The road to value is fraught with twists, turns, potholes, and side roads. It is rarely a straight line process to get from spreadsheets and dirty data stores to well structured data marts, data driven insights, and advanced machine learning models. Companies that are just starting down the path often think (or get pitched from some slick vendor) that there is a magic bullet or product that will solve their data management woes: that is, turn their data into something that drives business value.

Before starting out on your journey, it is a good idea to have an idea where you want to go. The conceived end state, all too often, is a utopian land full of rainbows and teams of happy, productive analysts, engineers, and data scientists churning out transformative insights and algorithms that propel the business to the next level. Unfortunately, reality usually falls short of oversold expectations. The following are a short list of things to consider as you proceed down the path to transform your business or organization with data.

Know what problems you are trying to solve. Also, your problem is oftentimes a bad process or other issue and not a lack of data. Being clear on the front end about the business problems you are trying to solve will greatly increase the odds of a successful data and analytics strategy implementation. The problems should be specific enough that you can create goals to action against. This also means an ability to measure and track progress in meeting these goals. One thing to note is that data and analysis are not a panacea. They will not fix broken or non-existent processes, poor communication, bad leadership, or organizational rivalries. Before starting down the road make sure the problem you are addressing is one that is fixable with data.

Plan to measure what you intend to fix. You cannot improve something that is not measured. The good news is that systems and services that have any sort of software component will usually generate the data you need. In fact, we generate more data now via devices, sensors, websites, and the like than at anytime in history. That’s the good news. The bad news is that you have to turn this into something you can track. Make sure to create and track the appropriate metrics that relate to your problem.

Be sure someone owns the metric. Data (i.e. a metric) without an owner is like a horse without a rider: it’s not going to take you where you want to go. Each metric should be tied to an initiative or business activity and should be owned by someone (or group of someones) who is ultimately responsible for that metric.

Know what success looks like. You need to know your destination if you are going to be successful. This should be easy if you’ve done the first step which is define the problem. Knowing your end state like hitting certain KPI’s or implementing a new capability is crucial in gauging success or failure.

Make sure you have the people, expertise, and tools to implement data driven solutions. Need a dashboard? You probably need a tool like Tableau or PowerBI. You also need someone who knows how to use these tools. The same goes for building a data platform or implementing a new algorithm. You need the right engineers and data scientists on hand to do the work in addition to the right tools. It can be hard in today’s job market to find the talent you need. Sometimes hiring a vendor to handle some or all of your data and analytical needs may be the route to take, even in the short term.

Choose the right approach, start simple then iterate. Some algorithms are like sports cars: they’re cool and flashy but may not be what you need to get the kids to little league practice. Algorithms like deep learners and collaborative filtering are used by many organizations to drive cross-sell opportunities and personalize customer experiences. However, it may be beneficial to start small and then scale. Can you get away with (and solve the same problem) using a single model? Or maybe simple business rules will get you off and running more quickly. The point is start simple and iterate towards a larger solution. This usually entails lower up-front project risk and faster solution deployment.

While this is not exhaustive list of everything you should consider, these are some of the most important ones. Having a well thought out roadmap with a good idea of the problems you are looking to solve is a necessary first step on the path to successfully leveraging data and analytics for your business. Watching out for and avoiding some these common mistakes will help to increase the odds of delivering long term value.