Data has become an essential resource for business growth. Just a couple a decades ago, a company could function just fine without capturing and analysing data. What changed in between that triggered this shift? For start, markets got larger, consumers got smarter and more demanding, companies expanded to multi-national markets. Understanding and adapting to different markets is far from easy. Companies have realized the value of data in making strategical business decisions. As a result, enterprise analytics has emerged as the solution for collecting, processing and analyzing data across the entire organisation.
How Enterprise Analytics helps businesses
Companies have been investing in enterprise analytics solutions in order to get the most value for their business. These are some ways enterprise analytics can streamline business processes, boost performance, improve customer satisfaction and get a return on their investment.
- Analysing data helps to identify new business opportunities that may be otherwise neglected.
- Companies that have their own distribution centres can better predict and plan delivery capacities.
- By evaluating customer requests for refunds and complaints, businesses can identify poorly performing suppliers and make sure their customers get quality and on-time deliveries.
- When a company can define which customers are most likely to repeat purchase, it allows them to optimize their marketing investments. This, in turn, results in building long-term customer relationships.
- Seasonal businesses that require more workforce and experience higher demand during certain periods can accurately forecast staffing needs.
- Relying on data and analytics leads to making better marketing decisions and increased marketing productivity. The loads of information a brand has for its customers help streamline the customer experience by more precise targeting.
- Product management can be improved thanks to the available data the company has for most popular or sold products. Using enterprise analytics, the company can decide to target a product to sell in the right region and time, which will lead to increased sales.
The problem with barriers is that data is always there, but not everyone can see it and access it.
The effect of Disconnected Enterprise Analytics
Despite the highly developed enterprise analytics ecosystem, businesses are still realising only a fraction of the full potential value from data and analytics, as McKinsey state in their report. And there is a prevalent reason for that – barriers. The goal of doing data science and analytics is to make sense of the huge amount of data businesses have in their systems and make data-driven decisions. But one of the biggest restrictions to making use of data-driven insight into the business processes is access to data. Data is isolated in silos across multiple departments in a company, which renders doing enterprise analytics nearly impossible.
Sharing his views at the Data Innovation Summit 2019, Shann Mistry from Alteryx gave us a really clear perspective of this pervasive disconnect in enterprise analytics and some solutions for disrupting it.
“The problem with barriers is that data is always there, but not everyone can see it and access it,” points out Shaan. The reality of today’s organisations is that there is a disconnect of data between these three departments: analytics, data science and IT. Each department functions within its own barriers and has its own challenges.
Interactions between these 3 siloed departments
Shaan gives a hypothetical description of the interactions between the siloed departments are carried out – interactions which lead to building the barriers in the first place.
1) Interaction between Analysts and IT. Analysts are trying to build their processes. But they need access to data to do their work. Analysts may not be the most technical users in terms of coding or scripting. Typically the technology and tools they are using are not chosen by themselves so they seek help from IT about access to a certain data source. But IT usually have other priority projects in their scope and don’t have time to help analysts. If analysts propose a different piece of technology, they come across a block from IT saying that it’s not in their approved vendor list or agreed upon software. So analysts already some across barriers in terms of technology they can use.
2) Interaction between IT and Data Scientists. Data scientists are building their models using R, Python etc.. Once they build the models, they give over to IT, DevOps or another integration team. IT again have hands full of work and the last thing they want is to take a model from their data science colleagues and have to rewrite it for integration using Java, C++ or PHP, which is time-consuming for them. Meanwhile, the data scientists are getting frustrated with the time it takes to get a model into production. And if it gets deployed after a certain period of time, the model might be out of date.
3) Interaction between Analysts and Data Scientists. The biggest barrier between analysts and data scientist is language. It’s most likely the analysts won’t know what R or Python is. But they depend on the models that data scientists build in order to do their work. The data scientists, in turn, are placing the blame on IT for the delay in building the model and ask the analysts to help them push their models into production in the IT department.
All three departments are working independently in their own silo. And inevitably, these barriers are producing consequences in data analytics. There is a great amount of hidden data in a company and analysts are spending around one-third of their time trying to find data. Even if they find it, there is no guarantee it’s meaningful. The data comes from multiple sources and areas in different forms – structured, unstructured or semi-structured. As to time and effort, a lot of resources are spent on manual, repetitive processes. And even when models are built, only 80% of the models are partially deployed.
Different technologies that are used in different teams also present a significant barrier. Shaan says that it commonly happens that a business that has the same department in several regions, and even in that same business, the departments are using different technology for doing the one same task. So if a model is built in one region, it cannot be reused for the other regions where that same business is located.
What should companies do to bring down the barriers
Companies that want to be successful in the era of big data, have to bring down the barriers and silos. How can they achieve this?
Shaan suggests that organisation need a solution, a tool, or a platform that makes it easy for all of the users to find what they are actually searching for. Once users have found what they are looking for, they should be able to document it and store it in a place so that piece of information is available to all users within the team, business and region. If users need to request access to it, it should be done in an easy and seamless way. Once flows and processes are developed, users should be able to leverage their skills with the tool that is code-free, or code-friendly if users have those skills. They should be able to share it will other users so they can understand it too.
The IT department, in particular, should be able to take the solution that has been built, automate, scale and govern it. Models should be built, deployed and integrated, regardless of the language that is used, in a much quicker way.
Shaan envisions the ultimate goal as a barrier-free platform where everyone is using the same technology, talking the same language regardless of the skills they have. He describes the solution as a single platform where users can discover and collaborate with their colleagues, build end-to-end processes, be able to share them with other technical or non-technical users, deploy the models in a centralized way and get them in production faster. But this kind of platform wouldn’t be possible without a data-driven analytical community that will unite users.
The solution for bringing down the barriers in enterprise analytics is the ability to work together, use a same easy-to-use platform for which the learning curve is not so high so that its use can be extended and repeated, concludes Shaan.