This article explores eight ways to help your organisation realise an improved, future state of business intelligence and data analytics. While some aspects will be easier to adopt than others, taking any of these steps will improve the outcomes that your business intelligence (BI) and data analytics (DA) unlocks.
1. Push past analytics into actions
As with many technology or digital projects, it can be easy to focus on the doing rather than the customer or the why. In the instance of data analytics, businesses need to think about what actions they want to take following the analytics, then let these actions inform which type of analytics is most suitable.
Failure to think about the next steps can result in 'wasted' analytics projects - getting findings but not doing anything with them or how you might benefit your customer. Is your organisation willing to change what it's doing as a result of the potential findings? Is you brave enough to challenge the perceptions of your current processes? Being open to future actions may lead you to explore areas that you haven’t even considered or developing customer experiences you didn’t know were needed.
2. Embrace the data openness and privacy paradox
In the context of GDPR and data sovereignty, businesses seeking to maximise the effectiveness of their BI and DA efforts face a significant hurdle with access to data. And it will only get worse. Businesses need to respect customer privacy, while being equally open and transparent with how customer data is used.
But can you be both open and privately minded when it comes to data? Aimee Whitcroft, Advisory Board member of the Open Data Charter, believes so. While there is a tension, if handled well openness ensures transparency, providing people with the confidence of knowing exactly what a business is using their data for and with whom they’re using it. From the privacy perspective, it’s protecting personal information to ensure it doesn’t get into the hands of people who will use it for ill gain.
Whitcroft urges organisations to constantly look for ways to anonymise their datasets, and recommends taking the following approach:
- De-identification. The removal of names and birth dates.
- Aggregation. The combination and simplification of data sets
- Perturbation. The addition of random noise to the data set.
- Suppression. The decision to not report some data.
- Limitation. The withholding of access to raw data.
The reason for taking such steps is to ensure the data set cannot be reverse engineered in an attempt to uncover the actual identities of people contained within.
3. Generate new data
While there is significant data in the world, ownership and access to this data can be problematic. The answer is to look in places your competitors normally wouldn’t and start looking as soon as possible.
“Data isn't like normal software where you can hold off upgrading to a new system then start receiving the full benefits as soon as you've pressed go. There's no making up for lost time. Delays in collecting data will result in missing out on competitive advantages and insights that could drive your business forward.” Syen Nik, Head of Machine Learning, Jade.
Mining unstructured documents is one such option. By extracting data from existing documentation, businesses can use machine learning and other statistical methodologies to cost-effectively generate their own data or enrich what data currently exists. While this might initially sound like an endeavour reserved purely for externally sourced documents, it can also be used to automate the extraction of internal information contained in historical documents.
Computer vision has the ability to generate volumes of data too. Businesses can often utilise existing infrastructure, such as CCTV networks, some of which have been in place for decades. For example, data from traffic cameras could be used in an app that helps people save time when finding carparks. Security cameras could be used to track customer movement through stores or for people counting. With these two examples, data that was ‘hiding’ in plain sight can now be cost-effectively used by businesses.
4. Decentralise your data
The enterprise sector has seen a significant uptake in microservices, SaaS-based products over the past five years. For all their benefits and beautiful design, the significant rise in popularity has led to the creation of even more data silos. This brings with it not just data privacy and security headaches for IT departments, but also for those who rely on up-to-date data when making decisions. One solution is to adopt a decentralised data approach to ensure data is reliable, clean, and complete!
Below is an overview of what is involved with decentralising data. Explore this topic in further detail in The case for decentralising data ebook.
- Customer centricity is foundational for decentralised data. The whole business, various processes, and of course data all need to support the delivery of products, services, and experiences to the customer. Privacy and security need to be comprehensively addressed in this process, particularly with the likes of GDPR. While decentralised data can be accomplished without a customer focus, businesses can easily invest in processes that offer no meaningful difference to their prospect and customer base - which raises questions of the processes in the first place.
- Decentralised data governance involves gaining an in-depth understanding of the high-level business entities or data blocks. In any given business, there is likely to be an entity around the customer, the product, or the service. There could also be entities with regards to projects, workers, tasks, invoices of expenses, and so on. Many businesses have already mapped this out as part of their enterprise architecture.
- In order to keep accurate records in a constantly changing environment, businesses need to assign a master (or main) system to each entity, for which other datasets all refer to when requesting information. Addresses and customers might be mastered by your CRM system. Product specifications like policies might be mastered in your policy administration system. Invoices and transactions might be mastered in your billing system or accounting system.
