ve identified your audience, consider that your demographic may grow in the future. It may be senior management today, but may include the rest of management, the board, external auditors and perhaps government bodies. Therefore, this deliverable, whether it’s a scorecard or a formalized report, needs to be constructed with audience in mind. It needs to be simple enough to use and easy to consume the information. Consider how to encourage new participants to grow your audience and establish as the source of key business analysis. This could be as simple as creating a data dictionary or a metric glossary to accompany the report you are delivering.
2) Understand the drivers (metrics) of revenue and cost
If you believe the old adage of what doesn't get measured, isn't managed, then you need to not only understand what to measure but how these elements work together in the financial performance of the organization. This understanding requires a deep understanding of the cause-and-effect relationships that exist within the income statement, the balance sheet, between operational silos and across the various business lines.
Beyond the financial perspective, there’s the customer perspective that challenges you to think about how your organization interacts with its customers. Perhaps it’s the length of call when responding to customer questions, time to close the sale, total number of new customers, lost customers, average churn ratio or net promoter scores. Measuring the touch points on customers (your revenue source) is as important as your financial metrics.
3) Look for things you can control and are truly relevant
There is no point in measuring things for the sake of measuring. For example, in the insurance business, we all know that weather has a great deal of impact on insurance claims from wicked winter driving to pulverizing hail storms to ravaging floodwaters. We are able to monitor these events and cross-reference them with atmospheric and historical data to conclude that climate change is impacting losses.
This “climate change” metric can be monitored, but why? What can you possible do about it? Unless you have the ability to control the weather by harnessing the powers of Thor’s infinity stone, it becomes a "that's nice" metric. A metric like this may help explain things or create a catalyst to look for ways to reduce losses, but ultimately it’s a weak KPI.
A stronger measurement might be examining the roofing materials used on homes, age of roofs and age of municipal infrastructure to determine whether the risk class warrants an increased risk premium. Or perhaps, in the handling of claims, using better roofing materials to mitigate hail damage, or working with municipalities on increasing the distance between homes to reduce multi-home fires. Tracking our policyholders’ roofing materials and comparing that data when assessing fire damage could lead to improved loss prevention.
In other industries, a similar argument could be made to better improve products to the end-user.
What are those strong KPIs for your industry?
4) Keep it simple
You can easily get mired in the process of developing 100+ metrics. There is no value and, in fact, excess metrics can often lead to confusing employees and derailing any potential benefits. It’s a simple case of the law of diminishing returns.
Metrics should be kept in the 20 to 30 range and only 5 or so should be financial (lagging); the rest should be leading indicators embracing the other key facets of the organization from customer metrics to internal business processes to people metrics and technology innovation. These will establish a solid foundation for the organization to develop its Balanced Scorecard and Strategy Map. In fact, organizations without a Balanced Scorecard ought to consider its adoption. (If you are considering it, stop here; designing performance metrics before a Strategy Map is putting the cart before the horse.)
5) Can you actually measure it?
After spending all this time in designing key data points, can they be reasonably measured without the use of any underlying assumptions? Consider that the more assumptions used or proxy applied simply waters down the value of the data point. Ideally, the data collected is unadultered and can be relied upon.
If you can’t measure it reasonably well or in a timely fashion, consider dropping it.
5) Talk with key subject-matter experts in the company from operations to marketing and IT to finance
How do managers run their departments? What steps do staff members take to do their jobs? How does change really affect them?
Consider the implications of what you are doing; think the entire process out before you act. By speaking to the subject matter experts in each department, you might be surprised at what seems like a micro-step can do to workloads.
In my own accounting department, I was guilty of just that. I revamped the chart of accounts from an 8-character value to 18-characters. The addition of 10-characters added, on average, an increase to workloads by 1½ hours each day,almost 400 hours annually. I had to decide if it was worth it, or were there mechanisms or efficiencies that could be used to offset the increased workload.
You may be asking whether this is worth the headache. Absolutely! The days of running your business on gut feel are long gone. Data doesn’t allow you to make decisions; it furnishes your arsenal to support decisions. It allows you to measure against your plan and, if you’re lucky, cultivate a niche market, re-examine processes and/or reveal new opportunities. None of this will happen right away. You need to have vision, a champion and the will to push ideas to the forefront.