- Once entities have been created or modified, they need a mechanism to distribute (publish subscribe pattern). Part of this process is defining one way or two-way synchronisation. It may be that some systems need to update others, even when they are not the master.
Decentralisation spreads the ownership of data across the organisation, which typically means businesses can have greater confidence with the data they’re using when making decisions. But more importantly, decentralising data unlocks a level of speed that enterprises can only achieve in their wildest dreams. Being able to implement new systems without impacting others or needing months of regression testing is also a massive benefit.
5. Narrate your data
At the beginning of 2020, Seed Scientific estimated the size of the digital universe to be about 44 zettabytes of data (byte, mega, giga, tera, peta, exa, zetta, yotta). And that was before the sea of recorded webinars and virtual meetings flooded the business world.
The magnitude of this volume of data can quickly paralyse a businesses’ decision-making process and productivity. This is where data analytics, and more so, data narration, comes into play. Extracting information then providing commentary (visual or written) around this is the best way we can make sense of such vast pools of data. Data narration uses natural language generation to turn tables or spreadsheets into what you could call executive summaries, directing people to the most relevant findings.
“The value of data is only realised when it is used, and when it is used often."
Aimee Whitcroft, Open Data Symposium, Apia.
Businesses can have the largest and cleanest datasets, and the most complete view of the customer, but if they can’t narrate or visualise the data, then its value is virtually zero.
6. Democratise business intelligence and data analytics
Making informed decisions at speed is one of the key outcomes that business intelligence and data analytics provides, which in turn help organisations to get ahead of the competition. In reality, decisions are often delayed because people do not have the information they need. Essentially, people from departments across the business all need to have access to business intelligence and data analytics tools.
It goes further than providing people with tools, it also means implementing processes, automation, and even artificial intelligence to remove bottlenecks and dependencies on certain people, as mentioned above, with the end goal of empowering and unlocking the productivity of others.
“Empowerment is about providing more power to more people, in order to bring value, not only to the organisation, but to the individuals themselves.” Terri Simpkin, Education Committee Chair, Infrastructure Masons.
The workforce of the future will have access to business intelligence and data analytics as easily as documents, spreadsheets, email, communication, and other apps. What’s more, BI and DA will either be built into these apps or integrated into existing workflows, which all maximise productivity and deliver the best customer and employee experiences. One example of this is prescriptive analytics, where businesses can use trained models and other AI smarts to provide recommendations on which course of action to take.
7. Breakdown infrastructure bottlenecks
IT infrastructure can strangle an organisation’s ability to provide the level of critical insights that business intelligence and data analytics offer. Peter Hind, a senior analyst at ADAPT, reiterated the need for organisations to identify situations where critical analytical applications are undermined by inadequate IT infrastructure. Hind recommends businesses conduct stress tests to identify such bottlenecks, and highlighted three potential issues being: I/O to disk, inadequate memory, and or concurrency issues.
8. Invest in new skillsets and capabilities
For some businesses, adopting effective business intelligence and data analytics practices is out of immediate consideration due to a lack of internal skills. Retraining, hiring, or outsourcing are three options to address the absence of these capabilities but each has its drawbacks.
Retraining takes time but once it’s done, employees can combine their new skills with their existing knowledge of the company works. Their learning can be done in the context of your business too. Hiring is an option but the market for such skills is very competitive. Outsourcing is a common solution but can be costly in the long-run.
"Data science is a combination of business knowledge, programming skills, and statistics. Finding a data scientist is like looking for a unicorn. Build a data science team instead." Syen Nik, Head of Machine Learning, Jade.
As a general approach, Jade believes in a combination of retraining and outsourcing. Our long-term partnership model enables our clients to initially utilise our heavy-lifting technical skills, while we help upskill their employees so they can manage or operate the various platforms we implement. This often involves providing training to maximise their investment and achieve their desired results.
Start realising the full potential of business intelligence and data analytics today.
The goal of implementing business intelligence and data analytics is to generate extra value for your customers, which leads to providing new opportunities to improve customer experiences, grow acquisition, increase share of wallet, drive loyalty, and more.
The sooner you invest in business intelligence and data analytics, the sooner your business can achieve all that results from clearer insights and better guidance – particularly with when making both strategic and operational decisions tomorrow. When you implement a data-rich environment with a strong customer focus, not only will your customers be delighted by the types of experiences you will provide, your shareholders will be more than delighted too.